CN117930786A - Intelligent digital twin simulation system for steel production process - Google Patents
Intelligent digital twin simulation system for steel production process Download PDFInfo
- Publication number
- CN117930786A CN117930786A CN202410324537.7A CN202410324537A CN117930786A CN 117930786 A CN117930786 A CN 117930786A CN 202410324537 A CN202410324537 A CN 202410324537A CN 117930786 A CN117930786 A CN 117930786A
- Authority
- CN
- China
- Prior art keywords
- production
- optimization
- module
- model
- network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 496
- 229910000831 Steel Inorganic materials 0.000 title claims abstract description 57
- 239000010959 steel Substances 0.000 title claims abstract description 57
- 238000004088 simulation Methods 0.000 title claims abstract description 48
- 238000005457 optimization Methods 0.000 claims abstract description 253
- 239000002994 raw material Substances 0.000 claims abstract description 47
- 238000005516 engineering process Methods 0.000 claims abstract description 38
- 230000008859 change Effects 0.000 claims abstract description 36
- 238000000034 method Methods 0.000 claims description 166
- 238000004458 analytical method Methods 0.000 claims description 119
- 230000008569 process Effects 0.000 claims description 63
- 230000006870 function Effects 0.000 claims description 57
- 244000005700 microbiome Species 0.000 claims description 57
- 238000004422 calculation algorithm Methods 0.000 claims description 56
- 230000004044 response Effects 0.000 claims description 35
- 238000001816 cooling Methods 0.000 claims description 31
- 230000007704 transition Effects 0.000 claims description 29
- 238000011217 control strategy Methods 0.000 claims description 26
- 238000012544 monitoring process Methods 0.000 claims description 26
- 238000005096 rolling process Methods 0.000 claims description 26
- 230000000813 microbial effect Effects 0.000 claims description 24
- 238000007726 management method Methods 0.000 claims description 23
- 238000011156 evaluation Methods 0.000 claims description 21
- 239000000203 mixture Substances 0.000 claims description 21
- 238000001514 detection method Methods 0.000 claims description 19
- 230000002068 genetic effect Effects 0.000 claims description 19
- 238000013527 convolutional neural network Methods 0.000 claims description 18
- 238000009826 distribution Methods 0.000 claims description 18
- 230000008901 benefit Effects 0.000 claims description 15
- 238000013461 design Methods 0.000 claims description 15
- 230000010354 integration Effects 0.000 claims description 15
- 239000011159 matrix material Substances 0.000 claims description 15
- 239000000498 cooling water Substances 0.000 claims description 14
- 238000013468 resource allocation Methods 0.000 claims description 14
- 238000004364 calculation method Methods 0.000 claims description 12
- 230000006385 cognitive dynamics Effects 0.000 claims description 12
- 230000001149 cognitive effect Effects 0.000 claims description 12
- 238000005265 energy consumption Methods 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 12
- 238000012549 training Methods 0.000 claims description 12
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 claims description 11
- 229910052799 carbon Inorganic materials 0.000 claims description 11
- 230000006872 improvement Effects 0.000 claims description 11
- 239000000463 material Substances 0.000 claims description 11
- 230000035772 mutation Effects 0.000 claims description 11
- 230000019771 cognition Effects 0.000 claims description 10
- 238000010801 machine learning Methods 0.000 claims description 10
- 238000009628 steelmaking Methods 0.000 claims description 10
- 239000000126 substance Substances 0.000 claims description 10
- 238000007476 Maximum Likelihood Methods 0.000 claims description 9
- 230000007613 environmental effect Effects 0.000 claims description 9
- 238000003012 network analysis Methods 0.000 claims description 9
- 230000010365 information processing Effects 0.000 claims description 8
- 230000007246 mechanism Effects 0.000 claims description 8
- 238000003909 pattern recognition Methods 0.000 claims description 8
- 230000003044 adaptive effect Effects 0.000 claims description 7
- 238000001311 chemical methods and process Methods 0.000 claims description 7
- 238000010276 construction Methods 0.000 claims description 7
- 238000012423 maintenance Methods 0.000 claims description 7
- 238000001712 DNA sequencing Methods 0.000 claims description 6
- 238000011478 gradient descent method Methods 0.000 claims description 6
- 230000003993 interaction Effects 0.000 claims description 6
- 230000000737 periodic effect Effects 0.000 claims description 6
- 238000012163 sequencing technique Methods 0.000 claims description 6
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 5
- 238000004140 cleaning Methods 0.000 claims description 5
- 230000001427 coherent effect Effects 0.000 claims description 5
- 230000010485 coping Effects 0.000 claims description 5
- 238000003066 decision tree Methods 0.000 claims description 5
- 238000009472 formulation Methods 0.000 claims description 5
- 230000004927 fusion Effects 0.000 claims description 5
- 238000012482 interaction analysis Methods 0.000 claims description 5
- 238000011176 pooling Methods 0.000 claims description 5
- 238000007637 random forest analysis Methods 0.000 claims description 5
- 238000003753 real-time PCR Methods 0.000 claims description 5
- 239000013598 vector Substances 0.000 claims description 5
- 208000025174 PANDAS Diseases 0.000 claims description 4
- 208000021155 Paediatric autoimmune neuropsychiatric disorders associated with streptococcal infection Diseases 0.000 claims description 4
- 235000016496 Panda oleosa Nutrition 0.000 claims description 4
- 238000013459 approach Methods 0.000 claims description 4
- 230000004888 barrier function Effects 0.000 claims description 4
- 238000006243 chemical reaction Methods 0.000 claims description 4
- 230000008713 feedback mechanism Effects 0.000 claims description 4
- 238000012165 high-throughput sequencing Methods 0.000 claims description 4
- 238000013178 mathematical model Methods 0.000 claims description 4
- 238000010206 sensitivity analysis Methods 0.000 claims description 4
- 230000006978 adaptation Effects 0.000 claims description 3
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 claims description 3
- 238000005520 cutting process Methods 0.000 claims description 3
- 238000005259 measurement Methods 0.000 claims description 3
- 229910052760 oxygen Inorganic materials 0.000 claims description 3
- 239000001301 oxygen Substances 0.000 claims description 3
- 230000002085 persistent effect Effects 0.000 claims description 3
- 238000005316 response function Methods 0.000 claims description 3
- 230000003213 activating effect Effects 0.000 claims description 2
- 230000003115 biocidal effect Effects 0.000 claims description 2
- 240000000220 Panda oleosa Species 0.000 claims 1
- 238000005094 computer simulation Methods 0.000 abstract description 2
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 10
- 238000010586 diagram Methods 0.000 description 10
- 238000009749 continuous casting Methods 0.000 description 6
- 230000004913 activation Effects 0.000 description 5
- 229910052742 iron Inorganic materials 0.000 description 5
- 230000001965 increasing effect Effects 0.000 description 4
- 241000196324 Embryophyta Species 0.000 description 3
- 240000004718 Panda Species 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 239000007789 gas Substances 0.000 description 3
- 238000002360 preparation method Methods 0.000 description 3
- 230000001105 regulatory effect Effects 0.000 description 3
- 239000000571 coke Substances 0.000 description 2
- 125000004122 cyclic group Chemical group 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000010921 in-depth analysis Methods 0.000 description 2
- 230000002906 microbiologic effect Effects 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 238000007670 refining Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 108020004465 16S ribosomal RNA Proteins 0.000 description 1
- 238000007400 DNA extraction Methods 0.000 description 1
- 241000566145 Otus Species 0.000 description 1
- 241000960400 Torenia Species 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000003321 amplification Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 239000003034 coal gas Substances 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 230000001276 controlling effect Effects 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 230000002354 daily effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000013210 evaluation model Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000007614 genetic variation Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 230000002045 lasting effect Effects 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 238000013486 operation strategy Methods 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 108090000623 proteins and genes Proteins 0.000 description 1
- 238000003908 quality control method Methods 0.000 description 1
- 238000013441 quality evaluation Methods 0.000 description 1
- 230000002787 reinforcement Effects 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000005059 solid analysis Methods 0.000 description 1
- 241000894007 species Species 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06312—Adjustment or analysis of established resource schedule, e.g. resource or task levelling, or dynamic rescheduling
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/418—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM]
- G05B19/41865—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM] characterised by job scheduling, process planning, material flow
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/092—Reinforcement learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/0985—Hyperparameter optimisation; Meta-learning; Learning-to-learn
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
- G06N5/048—Fuzzy inferencing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N7/00—Computing arrangements based on specific mathematical models
- G06N7/01—Probabilistic graphical models, e.g. probabilistic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
- G06Q10/047—Optimisation of routes or paths, e.g. travelling salesman problem
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
- G16B25/00—ICT specially adapted for hybridisation; ICT specially adapted for gene or protein expression
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
- G16B40/00—ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
- G16B40/10—Signal processing, e.g. from mass spectrometry [MS] or from PCR
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
- G16B40/00—ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
- G16B40/20—Supervised data analysis
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C10/00—Computational theoretical chemistry, i.e. ICT specially adapted for theoretical aspects of quantum chemistry, molecular mechanics, molecular dynamics or the like
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
- G16C20/10—Analysis or design of chemical reactions, syntheses or processes
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
- G16C20/70—Machine learning, data mining or chemometrics
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/30—Nc systems
- G05B2219/32—Operator till task planning
- G05B2219/32252—Scheduling production, machining, job shop
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2123/00—Data types
- G06F2123/02—Data types in the time domain, e.g. time-series data
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Business, Economics & Management (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Human Resources & Organizations (AREA)
- Molecular Biology (AREA)
- Medical Informatics (AREA)
- Computational Linguistics (AREA)
- Evolutionary Biology (AREA)
- Biomedical Technology (AREA)
- Strategic Management (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Economics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Databases & Information Systems (AREA)
- Chemical & Material Sciences (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Biotechnology (AREA)
- Game Theory and Decision Science (AREA)
- Epidemiology (AREA)
- Public Health (AREA)
- Bioethics (AREA)
Abstract
The invention relates to the technical field of intelligent control, in particular to an intelligent digital twin simulation system for a steel production process. The invention remarkably improves the intelligent level and adaptability of the steel production process by integrating advanced modeling, simulation, optimization and decision support technologies. The combination of dynamic system modeling and simulation, including the application of continuous time Markov chains and random differential equations, overcomes the shortcomings of the traditional system in the aspect of accurately simulating random fluctuation and complex dynamics in the production process. The system can more accurately predict the change of the production process, and adjust the model parameters in real time, so that the system effectively responds to the quality fluctuation of raw materials and the change of energy cost.
Description
Technical Field
The invention relates to the technical field of intelligent control, in particular to an intelligent digital twin simulation system for a steel production process.
Background
The intelligent control technology field is a highly comprehensive field, and comprises knowledge and technology of various disciplines such as automation technology, computer science, artificial intelligence, machine learning and the like. In this field, the design and implementation of intelligent control systems relies on accurate monitoring, prediction, optimization and control of complex systems. The systems can automatically adapt to environmental changes, and realize efficient and intelligent management and control of industrial processes, robots, traffic systems and the like. The goal of intelligent control technology is to enhance the performance, efficiency and reliability of the system by increasing the level of intelligence of the system while reducing the need for human manipulation.
The intelligent digital twin simulation system for the steel production process is a specific application example in the technical field of intelligent control, and aims to simulate, monitor and predict a real production process by constructing a virtual copy of the steel production process. The purpose is to optimize the production flow by a digital means, improve the production efficiency and the product quality, and reduce the energy consumption and the cost. Digital twinning techniques allow enterprises to test and verify production strategies in virtual environments, predict equipment failures, and implement maintenance plans to optimize operations without affecting actual production.
Although the prior art achieves remarkable achievement in improving the intelligent level of the system and enhancing the performance efficiency and the reliability, the problem of insufficient accuracy still exists in accurately simulating and predicting random fluctuation and complex reaction dynamics in the production process. Especially in the aspects of the synchronism and the sensitivity of the real-time data feedback and the model parameter updating, the conventional system is difficult to realize the optimal configuration, and the adaptability and the resource utilization efficiency of the production process are affected. In addition, for dynamic optimization processing of complex production decisions, especially in the face of external factors such as raw material quality fluctuation and energy cost variation, the existing scheme has yet to be improved in the ability to dynamically adjust production parameters to maintain product quality and process safety. In terms of morphological analysis of the production flow, complexity management of the production flow and system optimization based on microecology dynamics, the prior art fails to fully develop and apply mathematical and ecological principles to optimize and simplify the production path, so that there is a limit in realizing dual promotion of production efficiency and system stability.
Disclosure of Invention
The invention aims to solve the defects existing in the prior art, and provides an intelligent digital twin simulation system for the steel production process.
In order to achieve the above purpose, the present invention adopts the following technical scheme: the intelligent digital twin simulation system for the steel production process comprises an integrated modeling and simulation module, a production optimization and regulation module, a flow and morphology analysis module, a cognitive dynamic decision module, a network and stability management module, a microecology regulation and control optimization module, a decision support and strategy optimization module and a dynamic response and regulation module;
The integrated modeling and simulation module is based on the physical and chemical processes of steel production, adopts a continuous time Markov chain model to describe the state transition probability in the production process, introduces a random differential equation method, simulates random fluctuation in the process, adjusts model parameters according to real-time monitoring data through maximum likelihood estimation, optimizes a model structure by using a genetic algorithm, reflects the current production process and generates a preliminary dynamic model;
The production optimization and regulation module adopts a model prediction control method based on a preliminary dynamic model, designs a prediction model to predict future production trend, optimizes continuous and discrete production control variables by combining a mixed integer linear programming algorithm, and matches the changes of raw material quality and energy cost to generate optimized production parameters;
The flow and morphology analysis module analyzes topological structures of the material flow and the information flow by adopting a differential topology method based on optimized production parameters, identifies key structural features and bottlenecks in the production process by using Tonlon analysis, optimizes the production path by adopting a local linear embedding method, and generates a production path optimization scheme;
The cognitive dynamic decision module is based on a production path optimization scheme, adopts a convolutional neural network, analyzes historical production data and current state, identifies modes and trends, performs strategy optimization by using a deep Q network, and generates a cognitive optimization decision result in a cyclically-changed production environment;
The network and stability management module analyzes the complex network constructed in the production process based on the cognition optimization decision result by adopting a centrality analysis and community detection algorithm, identifies key nodes and fragile links in the network, adjusts the network structure by a modularized optimization technology, and generates a network optimization strategy;
The micro-ecological regulation and control optimization module is based on a network optimization strategy, adopts a DNA sequencing technology and a quantitative polymerase chain reaction method, analyzes the microbial community composition in cooling water, analyzes the interaction between the microbial composition and cooling efficiency by combining a machine learning model, and guides the formulation of a regulation and control strategy to generate a microbial regulation and control scheme;
the decision support and strategy optimization module integrates data by adopting a data fusion technology based on a preliminary dynamic model, optimized production parameters, a production path optimization scheme, a cognitive optimization decision result, a network optimization strategy and a microorganism regulation scheme, evaluates the advantages and disadvantages of various decision schemes by using a hierarchical analysis process, and generates a comprehensive regulation scheme;
the dynamic response and adjustment module adopts fuzzy logic control to adjust the operation parameters of the production line in real time based on the comprehensive adjustment scheme, combines detection and response of emergency events and rolling time domain optimization, continuously optimizes decision and control strategies in the production process, and generates an adaptive adjustment scheme of the production process.
The invention is improved in that the preliminary dynamic model comprises probability distribution of production state variables, fluctuation mode of time sequence and response function of key nodes of the production process, the optimized production parameters comprise regulated furnace temperature setting, carbon and oxygen input proportion and cooling rate in the steelmaking process, the production path optimization scheme comprises simplified material handling flow, recombined production line layout and optimized information feedback mechanism, the cognition optimization decision result comprises a raw material purchase plan based on trend prediction, self-adaptive regulation rules of production scheduling and decision framework of emergency response, the network optimization strategy comprises optimized production core network connection, logistics distribution route and information flow transparency measures, the microorganism regulation scheme comprises selected antibiotic biological pollution chemicals, periodic microbial community monitoring plan and periodic frequency of cooling water cleaning and maintenance, the comprehensive regulation scheme comprises production efficiency optimization measures, energy consumption regulation strategies and environment emission control schemes, and the production process adaptation regulation scheme comprises updated production efficiency indexes, regulation records responding to external changes and improved product quality.
The invention is improved in that the integrated modeling and simulation module comprises a model integration sub-module, a parameter estimation sub-module and a model structure optimization sub-module;
The model integration submodule adopts a continuous time Markov chain model based on the physical and chemical processes of steel production, a state space is used for defining production states, a transition matrix designates the probability from one state to the other state, the probability description of the transition between the states is carried out through the construction of the state transition matrix, a random differential equation is introduced, and the simulation of random fluctuation is carried out through the Euler-Maruyama approximation, so that a state transition and fluctuation model is generated;
The parameter estimation submodule carries out maximum likelihood estimation based on the state conversion and fluctuation model, adjusts parameters by using a gradient descent method, maximizes a likelihood function of observed data, gradually optimizes model parameters by setting an initial parameter estimation value, a learning rate and iteration times, enables the model parameters to be matched with production data, and generates a parameter optimization model;
The model structure optimization submodule is based on a parameter optimization model, adopts a genetic algorithm to optimize a model structure, evaluates model performance by defining a fitness function, adopts selection, crossing and mutation operations, simulates a natural evolution process, selects a model with excellent performance according to the fitness function, generates a new model by randomly selecting a cross point combination of model parameters through the crossing operation, and repeatedly executes the new model structure until the optimal model structure is captured by randomly changing partial values in the model parameters and introducing new genetic diversity to generate a preliminary dynamic model.
The invention is improved in that the production optimization and regulation module comprises a control strategy sub-module, a parameter adjustment sub-module and an external factor response sub-module;
The control strategy submodule carries out a model prediction control method based on a preliminary dynamic model, builds a control and prediction framework to predict future production trend, adjusts a control strategy according to the model prediction control method, sets a control target by defining state space representation of the prediction model, predicts the production process of each time period by utilizing a rolling time window technology, determines optimal setting of production parameters under given constraint conditions by adopting a linear programming solver, dynamically adjusts the operation parameters of the production line, and generates a prediction control scheme according to the predicted production demand and raw material supply condition;
The parameter adjustment submodule is based on a predictive control scheme, adopts a mixed integer linear programming algorithm to refine and optimize production control variables, sets an optimization objective function to minimize production cost by defining an optimization problem comprising continuous and discrete decision variables, meets production and safety constraint conditions, applies Gurobi to solve the problem, and generates a production parameter optimization scheme by adjusting algorithm parameters including a boundary or cutting plane strategy of a branch-and-bound method and optimizing a solution process;
The external factor response submodule analyzes and responds to external factors based on a production parameter optimization scheme, potential influences of the external factors on a production process are evaluated through collecting and analyzing market demand change, raw material price fluctuation and energy supply condition information and using sensitivity analysis, a scene analysis method is adopted to simulate production results under various conditions, and a production strategy is adjusted, wherein the method comprises the steps of changing a raw material use plan and adjusting energy utilization efficiency in the production process, and optimized production parameters are generated.
The invention is improved in that the flow and morphology analysis module comprises a topological structure analysis sub-module, a bottleneck identification sub-module and a path optimization sub-module;
The topological structure analysis submodule performs a differential topological method to refine and analyze the structures of logistics and information flows based on optimized production parameters, identifies key topological characteristics in the assembly by constructing the representation of the logistics and the information flows in a multidimensional space and applying coherent group and Betty number calculation, reveals potential connectivity and barrier points in the logistics and the information flows by a ring and a cavity, and comprises the steps of calculating topological invariants by MATLAB to generate a topological structure analysis result;
The bottleneck recognition sub-module is used for analyzing key structural features and bottlenecks in the production process by adopting a Tonlun analysis method based on a topological structure analysis result, recognizing persistent structural features influencing the production process by comparing topological models under various working conditions, selecting a part of the production process which still keeps the structural characteristics unchanged under small variation by utilizing path connectivity and space deformation, constructing and analyzing a mathematical model by using a SciPy library of Python, recognizing key bottlenecks obstructing production efficiency, and generating a bottleneck recognition result;
the path optimization submodule optimizes the production path by using a local linear embedding method based on the bottleneck recognition result, captures the optimal representation capable of keeping the local structure in a low-dimensional space by referring to the local similarity of each link in the production process, and comprises reconstructing the production path by using a local linear embedding algorithm provided by a Scikit-learn library in Python to generate a production path optimization scheme.
The invention is improved in that the cognitive dynamic decision module comprises an information processing sub-module, a learning optimization sub-module and a strategy adjustment sub-module;
The information processing submodule is used for executing a convolutional neural network based on a production path optimization scheme, analyzing historical production data and a current state, constructing a convolutional neural network model, extracting spatial features in the data by using a convolutional layer, enabling each layer of convolution to follow an activating layer, optimizing the nonlinear processing capacity of the model by using a ReLU function, reducing the spatial dimension of the features by using a pooling layer, reducing the calculated amount and retaining key information, converting the feature vectors into pattern recognition output by using a full-connection layer, setting an optimizer as Adam in the training process, adopting a loss function by adopting cross-entropy, adjusting the learning rate and the batch size, optimizing the model performance, and generating a pattern recognition result;
The learning optimization submodule applies a depth Q network to optimize a decision strategy based on a mode identification result, evaluates the accumulated benefits of the decision by defining a reward function, optimizes the learning efficiency by using an experience replay mechanism, optimizes the stability of the learning process by adopting a target network, relieves the problem of target movement by periodically updating the weight of the target network, refines and adjusts the learning rate, the updating frequency and the discount factor gamma, matches the change of the production environment and generates an optimized decision strategy;
The strategy adjustment submodule carries out real-time adjustment of strategies in continuous change of the production environment based on optimized decision strategies, compares the optimized strategies with the current production conditions by monitoring real-time production data, identifies strategy parameters needing to be adjusted, adopts a self-adaptive adjustment mechanism, dynamically modifies production decision parameters, including production speed and raw material input proportion, and generates a cognition optimization decision result.
The invention improves that the network and stability management module comprises a network analysis sub-module, a key node optimization sub-module and a fragile link improvement sub-module;
The network analysis submodule analyzes a complex network constructed in the production process based on a cognition optimization decision result, performs centrality analysis and a community detection algorithm, calculates the direct connection number of each node by adopting centrality, approaches to the average distance from a centrality measurement node to all other nodes, evaluates the importance of the nodes on the path between node pairs by intermediation centrality, identifies communities or groups with dense resources and information flow in the network by a modularity optimization method, comprises setting algorithm parameters by Gephi, revealing key structural characteristics of the production network, and generates a network structure analysis result;
The key node optimizing submodule optimizes the identified key nodes by adopting a modularized optimizing technology based on a network structure analysis result, and comprises the steps of optimizing the connectivity of the key nodes or improving the positions of the key nodes in a network by adopting shortest path optimization or network traffic redistribution, so that the nodes can coordinate resources and information flows in a production network, and a key node optimizing strategy is generated;
The fragile link improvement submodule identifies and improves the fragile links in the network based on the key node optimization strategy, identifies the areas which are easily affected due to weak structure or excessive concentration of resources by analyzing the flow connectivity and the resource distribution of the network, adopts the measures of adding redundant connection, balancing the resource distribution and adjusting the structural design of the fragile links, and generates the network optimization strategy.
The invention is improved in that the microecology regulation and control optimization module comprises a community monitoring sub-module, a performance analysis sub-module and a microecology management optimization sub-module;
the community monitoring submodule is used for carrying out high-throughput sequencing on samples through a IlluminaMiSeq platform based on a network optimization strategy, processing sequencing data by using QIIME2 software, clustering operation classification units, setting a 97% similarity threshold, carrying out microorganism abundance analysis by using a feature-tablesummarize command, carrying out microorganism classification by using a feature-CLASSIFIERCLASSIFY-sklearn command, revealing microorganism distribution and diversity in communities, and generating a microorganism community composition analysis result;
the performance analysis submodule builds an analysis result based on a microbial community, adopts a machine learning model, builds a random forest model through a Scikit-learn library of Python, sets an n_ estimators parameter as 100 to represent the number of trees, and a max_depth parameter as 10 to limit the maximum depth of the trees, uses a fit method to carry out model training, adopts a predict method to predict cooling efficiency, adopts a feature_ importances _attribute to evaluate the importance of the types of microorganisms, identifies microorganisms with influence on the cooling efficiency, and generates a cooling efficiency and microorganism interaction analysis result;
The micro-ecological management optimization submodule designs a microorganism regulation and control strategy based on the analysis result of the cooling efficiency and the microorganism interaction, adopts a decision tree algorithm to select regulation and control measures, uses DecisionTreeClassifier of a Scikit-learn library, sets criterion parameters as gini to evaluate splitting quality, uses a splitter parameter as best to capture optimal splitting, trains regulation and control strategy data by a fit method, and uses a predict method to adjust the strategy according to real-time monitoring data to generate a microorganism regulation and control scheme.
The invention is improved in that the decision support and strategy optimization module comprises a data integration sub-module, an analysis and evaluation sub-module and a comprehensive scheme generation sub-module;
The data integration submodule carries out comprehensive processing on data based on a preliminary dynamic model, optimized production parameters, a production path optimization scheme, a cognitive optimization decision result, a network optimization strategy and a microorganism regulation scheme, merges the data of multiple sources through a Pandas library, associates key fields of a data set by using a dataframe method, removes missing values by using a dropna method on the merged data set, eliminates repeated records by using a drop duplicates method, and generates an integrated data set;
The analysis and evaluation sub-module adopts a hierarchical analysis process to evaluate the quality of the decision scheme based on the integrated data set, a hierarchical structure model is constructed to be divided into a target layer, a criterion layer and a scheme layer, a compact method of a ahpy library is used for quantifying the relative importance among each layer, consistency of a consistency index check decision matrix is set, an optimal decision scheme is identified through a weight value obtained through calculation, and a decision scheme evaluation result is generated;
The comprehensive scheme generating submodule adopts a multi-objective decision analysis method based on the decision scheme evaluation result, selects production adjustment measures through logic judgment and weight comparison by referring to the current production demand and resource configuration, analyzes and selects adjustment measures capable of meeting a plurality of production targets simultaneously by utilizing pareto front analysis, and generates a comprehensive adjustment scheme.
The invention is improved in that the dynamic response and adjustment module comprises a flow adjustment sub-module, an operation optimization sub-module and an environment matching sub-module;
The flow adjustment submodule is based on a comprehensive adjustment scheme, adopts a fuzzy logic control method to adjust the production flow in real time, constructs a fuzzy logic controller, takes the production speed and the raw material input quantity as input variables, defines output variables as adjusted production parameters, sets membership functions and rule bases of the input variables by utilizing a fuzzy module in Matlab or Python, processes the input data through fuzzy reasoning, selects operation parameter values by applying a centroid method, matches the change of real-time production requirements, and generates a flow adjustment strategy;
The operation optimization submodule executes rolling time domain optimization based on a flow adjustment strategy, continuously optimizes operation parameters, sets a rolling time domain frame, updates an objective function and constraint by using current and predicted production data, applies a genetic algorithm, executes the operation optimization by using a deap library of Python, and adjusts algorithm parameters including population size, cross rate and mutation rate, wherein the objective function design simultaneously maximizes production efficiency and minimizes cost, constraint conditions include machine capacity and quality requirements, and adjusts the operation parameters according to instant data after each rolling window is finished to generate an operation optimization scheme;
The environment matching sub-module is used for implementing environment response measures based on an operation optimization scheme, coping with sudden events in a production environment, adopting a Python change detection technology to identify key change points in the production process, dynamically adjusting a production plan and resource allocation according to the detected event type and urgency, adopting a linear programming technology, optimizing through linprog functions of a SciPy library in the resource allocation, matching and recovering to an optimized production state, and generating an adaptive adjustment scheme of the production process.
Compared with the prior art, the invention has the advantages and positive effects that:
In the invention, the intelligent level and adaptability of the steel production process are obviously improved by integrating advanced modeling, simulation, optimization and decision support technologies. The combination of dynamic system modeling and simulation, including the application of continuous time Markov chains and random differential equations, overcomes the shortcomings of the traditional system in the aspect of accurately simulating random fluctuation and complex dynamics in the production process. The system can more accurately predict the change of the production process, and adjust the model parameters in real time, so that the system effectively responds to the quality fluctuation of raw materials and the change of energy cost. The introduction of the hybrid system theory further optimizes the complex decision process involving continuous and discrete action variables, and realizes the optimal dynamic adjustment of production parameters through model predictive control and hybrid integer linear programming technology. The production efficiency is improved, and the product quality and the process safety are ensured. The application of the differential topology method brings a new view angle in the optimization of logistics and information flow paths, and by identifying bottlenecks and redundant links in the production process, a more efficient and concise production path is designed, so that the efficiency and the simplification degree of the production flow are remarkably improved. The introduction of the cognitive dynamic system and the combination of the deep learning and the reinforcement learning algorithm not only improve the accuracy and the efficiency of decision making, but also enable the system to self-adaptively learn and optimize the production process, thereby realizing the dynamic response to the change of the production environment. The network science is applied to management and optimization of complex production flows, and the stability of the production network and the efficiency of resource allocation are enhanced by identifying and optimizing key nodes and fragile links. The cooling water system optimization strategy based on the dynamic microbial community effectively controls the formation of the biological film by monitoring and regulating the microbial composition in real time, reduces the energy consumption and the maintenance cost, and improves the reliability and the operation efficiency of the system. The integration of the series of innovative schemes not only optimizes the steel production process and improves the production efficiency and the product quality, but also realizes the efficient utilization of energy and resources.
Drawings
FIG. 1 is a block diagram of an intelligent digital twin simulation system for a steel production process according to the present invention;
FIG. 2 is a system frame diagram of an intelligent digital twin simulation system for a steel production process according to the present invention;
FIG. 3 is a schematic diagram of an integrated modeling and simulation module in an intelligent digital twin simulation system for a steel production process according to the present invention;
FIG. 4 is a schematic diagram of a production optimization and regulation module in an intelligent digital twin simulation system for a steel production process according to the present invention;
FIG. 5 is a schematic diagram showing a flow and morphology analysis module in an intelligent digital twin simulation system for steel production process according to the present invention;
FIG. 6 is a schematic diagram of a cognitive dynamic decision module in an intelligent digital twin simulation system for a steel production process;
FIG. 7 is a schematic diagram of a network and stability management module in an intelligent digital twin simulation system for steel production process according to the present invention;
FIG. 8 is a schematic diagram of a micro-ecological regulation optimization module in an intelligent digital twin simulation system for a steel production process;
FIG. 9 is a schematic diagram showing a decision support and strategy optimization module in an intelligent digital twin simulation system for steel production process;
FIG. 10 is a schematic diagram showing the dynamic response and adjustment module in the intelligent digital twin simulation system for the steel production process according to the present invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
In the description of the present invention, it should be understood that the terms upper, lower, front, rear, left, right, and the like of the length-width indicate the orientation or the positional relationship based on the orientation or the positional relationship shown in the drawings, are merely for convenience of describing the present invention and simplifying the description, and do not indicate or imply that the apparatus or element to be referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as limiting the present invention. Furthermore, in the description of the present invention, plural means two or more unless specifically defined otherwise.
Examples:
Referring to fig. 1, the present invention provides a technical solution: the intelligent digital twin simulation system for the steel production process comprises an integrated modeling and simulation module, a production optimization and regulation module, a flow and morphology analysis module, a cognitive dynamic decision module, a network and stability management module, a microecology regulation and control optimization module, a decision support and strategy optimization module and a dynamic response and regulation module;
The integrated modeling and simulation module is based on the physical and chemical processes of steel production, adopts a continuous time Markov chain model to describe the state transition probability in the production process, introduces a random differential equation method, simulates random fluctuation in the process, adjusts model parameters according to real-time monitoring data through maximum likelihood estimation, optimizes a model structure by using a genetic algorithm, reflects the current production process and generates a preliminary dynamic model;
The production optimization and regulation module adopts a model prediction control method based on a preliminary dynamic model, designs a prediction model to predict future production trend, optimizes continuous and discrete production control variables by combining a mixed integer linear programming algorithm, and matches the changes of raw material quality and energy cost to generate optimized production parameters;
The flow and morphology analysis module adopts a differential topology method to analyze the topological structure of the material flow and the information flow based on the optimized production parameters, uses Tonlon analysis to identify key structural features and bottlenecks in the production process, and adopts a local linear embedding method to optimize the production path so as to generate a production path optimization scheme;
the cognitive dynamic decision module is based on a production path optimization scheme, adopts a convolutional neural network to analyze historical production data and current state, identifies modes and trends, utilizes a deep Q network to perform strategy optimization and decision making in a cyclically-changed production environment, and generates a cognitive optimization decision result;
The network and stability management module analyzes the complex network constructed in the production process by adopting a centrality analysis and community detection algorithm based on the cognition optimization decision result, identifies key nodes and fragile links in the network, adjusts the network structure by a modularized optimization technology, and generates a network optimization strategy;
The micro-ecological regulation and control optimization module is based on a network optimization strategy, adopts a DNA sequencing technology and a quantitative polymerase chain reaction method, analyzes the microbial community composition in cooling water, combines a machine learning model, analyzes the interaction between the microbial composition and cooling efficiency, and guides the formulation of a regulation and control strategy to generate a microbial regulation and control scheme;
The decision support and strategy optimization module integrates data by adopting a data fusion technology based on a preliminary dynamic model, optimized production parameters, a production path optimization scheme, a cognitive optimization decision result, a network optimization strategy and a microorganism regulation scheme, evaluates the advantages and disadvantages of various decision schemes by using a hierarchical analysis process, and generates a comprehensive regulation scheme;
The dynamic response and adjustment module adopts fuzzy logic control to adjust the operation parameters of the production line in real time based on the comprehensive adjustment scheme, combines detection and response of sudden events and rolling time domain optimization, continuously optimizes decision and control strategies in the production process, and generates an adaptive adjustment scheme of the production process.
The preliminary dynamic model comprises probability distribution of production state variables, fluctuation mode of time sequence, response function of key nodes of the production process, optimized production parameters comprise regulated furnace temperature setting, carbon and oxygen input proportion and cooling rate in the steelmaking process, a production path optimization scheme comprises simplified material handling flow, recombined production line layout and an optimized information feedback mechanism, a cognitive optimization decision result comprises a raw material purchase plan based on trend prediction, a self-adaptive regulation rule of production scheduling and a decision framework of emergency response, a network optimization strategy comprises optimized production core network connection, logistics distribution route and information flow transparency measures, a microorganism regulation scheme comprises selected anti-biological pollution chemicals, a periodic microbial community monitoring plan and periodic frequency of cooling water cleaning and maintenance, a comprehensive regulation scheme comprises production efficiency optimization measures, an energy consumption regulation strategy and an environment emission control scheme, and a production process adaptation regulation scheme comprises updated production efficiency index, an external change response regulation record and improved product quality.
In the integrated modeling and simulation module, a continuous time Markov chain model and a random differential equation method are adopted, and the model is optimized by utilizing a genetic algorithm in combination with real-time monitoring data. A specific operational procedure involves describing transition probabilities of states in a production process using a continuous time markov chain model, which involves the definition of a state space and the construction of a transition matrix, where each state corresponds to a specific stage in the production process and the transition matrix contains probabilities of transitioning from one state to another. The random differential equation method is used for simulating random fluctuation in the production process, numerical solution is carried out through the Euler-Maruyama method, and fluctuation caused by various uncertain factors in the production process is simulated. The maximum likelihood estimation is used for adjusting model parameters according to the real-time monitoring data, so that the model can accurately reflect the current production state. The genetic algorithm is used for optimizing the model structure, and the model structure is continuously and iteratively updated by defining fitness functions, executing selection, crossing and mutation operations so as to better adapt to the dynamic change of the production process. The preliminary dynamic model finally generated by the module not only accurately describes the state transition probability and random fluctuation in the production process, but also realizes the high reflection and prediction of the current production process through the real-time adjustment and structural optimization of model parameters, and provides an accurate data basis for the follow-up production optimization and regulation.
In the production optimization and regulation module, based on a preliminary dynamic model, a model predictive control method and a mixed integer linear programming algorithm are adopted to optimize production control variables. The model prediction control method predicts the production trend in a period of time in the future by constructing a prediction model, and comprises the steps of defining a control target, constructing the prediction model and formulating a control strategy. The control target is set according to the production requirement, such as maximizing the output or minimizing the energy consumption, the prediction model is based on the preliminary dynamic model, and the basis is provided for control decision by simulating the future production state. The mixed integer linear programming algorithm is used for accurately optimizing continuous and discrete production control variables, the operation process comprises an objective function for defining an optimization problem and constraint conditions, the objective function reflects optimization targets such as production cost minimization, and the constraint conditions cover physical and chemical limitations in the production process. The module not only matches the change of raw material quality and energy cost, but also provides an optimal regulation strategy for the production process through fine prediction and optimization operation, and remarkably improves the production efficiency and economic benefit.
In the flow and morphology analysis module, a differential topology method and a local linear embedding method are adopted to optimize a production path based on optimized production parameters. The differential topology method identifies key structural features and bottlenecks in the production process by analyzing the topological structures of the logistics and the information flows, and comprises the steps of constructing topological models of the logistics and the information flows and revealing the structural features by applying Torenia analysis. The local linear embedding algorithm is used to reconstruct the production path in the low-dimensional space, optimize the global structure while maintaining local similarity, and the operation process includes selecting proper neighborhood size, calculating the linear relation between each point and its neighborhood, and searching the low-dimensional representation through the optimization technology. The module not only simplifies the material handling flow and reorganizes the layout of the production line through careful topology analysis and path reconstruction, but also optimizes an information feedback mechanism, thereby providing powerful support for improving the efficiency and flexibility of the production process.
In the cognitive dynamic decision module, based on a production path optimization scheme, a convolutional neural network and a deep Q network are adopted to perform pattern recognition and strategy optimization. The convolutional neural network extracts key features by analyzing historical production data and current states, and identifies patterns and trends in the production process, including designing network structures, selecting activation functions, extracting features and identifying patterns. The deep Q network is then used for policy optimization based on the identified patterns and trends, and the operational process includes defining a reward function, optimizing decision strategies through iterative training using empirical replay and a target network optimization learning process. The intelligent optimization of the production decision is realized by the module through an advanced machine learning technology, the self-adaptability and the decision efficiency of the production process are improved, and an effective coping strategy is provided for the cyclic variation in the production environment.
In the network and stability management module, a complex network constructed in the production process is carefully analyzed and optimized by adopting a centrality analysis and community detection algorithm. Centrality analysis is used to identify key nodes in a network, including computing the centrality, near centrality, and intermediate centrality of nodes, these metrics helping to reveal the importance and impact of nodes in the network. The community detection algorithm is used to discover modules or communities in the network, including applying a modularity optimization method to divide the network into communities to more effectively manage resources and information flows. In addition, modular optimization techniques are used to adjust network architecture to optimize resource allocation and flow connectivity by improving connectivity and location of critical nodes. The operation result of the module is that a network optimization strategy is generated, so that the stability and the robustness of a production network are enhanced, the efficiency and the response capability of the whole production system are improved, and powerful network support is provided for rapidly adapting to market changes and internal demand changes.
In the micro-ecological regulation and control optimization module, based on a network optimization strategy, a DNA sequencing technology and a quantitative polymerase chain reaction method are utilized, and a machine learning model is combined to carry out deep analysis on the microbial community composition in the cooling water and guide the formulation of the regulation and control strategy. DNA sequencing technology and quantitative polymerase chain reaction methods are used to accurately identify and quantify the type and quantity of microorganisms in cooling water, including sampling, DNA extraction, sequencing, and data analysis, which ensure the accuracy and reliability of microbial community analysis. Machine learning models, particularly random forest models, are used to analyze interactions between microorganism composition and cooling effectiveness, including feature selection, model training, and effectiveness prediction, to identify the microorganism species that have the greatest impact on cooling effectiveness. Based on these analysis results, microbiological control programs were designed, including selection of appropriate chemicals, adjustment of cleaning and maintenance frequency, to optimize the performance and efficiency of the cooling water system. The execution of the module not only improves the operation efficiency and reliability of the cooling system, but also prolongs the service life of the equipment and reduces the maintenance cost through fine microorganism management.
In the decision support and strategy optimization module, a data fusion technology and a hierarchical analysis process are adopted based on a preliminary dynamic model and other optimization schemes, the decision scheme is comprehensively evaluated, and strategy support is provided for adjustment and optimization of the production process. Data fusion techniques are used to integrate data sets from different sources, including data cleansing, merging, and preprocessing, to ensure that decision analysis is based on comprehensive and accurate information. The analytic hierarchy process is used for evaluating and comparing the merits of different decision schemes, and comprises the steps of establishing an evaluation model, determining evaluation standards and weights, and carrying out consistency test to ensure the rationality and reliability of an evaluation result. Finally, through comprehensive analysis and comparison, a comprehensive adjustment scheme comprising various measures such as production efficiency optimization, energy consumption adjustment, environmental emission control and the like is generated. The module provides powerful support for production optimization and resource allocation through accurate data analysis and scientific decision evaluation, and improves the overall efficiency and environmental sustainability of the production process.
In the dynamic response and adjustment module, fuzzy logic control, rolling time domain optimization and environmental response measures are comprehensively utilized to realize real-time adjustment and optimization of the operation parameters of the production line. The fuzzy logic controller processes fuzzy input by setting fuzzy rules and membership functions according to real-time production demand change, and dynamically adjusts production parameters such as production speed and raw material input quantity, including rule setting, fuzzy reasoning and defuzzification processes, so as to adapt to the instant demand of the production process. The rolling time domain optimization realizes double optimization of production efficiency and cost by periodically updating an objective function and constraint conditions and dynamically optimizing operation parameters by utilizing a genetic algorithm, wherein the operation parameters comprise population initialization, fitness evaluation, selection, crossover and variation. The environmental response measures are used for timely adjusting production plans and resource allocation by adopting a change detection technology aiming at sudden events in the production environment, and comprise event detection, emergency assessment and response strategy formulation, so that continuity and stability of the production process are ensured. The implementation of the module remarkably improves the adaptability and response speed of the production system, provides an effective solution for facing internal changes and external interference, and ensures the efficient and stable operation of the production process.
Referring to fig. 2 and 3, the integrated modeling and simulation module includes a model integration sub-module, a parameter estimation sub-module, and a model structure optimization sub-module;
the model integration submodule adopts a continuous time Markov chain model based on the physical and chemical processes of steel production, a state space is used for defining production states, a transition matrix designates the probability from one state to the other state, the probability description of the transition between the states is carried out through the construction of the state transition matrix, a random differential equation is introduced, and the simulation of random fluctuation is carried out through the Euler-Maruyama approximation, so that a state transition and fluctuation model is generated;
The parameter estimation submodule carries out maximum likelihood estimation based on the state conversion and fluctuation model, adjusts parameters by using a gradient descent method, maximizes a likelihood function of observed data, gradually optimizes model parameters by setting an initial parameter estimation value, a learning rate and iteration times, enables the model parameters to be matched with production data, and generates a parameter optimization model;
The model structure optimization submodule is based on a parameter optimization model, adopts a genetic algorithm to optimize a model structure, evaluates model performance by defining a fitness function, adopts selection, crossover and mutation operations, simulates a natural evolution process, selects a model with excellent performance according to the fitness function, generates a new model by crossover operation through cross point combination of randomly selected model parameters, and repeatedly executes the mutation operation until the optimal model structure is captured by randomly changing partial values in the model parameters and introducing new genetic diversity to generate a preliminary dynamic model.
In the model integration submodule, a composite model capable of accurately describing the state transition probability and random fluctuation in the steel production process is constructed through the combination of a continuous time Markov chain model and a random differential equation. The continuous time Markov chain model defines the various states in the production process using a state space approach, while a transition matrix is used to specify the probability of transitioning from one state to another. Wherein the construction of the state transition matrix is based on in-depth analysis of historical production data and comprehensive consideration of physical and chemical processes. The random fluctuation in the production process is simulated through an Euler-Maruyama approximation method by introducing a random differential equation, and the capability of describing uncertainty in the production process is added for the model. The step involves a complex numerical calculation process, including modeling and simulating randomness of influencing factors in the production process, so as to ensure that the model can accurately reflect actual conditions of the production site. Finally, the submodule generates a state transition and fluctuation model which comprehensively considers the physicochemical process, the state transition probability and the random fluctuation of the steel production, and provides a solid foundation for subsequent optimization and decision.
The parameter estimation submodule adjusts and optimizes model parameters by adopting a maximum likelihood estimation method and a gradient descent method based on the generated state transition and fluctuation model. The key of the process is that the model parameters are gradually adjusted through iterative calculation to maximize the likelihood function of the observed data, so that the model can reflect the actual condition of the production process more accurately. To achieve this goal, the submodule first sets initial estimates of the model parameters, and then adjusts these parameters step by a gradient descent method according to the set learning rate and the number of iterations. In each iteration process, the sub-module calculates likelihood function values under the current parameters, and adjusts the parameters according to the gradient direction so as to maximize the likelihood function. The implementation of this process involves complex mathematical calculations and arithmetic logic, ensuring that the model parameters can be ultimately matched to the actual production data by precise adjustment. After the process is completed, the parameter estimation submodule generates a parameter optimization model which has strong interpretation capability in theory and can provide highly accurate production process simulation in practical application.
The model structure optimization submodule adopts a genetic algorithm to carry out further structure optimization on the parameter optimization model. The submodule evaluates the performance of the model by defining an fitness function, and then simulates natural selection and genetic variation processes through selection, crossover and variation operations to continuously optimize the model structure. The selecting operation selects a model with better performance according to the fitness function value of the model, and the intersecting operation combines parameter parts of the two models through randomly selecting intersection points of model parameters to generate a new model structure. The mutation operation introduces new genetic diversity by randomly changing part of the values in the model parameters. This series of operations is repeatedly performed until an optimized model structure is found. By the mode, the model structure optimization sub-module finally generates an optimized preliminary dynamic model, and the model not only can accurately simulate the steel production process, but also can provide powerful support for production optimization and decision.
It is assumed that in the steel making process, it is necessary to construct a state transition and fluctuation model capable of simulating the temperature change and chemical composition fluctuation of molten steel. The model describes the transition of the temperature state using a continuous time Markov chain, and a random differential equation to simulate random fluctuation of the carbon content, wherein the initial temperature of molten steel is 1600 ℃, the temperature state interval is 10 ℃, the initial value of the carbon content is 0.4%, and the fluctuation of the carbon content follows a normal distribution with a standard deviation of 0.05%. The change per minute was simulated by the Euler-Maruyama approximation. Based on the actual observation data, the model parameters are optimized by using a maximum likelihood estimation and gradient descent method, the initial temperature transition probability is estimated to be 0.8, the standard deviation of the carbon content fluctuation is estimated to be 0.1%, and the final optimization is iterated until the temperature transition probability is 0.85 and the standard deviation of the carbon content fluctuation is 0.04%. Further using genetic algorithm to optimize the model structure, and finding the optimal model structure after 200 generations of iteration by defining fitness function, selection, crossing and mutation operation, so as to accurately simulate physical and chemical changes in steel production and improve the optimization and adaptability of the production process.
Referring to fig. 2 and 4, the production optimization and regulation module includes a control strategy sub-module, a parameter adjustment sub-module, and an external factor response sub-module;
The control strategy submodule carries out a model prediction control method based on a preliminary dynamic model, builds a control and prediction framework to predict future production trend, adjusts a control strategy according to the model prediction control method, sets a control target by defining a state space representation of the prediction model, predicts a production process of each time period by utilizing a rolling time window technology, adopts a linear programming solver to determine optimal setting of production parameters under given constraint conditions, dynamically adjusts operation parameters of a production line, and corresponds to predicted production requirements and raw material supply conditions to generate a prediction control scheme;
The parameter adjustment submodule is based on a predictive control scheme, adopts a mixed integer linear programming algorithm to refine and optimize production control variables, sets an optimization objective function to minimize production cost by defining an optimization problem comprising continuous and discrete decision variables and meets production and safety constraint conditions, applies Gurobi to solve the problem, and generates a production parameter optimization scheme by adjusting algorithm parameters including a boundary or cutting plane strategy of a branch-and-bound method and optimizing a solution process;
The external factor response submodule analyzes and responds to external factors based on a production parameter optimization scheme, potential influences of the external factors on a production process are evaluated through collecting and analyzing market demand change, raw material price fluctuation and energy supply condition information and using sensitivity analysis, a scene analysis method is adopted to simulate production results under various conditions, and a production strategy is adjusted, wherein the method comprises the steps of changing a raw material use plan and adjusting energy utilization efficiency in the production process, and optimized production parameters are generated.
In the control strategy sub-module, a Model Predictive Control (MPC) method is performed to build a control and prediction framework based on the preliminary dynamic model, predict future production trends, and adjust the control strategy accordingly. In particular, the submodule defines a state space representation of a predictive model, which representation includes various variables and parameters of the production process, such as the temperature of the molten steel, the proportion of chemical components, the pressure in the furnace, etc., and the law of variation thereof over time. The control targets are set to maximize yield or minimize energy consumption, etc., which requires that the model be able to take into account the trade-off of multiple production targets at the same time. Using the rolling time window technique, the sub-module makes predictions of the production process for each future time period that take into account constraints such as device capabilities and environmental criteria. The predicted results are then used in a linear programming solver to determine the optimal production parameter settings under given constraints. Adjustment of the control strategy is based on the output of the MPC framework to dynamically adjust the operating parameters of the production line to account for the predicted production demand and raw material supply conditions. By the method, a predictive control scheme is generated, the scheme can dynamically optimize the production process, improve the production efficiency and the resource utilization rate, and simultaneously reduce the energy consumption and the production cost.
The parameter adjustment submodule adopts a Mixed Integer Linear Programming (MILP) algorithm to refine and optimize production control variables based on a predictive control scheme. In this sub-module, the objective function is set to minimize production costs while meeting production and safety constraints by defining an optimization problem that includes both continuous and discrete decision variables. The MILP problem is solved by using Gurobi optimizers, in which algorithm parameters, such as boundary of branch-and-bound method or cut plane strategy, are carefully tuned to optimize the solving process, speeding up the solving and improving the solving quality. By refining and optimizing the production control variables, the submodule generates a production parameter optimization scheme, which further reduces the production cost and ensures the stability and safety of the production flow. While minimizing costs.
The external factor response submodule is used for carrying out deep analysis and response on the external factor based on the production parameter optimization scheme. By collecting and analyzing information on market demand changes, raw material price fluctuations and energy supply, the sub-module uses sensitivity analysis to evaluate the potential impact of these external factors on the production process. Based on the results of the scene analysis, the production strategy is adjusted, including changing the usage plan of raw materials and adjusting the energy utilization efficiency in the production process to accommodate external changes and optimize production parameters. In this way, the external factor coping submodule generates an optimized production parameter scheme which not only considers the efficiency and cost of the internal production process, but also considers the influence of external environment changes, thereby ensuring the flexibility and adaptability of the production system.
Suppose a steel plant needs to optimize the operation of its blast furnace to maximize production and minimize energy consumption. And executing a model prediction control method based on the preliminary dynamic model. The state space representation of the model includes: the molten steel temperature is 1500 ℃ to 1600 ℃, the carbon content is 0.2 to 0.8 percent, and the furnace pressure is 1 to 5atm. The optimal production parameters, such as gas flow and raw material input proportion, are determined by a rolling time window technology and Gurobi linear programming solver. Simulation results show that the coal gas flow rate is adjusted to 5000 cubic meters per hour, and the raw material proportion is iron ore: coke=2:1, can increase production by 5% per day and reduce energy consumption by 10%. The mixed integer linear programming algorithm is further used to refine the optimization control variables with the goal of minimizing production costs including raw materials, energy consumption, and operating costs. After optimization, the raw material ratio is adjusted to 2.2:1, the gas flow is reduced to 4800 cubic meters per hour, and the cost is reduced by 2% every day. In face of external factors of increasing market demands and rising raw material prices, iron ores are prepared by adjusting the raw material ratio: coke=2.1:1, increasing gas flow to 4900 cubic meters per hour, meeting the growing demand and controlling the cost increase within 5%, keeping the enterprise profitability and market competitiveness.
Referring to fig. 2 and 5, the flow and morphology analysis module includes a topology analysis sub-module, a bottleneck identification sub-module, and a path optimization sub-module;
The topological structure analysis submodule performs a differential topological method to refine the structures of the logistics and the information flow based on the optimized production parameters, identifies key topological characteristics in the assembly by constructing the representation of the logistics and the information flow in a multidimensional space and applying coherent group and Betty number calculation, reveals potential connectivity and barrier points in the logistics and the information flow by a ring and a cavity, and comprises the steps of calculating topological invariants by MATLAB to generate a topological structure analysis result;
The bottleneck recognition sub-module is used for analyzing key structural features and bottlenecks in the production process by adopting a Tonlun analysis method based on a topological structure analysis result, recognizing persistent structural features influencing the production process by comparing topological models under various working conditions, selecting a part of the production process which still keeps the structural characteristics unchanged under small fluctuation by utilizing path connectivity and space deformation, constructing and analyzing a mathematical model by using a SciPy library of Python, recognizing the key bottlenecks obstructing the production efficiency, and generating a bottleneck recognition result;
The path optimization submodule optimizes the production path by using a local linear embedding method based on the bottleneck recognition result, captures the optimal representation capable of keeping the local structure in a low-dimensional space by referring to the local similarity of each link in the production process, and comprises reconstructing the production path by using a local linear embedding algorithm provided by a Scikit-learn library in Python to generate a production path optimization scheme.
In the topological structure analysis submodule, the structures of the logistics and the information flow are subjected to refinement analysis through a differential topological method. First, the sub-module collects and processes logistics and information flow data on the production line, including but not limited to input and output records of various stations on the production line, material transfer paths, production scheduling information, and the like. These data are converted into representations in a multidimensional space, expressing the interrelationship of streams and information streams in the form of points, lines and planes. And calculating the coherent group and Betty number by utilizing MATLAB, and identifying rings and holes in the complex network by establishing and analyzing the topological structure of the complex network. The rings represent a cyclic dependence in the stream or information stream and the voids represent potential points of fracture or areas of inefficiency. From these calculations, key topological features in the production flow, such as connectivity and potential obstacle points, are revealed. The analysis result generated by the refinement operation is a detailed topology report including topology characteristics of the line flow and the information flow, key nodes, potential bottleneck areas, etc.
The bottleneck recognition sub-module is based on the topological structure analysis result, and further refines and analyzes key structural features and bottlenecks in the production process. By adopting the homotopy analysis method, the submodule compares topological models under different working conditions, and identifies and locates the lasting structural characteristics affecting the production flow efficiency. The analysis of path connectivity and spatial deformations is performed by constructing a mathematical model from the SciPy library of Python, which identifies the parts of the production flow that remain unchanged in their structural characteristics, i.e. stable bottleneck regions. These bottlenecks are due to physical space limitations, production equipment capacity limitations, or unreasonable production schedules. The bottleneck identification result is a detailed bottleneck analysis report containing all identified bottlenecks, locations, properties and effects on production efficiency.
The path optimization submodule optimizes the production path by using a local linear embedding method based on the bottleneck recognition result. By analyzing the local similarity of each link in the production process, an optimal representation of the local structure that can be maintained in a low-dimensional space is captured, thereby reconstructing a more efficient production path. The local linear embedding algorithm is realized by utilizing the Scikit-learn library in Python, and the process not only optimizes the production path, but also considers the flow efficiency of materials and the balance of the production line. The optimized production path reduces unnecessary material transportation and treatment steps, and improves the overall efficiency of the production line. The generated production path optimization scheme is a detailed optimization report, and comprises the comparison of production paths before and after optimization, expected efficiency improvement and suggestion for implementing the optimization scheme.
It is assumed that a production line of a steel plant includes main processes of raw material preparation, iron making, steel making, continuous casting, steel rolling, and the like. For analysis of the topology of the streams and information streams, data is collected including raw material transport paths, production scheduling instructions, streams between processes, and information streams. These data are converted into vector and matrix forms in multidimensional space, expressed as: raw material arrival frequency: every hour; treatment time per procedure: ironmaking for 2 hours/batch, steelmaking for 1.5 hours/batch, continuous casting for 1 hour/batch, and rolling for 2 hours/batch; material and information flow paths: raw material preparation, iron making, steelmaking, continuous casting and steel rolling; and calculating the coherent group and the Betty number by utilizing MATLAB, and identifying the efficiency barrier points from raw material preparation to the ironmaking stage and from steelmaking to the continuous casting stage. The use of the SciPy library of Python for homotopy analysis further determined that the raw material supply was unstable and the scheduling delay was a key factor in inefficiency. The production path is optimized through a local linear embedding method and the Scikit-learn library, so that the raw material supply scheduling and the information flow and logistics scheduling from steelmaking to continuous casting are improved, and the waiting time is reduced. These optimizations reduced the overall production time from 8 hours to 7 hours, improving production efficiency by 15%. The optimized production path scheme report details the measures, expected benefits and implementation steps, and provides a clear improvement plan and guidance for the efficiency improvement of the steel plant.
Referring to fig. 2 and 6, the cognitive dynamic decision module includes an information processing sub-module, a learning optimization sub-module, and a policy adjustment sub-module;
The information processing sub-module is used for executing a convolutional neural network based on a production path optimization scheme, analyzing historical production data and a current state, and by constructing a convolutional neural network model, extracting spatial features in the data by using a convolutional layer, following an activation layer after each layer of convolution, using the nonlinear processing capacity of a ReLU function optimization model, reducing the spatial dimension of the features by using a pooling layer, reducing the calculated amount and retaining key information, converting the feature vectors into pattern recognition output by using a full-connection layer, setting an optimizer as Adam in the training process, adopting a cross-entcopy as a loss function, adjusting the learning rate and the batch size to optimize the model performance, and generating a pattern recognition result;
The learning optimization submodule applies a depth Q network to optimize a decision strategy based on a mode identification result, evaluates the accumulated benefits of the decision by defining a reward function, optimizes the learning efficiency by using an experience replay mechanism, optimizes the stability of the learning process by adopting a target network, relieves the problem of target movement by periodically updating the weight of the target network, refines and adjusts the learning rate, the updating frequency and the discount factor gamma, matches the production environment change, and generates an optimized decision strategy;
The strategy adjustment submodule carries out real-time adjustment of strategies in continuous change of the production environment based on the optimized decision strategy, compares the optimized strategies with the current production conditions by monitoring real-time production data, identifies strategy parameters needing to be adjusted, adopts a self-adaptive adjustment mechanism, dynamically modifies production decision parameters, comprises production speed and raw material input proportion, and generates a cognition optimization decision result.
The information processing sub-module analyzes the historical production data and the current state by constructing a Convolutional Neural Network (CNN) model. The model utilizes a convolution layer to extract spatial features in data, and each layer of convolution is followed by an activation layer to use a ReLU function to enhance the nonlinear processing capacity of the model, and a pooling layer is used to reduce the spatial dimension of the features, reduce the calculated amount and retain key information. The full connection layer converts the extracted feature vector into an output of pattern recognition. The data format includes time series production parameters such as temperature, pressure, chemical composition ratio, etc., and quality index of the production result. The data are standardized and then input into a CNN model for training, and the model performance is optimized by adjusting the learning rate and the batch size through an Adam optimizer and a cross-entopy loss function. In the execution process, the CNN model gradually improves the recognition accuracy of the production process state through repeated iterative training, and finally generates a mode recognition result. The result can accurately reflect the production state and potential quality problems.
The learning optimization submodule adopts a Deep Q Network (DQN) to optimize a decision strategy based on the mode identification result of the information processing submodule. DQN uses an empirical replay mechanism to optimize learning efficiency by defining cumulative benefits of the reward function evaluation decisions while employing a target network to improve the stability of the learning process. In this process, the model inputs include the current production state and optional operating strategies, with the outputs being the expected benefits of each strategy. The reward function is defined based on the production efficiency and product quality, and the like, with the goal of finding an operation strategy that maximizes production benefits by learning. The DQN model adjusts the strategy stepwise by interactive learning to accommodate changes in the production environment. In the process, the model is continuously updated to capture complex relations and dynamic changes in the production process and generate an optimized decision strategy.
The strategy adjustment submodule carries out strategy real-time adjustment in continuous change of the production environment based on the optimization decision strategy generated by the learning optimization submodule. The submodule compares the current production state with the optimized strategy by monitoring the real-time production data, and identifies the strategy parameters to be adjusted. Through the self-adaptive adjustment mechanism, the module dynamically modifies production decision parameters such as production speed, raw material input proportion and the like so as to adapt to the change in the production process. In the process, the module continuously collects feedback, evaluates the adjustment effect and ensures that the strategy is always kept in an optimal state.
The data set of the production line is assumed to contain daily production parameters and product quality records for the past year. Including date, time, molten steel temperature (1500 ℃ to 1600 ℃), carbon content (0.2% -0.8%), pressure (1-5 atm) and product quality rating (A, B, C). These data are used to train a Convolutional Neural Network (CNN) model in order to identify patterns of production parameters that lead to degradation of product quality. The CNN model structure contains three convolutional layers, each followed by a ReLU activation layer and a max pooling layer, and two fully connected layers, ultimately outputting the probability distribution of the product quality ratings. The model uses Adam optimizer and cross-entopy loss function to reach 92% validation set accuracy. Based on this, a Deep Q Network (DQN) is applied to optimize production decisions to maximize high quality product yield. The DQN model, the structure comprises two hidden layers and a ReLU activation function, and the strategy for improving the product quality is successfully learned through 2000 training periods. The real-time monitoring system automatically adjusts production parameters according to the output of the DQN model, for example, the temperature of molten steel is reduced when the carbon content is increased, the fluctuation of product quality is effectively reduced, and the production stability and reliability are improved.
Referring to fig. 2 and 7, the network and stability management module includes a network analysis sub-module, a key node optimization sub-module, and a weak link improvement sub-module;
The network analysis submodule analyzes a complex network constructed in the production process based on a cognition optimization decision result, performs centrality analysis and a community detection algorithm, calculates the direct connection number of each node by adopting centrality, approaches to the average distance from a centrality measurement node to all other nodes, evaluates the importance of the nodes on the path between node pairs through the intermediation centrality, identifies communities or groups with dense resources and information flow in the network through a modularity optimization method, and comprises the steps of setting algorithm parameters by Gephi, revealing key structural characteristics of the production network and generating a network structure analysis result;
the key node optimizing submodule optimizes the identified key nodes by adopting a modularized optimizing technology based on a network structure analysis result, and the method comprises the steps of optimizing the connectivity of the key nodes or improving the positions of the key nodes in a network by adopting shortest path optimization or network traffic redistribution, so that the nodes can coordinate resources and information flows in a production network, and generating a key node optimizing strategy;
The fragile link improvement submodule identifies and improves the fragile links in the network based on the key node optimization strategy, identifies the areas which are easily affected due to weak structure or excessive concentration of resources by analyzing the flow connectivity and the resource distribution of the network, adopts the measures of adding redundant connection, balancing the resource distribution and adjusting the structural design of the fragile links, and generates the network optimization strategy.
In the network analysis submodule, the aim is to deeply understand a complex network structure constructed in the steel production process by using a centrality analysis and community detection algorithm, and identify key nodes and dense communities or groups in the network so as to provide a direction for optimization. First, data is collected and consolidated in the form of various links and information streams in the production process, including but not limited to logistics and information flow data between links such as raw material procurement, iron making, steel making, continuous casting, and steel rolling. And adopting Gephi and other network analysis tools to execute degree centrality analysis to calculate the direct connection number of each node, measuring the average distance from the node to all other nodes by the proximity centrality analysis, and evaluating the importance of the node on paths between other node pairs by the intermediate centrality analysis. The community detection algorithm further reveals communities or groups with dense resource and information flow, and provides key insight for subsequent optimization. The network structure analysis result generated by the process is a detailed report, which contains the key structure characteristics, key nodes and community distribution situation of the network.
The key node optimization submodule then adopts a modularized optimization technology to improve the connectivity or the position of the identified key nodes in the network based on the result of the network analysis submodule. The process includes evaluating the impact of critical nodes on production efficiency, determining shortest path optimization or network traffic redistribution policies, and implementing these policies to improve the efficiency of the nodes and mobility throughout the network. The submodule ensures that the key nodes can coordinate resources and information flows more effectively in the production network through algorithm optimization, reduces production delay and improves overall efficiency. The optimized key node strategies are presented in a report form, and the optimization measures, expected benefits and implementation suggestions are specified.
The goal of the vulnerability development sub-module is to identify and improve vulnerabilities in the network based on a key node optimization strategy. This includes in-depth analysis of production flow connectivity and resource allocation, identifying areas that are susceptible to structural weakness or excessive concentration of resources, and implementing measures to join redundant connections, balance resource allocation, and adjust the structural design of the frangible links. The sub-module improves the stability and toughness of the network by refining the operation, reducing the risk of production breaks. The implementation of the improvement and the impact on production network stability are recorded in detail in a network optimization strategy report, providing a complete solution to improve the efficiency and stability of the overall production process.
Referring to fig. 2 and 8, the micro-ecological regulation and control optimization module includes a community monitoring sub-module, a performance analysis sub-module and a micro-ecological management optimization sub-module;
The community monitoring submodule is used for carrying out high-throughput sequencing of samples through a IlluminaMiSeq platform based on a network optimization strategy, processing sequencing data by using QIIME2 software, clustering operation classification units, setting a 97% similarity threshold, carrying out microorganism abundance analysis by using a feature-tablesummarize command, carrying out microorganism classification by using a feature-CLASSIFIERCLASSIFY-sklearn command, revealing microorganism distribution and diversity in communities, and generating a microorganism community composition analysis result;
The performance analysis submodule builds a random forest model through a Scikit-learn library of Python based on a microbial community composition analysis result, sets an n_ estimators parameter as 100 to represent the number of trees, and a max_depth parameter as 10 to limit the maximum depth of the trees, performs model training by using a fit method, predicts cooling efficiency by using a predict method, evaluates the importance of microorganism types by using a feature_ importances _attribute, identifies microorganisms having an influence on the cooling efficiency, and generates a cooling efficiency and microorganism interaction analysis result;
The micro-ecological management optimization submodule designs a microorganism regulation and control strategy based on the cooling efficiency and microorganism interaction analysis result, adopts a decision tree algorithm to select regulation and control measures, uses DecisionTreeClassifier of a Scikit-learn library, sets criterion parameters as gini to evaluate splitting quality, uses the slit parameters as best to capture optimal splitting, trains regulation and control strategy data by using a fit method, and uses a predict method to adjust the strategy according to real-time monitoring data to generate a microorganism regulation and control scheme.
In the community monitoring submodule, high throughput sequencing of the microbial community in the cooling water was performed by applying DNA sequencing technology, including using Illumina MiSeq platform. The process begins with the collection of a sample of cooling water and the extraction of DNA therefrom, followed by the amplification of the 16S rRNA genes of the microorganism with specific primers to facilitate subsequent sequencing analysis. Sequencing data were processed by QIIME2 software, including quality control, clustering of Operational Taxonomies (OTUs), and analysis of microbial diversity. OTU was determined using a 97% similarity threshold to ensure accuracy of species identification. Analysis of the abundance and classification of microorganisms by the feature-table summarize command and the feature-CLASSIFIER CLASSIFY-sklearn command revealed the composition and diversity of microbial communities in the cooling water system. The microbial community generated in the process forms an analysis result, provides detailed data support for understanding the microbial ecology of the cooling system, and helps identify the microbial species affecting the system performance.
The performance analysis sub-module adopts a machine learning model, in particular a random forest model, to analyze the influence of different microorganism types on the cooling efficiency based on the result of the community monitoring sub-module. In a Python environment, a model is built through a Scikit-learn library, the parameter n_ estimators is set to be 100, the construction of 100 decision trees is represented, the max_depth is 10, and the maximum depth of the tree is limited so as to avoid over fitting. In the model training process, the input data comprise abundance data of microorganisms and corresponding cooling efficiency indexes, such as energy ratio, failure rate and the like. The model was trained using the fit method, followed by a predict method to predict cooling effectiveness, while using the feature_ importances _attribute to evaluate the extent of the effect of different microorganism species on cooling effectiveness. The analysis reveals a positive correlation between certain microorganism species and high cooling performance, while certain are related to low performance, and the generated cooling performance and microorganism interaction analysis result provide scientific basis for formulating a regulation strategy.
The micro-ecological management optimization submodule designs a micro-organism regulation strategy based on the result of the performance analysis submodule so as to optimize the performance of the cooling system. A model was constructed to select the best microbial control using decision tree algorithms, including DecisionTreeClassifier in the Scikit-learn library. In the model setup, criterion parameters were chosen to be gini, the quality of the split was evaluated, and the split parameters were set to be to capture the optimal split point. The model is trained by abundance of microorganism types and known cooling efficiency data through a fit method, and then a predict method is used for adjusting a regulation strategy according to the microorganism data monitored in real time. The resulting microbiological control program dictates that the overall effectiveness and stability of the cooling system is improved by adjusting the dosing, cleaning frequency, and other physical control measures of the cooling water treatment chemical to optimize a particular microorganism population.
Referring to fig. 2 and 9, the decision support and policy optimization module includes a data integration sub-module, an analysis and evaluation sub-module, and a comprehensive scheme generation sub-module;
the data integration sub-module performs comprehensive processing on data based on a preliminary dynamic model, optimized production parameters, a production path optimization scheme, a cognitive optimization decision result, a network optimization strategy and a microorganism regulation scheme, merges the data of multiple sources through a Pandas library, associates key fields of a data set by using a DataFrame. Merge method, removes missing values by using a dropna method on the merged data set, eliminates repeated records by using a drop duplicates method, and generates an integrated data set;
The analysis and evaluation sub-module adopts a hierarchical analysis process to evaluate the quality of the decision scheme based on the integrated data set, the hierarchical structure model is constructed to be divided into a target layer, a criterion layer and a scheme layer, the compare method of the ahpy library is used for quantifying the relative importance among each layer, consistency of a consistency index check decision matrix is set, the optimal decision scheme is identified through the weight value obtained by calculation, and a decision scheme evaluation result is generated;
The comprehensive scheme generating submodule adopts a multi-objective decision analysis method based on the decision scheme evaluation result, selects production adjustment measures through logic judgment and weight comparison by referring to the current production demand and resource allocation, analyzes and selects adjustment measures capable of meeting a plurality of production targets simultaneously by utilizing pareto front analysis, and generates a comprehensive adjustment scheme.
The data integration submodule aims to collect and arrange data from all links in the steel production process to form a unified data set for subsequent analysis. Firstly, merging the data of multiple sources such as a preliminary dynamic model, optimized production parameters, a production path optimization scheme, a cognitive optimization decision result, a network optimization strategy, a microorganism regulation scheme and the like through a Pandas library of Python. In the merging process, different data sets are associated based on key fields by using a dataframe. Merge method, then records containing missing values are removed by using a dropna method, and duplicate records are deleted by using a drop_ duplicates method. The result of this process is a clean and complete integrated data set that provides a reliable data basis for subsequent analytical evaluation and decision support.
The analysis and evaluation sub-module performs quality evaluation on different decision schemes by using a hierarchical analysis process based on the integrated data set. First, a hierarchical model is constructed to divide the decision problem into a target layer, a criterion layer and a scheme layer. The compare method of ahpy library is then used to quantify the relative importance between each tier and to set a consistency index to verify the consistency of the decision matrix. And identifying an optimal decision scheme according to the weight value obtained by calculation through the series of calculation, thereby generating a decision scheme evaluation result. The process not only improves the scientificity and accuracy of the decision, but also provides a solid analysis basis for the generation of the comprehensive scheme.
The comprehensive scheme generating sub-module adopts a multi-objective decision analysis method to comprehensively consider the current production demand and resource configuration on the basis of the analysis and evaluation sub-module, and selects the most suitable production adjustment measures through logic judgment and weight comparison. In addition, pareto front analysis is also used to analyze and select the best adjustment measures on the premise of meeting a plurality of production targets. The final output of the process is a comprehensive adjustment scheme, which not only meets the improvement requirements of production efficiency and quality, but also considers the sustainable utilization of resources and environmental influence, and provides comprehensive decision support for intelligent optimization of the steel production process.
Referring to fig. 2 and 10, the dynamic response and adjustment module includes a flow adjustment sub-module, an operation optimization sub-module, and an environment matching sub-module;
The flow adjustment submodule adjusts the production flow in real time by adopting a fuzzy logic control method based on a comprehensive adjustment scheme, a fuzzy logic controller is constructed, the controller comprises a fuzzy rule set, the production speed and the raw material input quantity are used as input variables, output variables are defined as adjusted production parameters, a fuzzy module in Matlab or Python is used for setting membership functions and rule bases of the input variables, input data are processed through fuzzy reasoning, an operation parameter value is selected by applying a centroid method, and the change of real-time production requirements is matched, so that a flow adjustment strategy is generated;
The operation optimization submodule executes rolling time domain optimization based on a flow adjustment strategy, continuously optimizes operation parameters, sets a rolling time domain frame, updates an objective function and constraint by using current and predicted production data, applies a genetic algorithm, executes the operation optimization by using a deap library of Python, and adjusts algorithm parameters including population size, cross rate and mutation rate, wherein the objective function design simultaneously maximizes production efficiency and minimizes cost, constraint conditions include machine capacity and quality requirements, and adjusts the operation parameters according to instant data after each rolling window is finished to generate an operation optimization scheme;
the environment matching sub-module is used for implementing environment response measures based on an operation optimization scheme, coping with sudden events in a production environment, adopting a Python change detection technology to identify key change points in the production process, dynamically adjusting a production plan and resource allocation according to the detected event type and urgency, adopting a linear programming technology, optimizing through linprog functions of a SciPy library in the resource allocation, matching and recovering to an optimized production state, and generating an adaptive adjustment scheme of the production process.
In the flow adjustment sub-module, the production flow is adjusted in real time by a fuzzy logic control method, so that the dynamic management of the production speed and the raw material input amount is realized, and the change of real-time production requirements is dealt with. First, input variables (production speed and raw material input amount) and output variables (adjusted production parameters) are defined, and a fuzzy logic controller is constructed by using a fuzzy module in a Matlab or Python environment. The construction of the controller includes setting membership functions for the input variables and a rule base, which are formulated based on previous production experience and expert knowledge. And processing input data by a fuzzy inference system, and calculating by adopting a centroid method to obtain a final operation parameter value. The result of the process is that a set of flow adjustment strategy is generated, the change in the production process can be responded flexibly, and the production parameter setting is optimized, so that the production efficiency and the adaptability are improved.
The operation optimization submodule performs rolling time domain optimization to continuously improve the operation parameters based on the flow adjustment strategy. The module adopts a genetic algorithm, is realized through a deap library of Python, and optimizes the operation parameters of the production process, including the adjustment of population size, crossing rate and mutation rate. The objective function aims to maximize production efficiency and minimize cost while meeting constraints on machine capacity and quality requirements. By setting a rolling time domain frame, the module dynamically updates an objective function and constraint conditions by using current and predicted production data, and adjusts operation parameters according to instant data after each rolling window is finished. The process not only realizes continuous optimization of production parameters, but also improves flexibility and economic benefit of the production process.
The environment matching sub-module takes measures to cope with sudden events in the production environment based on the operation optimization scheme. The module uses change detection techniques of Python to identify key points of change in the production process, such as raw material quality fluctuations, equipment failures, etc., and then dynamically adjusts the production plan and resource allocation according to the type and urgency of the event. By applying the linear programming technology, the linprog function of the SciPy library is utilized to optimize the resource configuration, so that the production system can be ensured to be quickly adapted to environmental changes and be restored to an optimized state. The implementation of the module improves the toughness and stability of the production system, ensures that the production process can run smoothly even in the face of unpredictable external disturbances.
The present invention is not limited to the above embodiments, and any equivalent embodiments which can be changed or modified by the technical disclosure described above can be applied to other fields, but any simple modification, equivalent changes and modification made to the above embodiments according to the technical matter of the present invention will still fall within the scope of the technical disclosure.
Claims (10)
1. The intelligent digital twin simulation system for the steel production process is characterized in that: the system comprises an integrated modeling and simulation module, a production optimization and regulation module, a flow and morphology analysis module, a cognitive dynamic decision module, a network and stability management module, a microecological regulation and control optimization module, a decision support and strategy optimization module and a dynamic response and adjustment module;
The integrated modeling and simulation module is based on the physical and chemical processes of steel production, adopts a continuous time Markov chain model to describe the state transition probability in the production process, introduces a random differential equation method, simulates random fluctuation in the process, adjusts model parameters according to real-time monitoring data through maximum likelihood estimation, optimizes a model structure by using a genetic algorithm, reflects the current production process and generates a preliminary dynamic model;
The production optimization and regulation module adopts a model prediction control method based on a preliminary dynamic model, designs a prediction model to predict future production trend, optimizes continuous and discrete production control variables by combining a mixed integer linear programming algorithm, and matches the changes of raw material quality and energy cost to generate optimized production parameters;
The flow and morphology analysis module analyzes topological structures of the material flow and the information flow by adopting a differential topology method based on optimized production parameters, identifies key structural features and bottlenecks in the production process by using Tonlon analysis, optimizes the production path by adopting a local linear embedding method, and generates a production path optimization scheme;
The cognitive dynamic decision module is based on a production path optimization scheme, adopts a convolutional neural network, analyzes historical production data and current state, identifies modes and trends, performs strategy optimization by using a deep Q network, and generates a cognitive optimization decision result in a cyclically-changed production environment;
The network and stability management module analyzes the complex network constructed in the production process based on the cognition optimization decision result by adopting a centrality analysis and community detection algorithm, identifies key nodes and fragile links in the network, adjusts the network structure by a modularized optimization technology, and generates a network optimization strategy;
The micro-ecological regulation and control optimization module is based on a network optimization strategy, adopts a DNA sequencing technology and a quantitative polymerase chain reaction method, analyzes the microbial community composition in cooling water, analyzes the interaction between the microbial composition and cooling efficiency by combining a machine learning model, and guides the formulation of a regulation and control strategy to generate a microbial regulation and control scheme;
the decision support and strategy optimization module integrates data by adopting a data fusion technology based on a preliminary dynamic model, optimized production parameters, a production path optimization scheme, a cognitive optimization decision result, a network optimization strategy and a microorganism regulation scheme, evaluates the advantages and disadvantages of various decision schemes by using a hierarchical analysis process, and generates a comprehensive regulation scheme;
the dynamic response and adjustment module adopts fuzzy logic control to adjust the operation parameters of the production line in real time based on the comprehensive adjustment scheme, combines detection and response of emergency events and rolling time domain optimization, continuously optimizes decision and control strategies in the production process, and generates an adaptive adjustment scheme of the production process.
2. The intelligent digital twin simulation system for steel production process according to claim 1, wherein: the preliminary dynamic model comprises probability distribution of production state variables, fluctuation modes of time sequences, response functions of key nodes of a production process, the optimized production parameters comprise adjusted furnace temperature setting, carbon and oxygen input proportion and cooling rate in a steelmaking process, the production path optimization scheme comprises a simplified material handling flow, a recombined production line layout and an optimized information feedback mechanism, the cognitive optimization decision result comprises a raw material purchase plan based on trend prediction, an adaptive adjustment rule of production scheduling and a decision framework of emergency response, the network optimization strategy comprises optimized production core network connection, logistics distribution routes and information flow transparency measures, the microorganism regulation scheme comprises selected antibiotic biological pollution chemicals, a periodic microorganism community monitoring plan, cooling water cleaning and maintenance periodic frequency, the comprehensive adjustment scheme comprises production efficiency optimization measures, energy consumption adjustment strategies and environmental emission control schemes, and the production process adaptation adjustment scheme comprises updated production efficiency indexes, adjustment records responding to external changes and improved product quality.
3. The intelligent digital twin simulation system for steel production process according to claim 1, wherein: the integrated modeling and simulation module comprises a model integration sub-module, a parameter estimation sub-module and a model structure optimization sub-module;
The model integration submodule adopts a continuous time Markov chain model based on the physical and chemical processes of steel production, a state space is used for defining production states, a transition matrix designates the probability from one state to the other state, the probability description of the transition between the states is carried out through the construction of the state transition matrix, a random differential equation is introduced, and the simulation of random fluctuation is carried out through the Euler-Maruyama approximation, so that a state transition and fluctuation model is generated;
The parameter estimation submodule carries out maximum likelihood estimation based on the state conversion and fluctuation model, adjusts parameters by using a gradient descent method, maximizes a likelihood function of observed data, gradually optimizes model parameters by setting an initial parameter estimation value, a learning rate and iteration times, enables the model parameters to be matched with production data, and generates a parameter optimization model;
The model structure optimization submodule is based on a parameter optimization model, adopts a genetic algorithm to optimize a model structure, evaluates model performance by defining a fitness function, adopts selection, crossing and mutation operations, simulates a natural evolution process, selects a model with excellent performance according to the fitness function, generates a new model by randomly selecting a cross point combination of model parameters through the crossing operation, and repeatedly executes the new model structure until the optimal model structure is captured by randomly changing partial values in the model parameters and introducing new genetic diversity to generate a preliminary dynamic model.
4. The intelligent digital twin simulation system for steel production process according to claim 1, wherein: the production optimization and regulation module comprises a control strategy sub-module, a parameter adjustment sub-module and an external factor response sub-module;
The control strategy submodule carries out a model prediction control method based on a preliminary dynamic model, builds a control and prediction framework to predict future production trend, adjusts a control strategy according to the model prediction control method, sets a control target by defining state space representation of the prediction model, predicts the production process of each time period by utilizing a rolling time window technology, determines optimal setting of production parameters under given constraint conditions by adopting a linear programming solver, dynamically adjusts the operation parameters of the production line, and generates a prediction control scheme according to the predicted production demand and raw material supply condition;
The parameter adjustment submodule is based on a predictive control scheme, adopts a mixed integer linear programming algorithm to refine and optimize production control variables, sets an optimization objective function to minimize production cost by defining an optimization problem comprising continuous and discrete decision variables, meets production and safety constraint conditions, applies Gurobi to solve the problem, and generates a production parameter optimization scheme by adjusting algorithm parameters including a boundary or cutting plane strategy of a branch-and-bound method and optimizing a solution process;
The external factor response submodule analyzes and responds to external factors based on a production parameter optimization scheme, potential influences of the external factors on a production process are evaluated through collecting and analyzing market demand change, raw material price fluctuation and energy supply condition information and using sensitivity analysis, a scene analysis method is adopted to simulate production results under various conditions, and a production strategy is adjusted, wherein the method comprises the steps of changing a raw material use plan and adjusting energy utilization efficiency in the production process, and optimized production parameters are generated.
5. The intelligent digital twin simulation system for steel production process according to claim 1, wherein: the flow and morphology analysis module comprises a topology structure analysis sub-module, a bottleneck identification sub-module and a path optimization sub-module;
The topological structure analysis submodule performs a differential topological method to refine and analyze the structures of logistics and information flows based on optimized production parameters, identifies key topological characteristics in the assembly by constructing the representation of the logistics and the information flows in a multidimensional space and applying coherent group and Betty number calculation, reveals potential connectivity and barrier points in the logistics and the information flows by a ring and a cavity, and comprises the steps of calculating topological invariants by MATLAB to generate a topological structure analysis result;
The bottleneck recognition sub-module is used for analyzing key structural features and bottlenecks in the production process by adopting a Tonlun analysis method based on a topological structure analysis result, recognizing persistent structural features influencing the production process by comparing topological models under various working conditions, selecting a part of the production process which still keeps the structural characteristics unchanged under small variation by utilizing path connectivity and space deformation, constructing and analyzing a mathematical model by using a SciPy library of Python, recognizing key bottlenecks obstructing production efficiency, and generating a bottleneck recognition result;
the path optimization submodule optimizes the production path by using a local linear embedding method based on the bottleneck recognition result, captures the optimal representation capable of keeping the local structure in a low-dimensional space by referring to the local similarity of each link in the production process, and comprises reconstructing the production path by using a local linear embedding algorithm provided by a Scikit-learn library in Python to generate a production path optimization scheme.
6. The intelligent digital twin simulation system for steel production process according to claim 1, wherein: the cognitive dynamic decision module comprises an information processing sub-module, a learning optimization sub-module and a strategy adjustment sub-module;
The information processing submodule is used for executing a convolutional neural network based on a production path optimization scheme, analyzing historical production data and a current state, constructing a convolutional neural network model, extracting spatial features in the data by using a convolutional layer, enabling each layer of convolution to follow an activating layer, optimizing the nonlinear processing capacity of the model by using a ReLU function, reducing the spatial dimension of the features by using a pooling layer, reducing the calculated amount and retaining key information, converting the feature vectors into pattern recognition output by using a full-connection layer, setting an optimizer as Adam in the training process, adopting a loss function by adopting cross-entropy, adjusting the learning rate and the batch size, optimizing the model performance, and generating a pattern recognition result;
The learning optimization submodule applies a depth Q network to optimize a decision strategy based on a mode identification result, evaluates the accumulated benefits of the decision by defining a reward function, optimizes the learning efficiency by using an experience replay mechanism, optimizes the stability of the learning process by adopting a target network, relieves the problem of target movement by periodically updating the weight of the target network, refines and adjusts the learning rate, the updating frequency and the discount factor gamma, matches the change of the production environment and generates an optimized decision strategy;
The strategy adjustment submodule carries out real-time adjustment of strategies in continuous change of the production environment based on optimized decision strategies, compares the optimized strategies with the current production conditions by monitoring real-time production data, identifies strategy parameters needing to be adjusted, adopts a self-adaptive adjustment mechanism, dynamically modifies production decision parameters, including production speed and raw material input proportion, and generates a cognition optimization decision result.
7. The intelligent digital twin simulation system for steel production process according to claim 1, wherein: the network and stability management module comprises a network analysis sub-module, a key node optimization sub-module and a fragile link improvement sub-module;
The network analysis submodule analyzes a complex network constructed in the production process based on a cognition optimization decision result, performs centrality analysis and a community detection algorithm, calculates the direct connection number of each node by adopting centrality, approaches to the average distance from a centrality measurement node to all other nodes, evaluates the importance of the nodes on the path between node pairs by intermediation centrality, identifies communities or groups with dense resources and information flow in the network by a modularity optimization method, comprises setting algorithm parameters by Gephi, revealing key structural characteristics of the production network, and generates a network structure analysis result;
The key node optimizing submodule optimizes the identified key nodes by adopting a modularized optimizing technology based on a network structure analysis result, and comprises the steps of optimizing the connectivity of the key nodes or improving the positions of the key nodes in a network by adopting shortest path optimization or network traffic redistribution, so that the nodes can coordinate resources and information flows in a production network, and a key node optimizing strategy is generated;
The fragile link improvement submodule identifies and improves the fragile links in the network based on the key node optimization strategy, identifies the areas which are easily affected due to weak structure or excessive concentration of resources by analyzing the flow connectivity and the resource distribution of the network, adopts the measures of adding redundant connection, balancing the resource distribution and adjusting the structural design of the fragile links, and generates the network optimization strategy.
8. The intelligent digital twin simulation system for steel production process according to claim 1, wherein: the micro-ecological regulation and control optimization module comprises a community monitoring sub-module, a performance analysis sub-module and a micro-ecological management optimization sub-module;
the community monitoring submodule is used for carrying out high-throughput sequencing on samples through a IlluminaMiSeq platform based on a network optimization strategy, processing sequencing data by using QIIME2 software, clustering operation classification units, setting a 97% similarity threshold, carrying out microorganism abundance analysis by using a feature-tablesummarize command, carrying out microorganism classification by using a feature-CLASSIFIERCLASSIFY-sklearn command, revealing microorganism distribution and diversity in communities, and generating a microorganism community composition analysis result;
the performance analysis submodule builds an analysis result based on a microbial community, adopts a machine learning model, builds a random forest model through a Scikit-learn library of Python, sets an n_ estimators parameter as 100 to represent the number of trees, and a max_depth parameter as 10 to limit the maximum depth of the trees, uses a fit method to carry out model training, adopts a predict method to predict cooling efficiency, adopts a feature_ importances _attribute to evaluate the importance of the types of microorganisms, identifies microorganisms with influence on the cooling efficiency, and generates a cooling efficiency and microorganism interaction analysis result;
The micro-ecological management optimization submodule designs a microorganism regulation and control strategy based on the analysis result of the cooling efficiency and the microorganism interaction, adopts a decision tree algorithm to select regulation and control measures, uses DecisionTreeClassifier of a Scikit-learn library, sets criterion parameters as gini to evaluate splitting quality, uses a splitter parameter as best to capture optimal splitting, trains regulation and control strategy data by a fit method, and uses a predict method to adjust the strategy according to real-time monitoring data to generate a microorganism regulation and control scheme.
9. The intelligent digital twin simulation system for steel production process according to claim 1, wherein: the decision support and strategy optimization module comprises a data integration sub-module, an analysis and evaluation sub-module and a comprehensive scheme generation sub-module;
The data integration submodule carries out comprehensive processing on data based on a preliminary dynamic model, optimized production parameters, a production path optimization scheme, a cognitive optimization decision result, a network optimization strategy and a microorganism regulation scheme, merges the data of multiple sources through a Pandas library, associates key fields of a data set by using a dataframe method, removes missing values by using a dropna method on the merged data set, eliminates repeated records by using a drop duplicates method, and generates an integrated data set;
The analysis and evaluation sub-module adopts a hierarchical analysis process to evaluate the quality of the decision scheme based on the integrated data set, a hierarchical structure model is constructed to be divided into a target layer, a criterion layer and a scheme layer, a compact method of a ahpy library is used for quantifying the relative importance among each layer, consistency of a consistency index check decision matrix is set, an optimal decision scheme is identified through a weight value obtained through calculation, and a decision scheme evaluation result is generated;
The comprehensive scheme generating submodule adopts a multi-objective decision analysis method based on the decision scheme evaluation result, selects production adjustment measures through logic judgment and weight comparison by referring to the current production demand and resource configuration, analyzes and selects adjustment measures capable of meeting a plurality of production targets simultaneously by utilizing pareto front analysis, and generates a comprehensive adjustment scheme.
10. The intelligent digital twin simulation system for steel production process according to claim 1, wherein: the dynamic response and adjustment module comprises a flow adjustment sub-module, an operation optimization sub-module and an environment matching sub-module;
The flow adjustment submodule is based on a comprehensive adjustment scheme, adopts a fuzzy logic control method to adjust the production flow in real time, constructs a fuzzy logic controller, takes the production speed and the raw material input quantity as input variables, defines output variables as adjusted production parameters, sets membership functions and rule bases of the input variables by utilizing a fuzzy module in Matlab or Python, processes the input data through fuzzy reasoning, selects operation parameter values by applying a centroid method, matches the change of real-time production requirements, and generates a flow adjustment strategy;
The operation optimization submodule executes rolling time domain optimization based on a flow adjustment strategy, continuously optimizes operation parameters, sets a rolling time domain frame, updates an objective function and constraint by using current and predicted production data, applies a genetic algorithm, executes the operation optimization by using a deap library of Python, and adjusts algorithm parameters including population size, cross rate and mutation rate, wherein the objective function design simultaneously maximizes production efficiency and minimizes cost, constraint conditions include machine capacity and quality requirements, and adjusts the operation parameters according to instant data after each rolling window is finished to generate an operation optimization scheme;
The environment matching sub-module is used for implementing environment response measures based on an operation optimization scheme, coping with sudden events in a production environment, adopting a Python change detection technology to identify key change points in the production process, dynamically adjusting a production plan and resource allocation according to the detected event type and urgency, adopting a linear programming technology, optimizing through linprog functions of a SciPy library in the resource allocation, matching and recovering to an optimized production state, and generating an adaptive adjustment scheme of the production process.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410324537.7A CN117930786B (en) | 2024-03-21 | 2024-03-21 | Intelligent digital twin simulation system for steel production process |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410324537.7A CN117930786B (en) | 2024-03-21 | 2024-03-21 | Intelligent digital twin simulation system for steel production process |
Publications (2)
Publication Number | Publication Date |
---|---|
CN117930786A true CN117930786A (en) | 2024-04-26 |
CN117930786B CN117930786B (en) | 2024-06-11 |
Family
ID=90755916
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202410324537.7A Active CN117930786B (en) | 2024-03-21 | 2024-03-21 | Intelligent digital twin simulation system for steel production process |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117930786B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN118396169A (en) * | 2024-05-07 | 2024-07-26 | 深圳市益普科技有限公司 | Data processing method, system, device and storage medium of intelligent manufacturing factory |
CN118430724A (en) * | 2024-07-03 | 2024-08-02 | 南通蝶绣纺织有限公司 | Efficient material retrieval method and system for textile production |
CN118751435A (en) * | 2024-09-05 | 2024-10-11 | 南通优好佳包装材料有限公司 | PET bottle automatic spraying coating control system |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2021227325A1 (en) * | 2020-05-13 | 2021-11-18 | 中国电子科技集团公司第十四研究所 | Digital twin-based production process simulation and optimization method |
CN114115147A (en) * | 2021-11-18 | 2022-03-01 | 中国电子科技集团公司第三十八研究所 | SMT manufacturing process intelligent management and control system based on digital twins |
CN116118163A (en) * | 2021-11-15 | 2023-05-16 | 颖华科技股份有限公司 | Curved plastic panel and processing method and processing device thereof |
CN116931530A (en) * | 2023-09-15 | 2023-10-24 | 山东星科智能科技股份有限公司 | Dynamic prediction system and dynamic prediction method for iron-making steel rolling production process |
CN117393076A (en) * | 2023-12-13 | 2024-01-12 | 山东三岳化工有限公司 | Intelligent monitoring method and system for heat-resistant epoxy resin production process |
CN117495210A (en) * | 2024-01-02 | 2024-02-02 | 衡水烨通建设工程有限公司 | Highway concrete construction quality management system |
CN117575108A (en) * | 2024-01-16 | 2024-02-20 | 山东三岳化工有限公司 | Chemical plant energy data analysis system |
CN117611015A (en) * | 2024-01-22 | 2024-02-27 | 衡水烨通建设工程有限公司 | Real-time monitoring system for quality of building engineering |
CN117687891A (en) * | 2024-02-02 | 2024-03-12 | 深圳迅策科技股份有限公司 | Index calculation optimization system based on AI |
-
2024
- 2024-03-21 CN CN202410324537.7A patent/CN117930786B/en active Active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2021227325A1 (en) * | 2020-05-13 | 2021-11-18 | 中国电子科技集团公司第十四研究所 | Digital twin-based production process simulation and optimization method |
CN116118163A (en) * | 2021-11-15 | 2023-05-16 | 颖华科技股份有限公司 | Curved plastic panel and processing method and processing device thereof |
CN114115147A (en) * | 2021-11-18 | 2022-03-01 | 中国电子科技集团公司第三十八研究所 | SMT manufacturing process intelligent management and control system based on digital twins |
CN116931530A (en) * | 2023-09-15 | 2023-10-24 | 山东星科智能科技股份有限公司 | Dynamic prediction system and dynamic prediction method for iron-making steel rolling production process |
CN117393076A (en) * | 2023-12-13 | 2024-01-12 | 山东三岳化工有限公司 | Intelligent monitoring method and system for heat-resistant epoxy resin production process |
CN117495210A (en) * | 2024-01-02 | 2024-02-02 | 衡水烨通建设工程有限公司 | Highway concrete construction quality management system |
CN117575108A (en) * | 2024-01-16 | 2024-02-20 | 山东三岳化工有限公司 | Chemical plant energy data analysis system |
CN117611015A (en) * | 2024-01-22 | 2024-02-27 | 衡水烨通建设工程有限公司 | Real-time monitoring system for quality of building engineering |
CN117687891A (en) * | 2024-02-02 | 2024-03-12 | 深圳迅策科技股份有限公司 | Index calculation optimization system based on AI |
Non-Patent Citations (1)
Title |
---|
陈志鼎;梅李萍;: "基于数字孪生技术的水轮机虚实交互系统设计", 水电能源科学, no. 09, 15 September 2020 (2020-09-15) * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN118396169A (en) * | 2024-05-07 | 2024-07-26 | 深圳市益普科技有限公司 | Data processing method, system, device and storage medium of intelligent manufacturing factory |
CN118430724A (en) * | 2024-07-03 | 2024-08-02 | 南通蝶绣纺织有限公司 | Efficient material retrieval method and system for textile production |
CN118751435A (en) * | 2024-09-05 | 2024-10-11 | 南通优好佳包装材料有限公司 | PET bottle automatic spraying coating control system |
Also Published As
Publication number | Publication date |
---|---|
CN117930786B (en) | 2024-06-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN117930786B (en) | Intelligent digital twin simulation system for steel production process | |
US11755976B2 (en) | Method for predicting oxygen load in iron and steel enterprises based on production plan | |
Chen et al. | Modelling and optimization of fed-batch fermentation processes using dynamic neural networks and genetic algorithms | |
Plathottam et al. | A review of artificial intelligence applications in manufacturing operations | |
CN112800668A (en) | Digital twin-based complex product operation, maintenance and manufacturing integrated overall model | |
CN117389236B (en) | Propylene oxide production process optimization method and system | |
Matheri et al. | Sustainable circularity and intelligent data-driven operations and control of the wastewater treatment plant | |
CN110675005B (en) | Intelligent decision-making method based on artificial intelligence technology and ROPN technology | |
CN113704956A (en) | Urban road online microscopic simulation method and system based on digital twin technology | |
CN114678080B (en) | Converter end point phosphorus content prediction model, construction method and phosphorus content prediction method | |
CN117725537A (en) | Real-time metering data processing platform | |
CN117808166A (en) | Chemical industry safety automation detection monitoring system of clouding PLC | |
Hu et al. | Weighted kernel fuzzy C-means-based broad learning model for time-series prediction of carbon efficiency in iron ore sintering process | |
CN114418177B (en) | New product material distribution prediction method based on digital twin workshops for generating countermeasure network | |
CN117669895A (en) | Highway engineering environment influence evaluation system | |
Ding et al. | Recent progress on data-based optimization for mineral processing plants | |
Saldivar et al. | Identifying smart design attributes for Industry 4.0 customization using a clustering Genetic Algorithm | |
Alizadeh et al. | Using data mining for learning and clustering FCM | |
CN115660293A (en) | Comprehensive evaluation method for complete life cycle of complex electromechanical product based on digital twinning | |
CN104732067A (en) | Industrial process modeling forecasting method oriented at flow object | |
Tian et al. | Dynamic operation optimization based on improved dynamic multi-objective dragonfly algorithm in continuous annealing process. | |
Wu | Computational intelligence-based intelligent business intelligence system: concept and framework | |
Li et al. | Complementary learning-team machines to enlighten and exploit human expertise | |
CN114740713B (en) | Multi-objective optimization control method for wet flue gas desulfurization process | |
Jasiulewicz-Kaczmarek et al. | Modeling a hierarchical structure of factors influencing exploitation policy for water distribution systems using ISM approach |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |