CN115146455B - Complex supply chain multi-objective decision method supported by calculation experiment - Google Patents
Complex supply chain multi-objective decision method supported by calculation experiment Download PDFInfo
- Publication number
- CN115146455B CN115146455B CN202210751540.8A CN202210751540A CN115146455B CN 115146455 B CN115146455 B CN 115146455B CN 202210751540 A CN202210751540 A CN 202210751540A CN 115146455 B CN115146455 B CN 115146455B
- Authority
- CN
- China
- Prior art keywords
- supply chain
- scheme
- decision
- objective
- model
- 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.)
- Active
Links
- 238000002474 experimental method Methods 0.000 title claims abstract description 100
- 238000000034 method Methods 0.000 title claims abstract description 50
- 238000004364 calculation method Methods 0.000 title claims abstract description 48
- 238000005457 optimization Methods 0.000 claims abstract description 60
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 27
- 238000010606 normalization Methods 0.000 claims abstract description 18
- 238000012163 sequencing technique Methods 0.000 claims abstract description 11
- 238000011156 evaluation Methods 0.000 claims description 38
- 230000035945 sensitivity Effects 0.000 claims description 16
- 239000011159 matrix material Substances 0.000 claims description 13
- 230000007246 mechanism Effects 0.000 claims description 13
- 238000011161 development Methods 0.000 claims description 12
- 230000008569 process Effects 0.000 claims description 11
- 238000012795 verification Methods 0.000 claims description 10
- 230000003993 interaction Effects 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 6
- 239000000203 mixture Substances 0.000 claims description 5
- 230000003068 static effect Effects 0.000 claims description 5
- 238000010586 diagram Methods 0.000 claims description 3
- 230000014759 maintenance of location Effects 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 235000013399 edible fruits Nutrition 0.000 claims 1
- 238000013461 design Methods 0.000 abstract description 5
- 230000007547 defect Effects 0.000 abstract description 3
- 230000000694 effects Effects 0.000 abstract description 2
- 238000009826 distribution Methods 0.000 description 14
- 238000004519 manufacturing process Methods 0.000 description 13
- 238000004088 simulation Methods 0.000 description 10
- OKKJLVBELUTLKV-UHFFFAOYSA-N Methanol Chemical compound OC OKKJLVBELUTLKV-UHFFFAOYSA-N 0.000 description 9
- 238000013459 approach Methods 0.000 description 8
- 230000002068 genetic effect Effects 0.000 description 7
- 238000011160 research Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 239000003795 chemical substances by application Substances 0.000 description 4
- 230000009326 social learning Effects 0.000 description 4
- 238000007726 management method Methods 0.000 description 3
- 239000013598 vector Substances 0.000 description 3
- 241000196324 Embryophyta Species 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 230000006399 behavior Effects 0.000 description 2
- 229910052799 carbon Inorganic materials 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 239000006185 dispersion Substances 0.000 description 2
- 238000013508 migration Methods 0.000 description 2
- 230000005012 migration Effects 0.000 description 2
- 230000003612 virological effect Effects 0.000 description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- 206010012186 Delayed delivery Diseases 0.000 description 1
- 235000013400 Quercus lobata Nutrition 0.000 description 1
- 240000001749 Quercus lobata Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000012552 review Methods 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- 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
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Economics (AREA)
- Computational Mathematics (AREA)
- Educational Administration (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- General Engineering & Computer Science (AREA)
- Pure & Applied Mathematics (AREA)
- Development Economics (AREA)
- Data Mining & Analysis (AREA)
- Entrepreneurship & Innovation (AREA)
- Strategic Management (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- Algebra (AREA)
- Computing Systems (AREA)
- Geometry (AREA)
- Evolutionary Computation (AREA)
- Game Theory and Decision Science (AREA)
- Computer Hardware Design (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a multi-objective decision method of a complex supply chain supported by a calculation experiment, which can explore and predict multi-element emergence of a scheme in the complex supply chain, and perform multi-objective optimization and sequencing on the scheme based on the emergence result to obtain decision suggestions. Firstly, designing a calculation experiment to construct a supply chain model; after the design of the calculation experiment model is completed, the calculation experiment and the multi-objective evolutionary algorithm NSGA-II are integrated in a closed loop mode, so that the scheme multi-element evolution and optimization are realized; after scheme multi-objective optimization is completed, sorting and selecting the pareto optimization scheme through an entropy weight-TOPSIS method. The invention provides a multi-target normalization decision method of a supply chain under the theoretical background of a complex system for the first time, and the proposed decision method can make up for the decision defect of a manager based on subjective preference, increase the decision alternative space of the manager and improve the decision effect.
Description
Technical Field
The invention belongs to the field of computer simulation optimization, and particularly relates to a complex supply chain multi-objective decision method supported by a calculation experiment.
Background
Based on computational experiments, the multiple emerging of schemes can be explored, enabling predictive decisions, with applications including but not limited to: the supply chain network coordination mechanisms (Q.Q.Long.2014. An agent-based distributed computational experiment framework for virtual supply chain network development.Expert Systems with Applications,41 (9), 4094-4112; Q.Q.Long.2015.three-dimensional-flow model of agent-based computational experiment for complex supply network solutions.Experion Systems with Applications,42 (5), 2525-2537), customer service supply chains (X.Xue, S.F.Wang, L.J.Zhang, Z.Y.Feng, Y.D.Guo.2019.Social learning evolution (SLE): computational experiment-based modeling framework of social management.IEEE Transactions on Industrial Informatics,15 (6), 3343-3355;X.Xue,S.F.Wang,L.J.Zhang,Z.Y.Feng.2019.Evaluating of dynamic service matching strategy for social manufacturing in cloud environment.Future Generation Computer Systems,91,311-326;X.Xue,H.F.Han,S.F.Wang,C.Z.Qin.2019.Computational experiment-based evaluation on context-Aware O2O service recovery.IEEE Transactions on Services Computing,12 (6), 910-924;X.Xue,Y.M.Kou,S.F.Wang,Z.Z.Liu.2018.Computational experiment research on the equalization-oriented service strategy in collaborative management sciences. Transactions on Services Computing,11 (2), 369-383;X.Xue,S.F.Wang,B.Gui,Z.W.Hou.2016.A computational experiment-based evaluation method for context-Aware services in complicated management Sciences, IEEE 269-286), social business knowledge sharing behaviors (G.Y.Jiang, F.C.Ma, J.Shang, P.Y.K.Chau.2014.Evolution of knowledge sharing behavior in social commerce: an agent-based computational, 278-250).
Currently, few documents propose multi-objective decision methods supported by computational experiments, but some studies have utilized methods combining simulation and evolutionary algorithms to solve the complex supply chain decision problem. Noordhoek et al (M.Noordhoek, W.Dullaert, D.S.W.Lai, S.de Leeuw.2018.A formulation-optimization approach for a service-constrained multi-echelon distribution network. Delivery Research Part E: logistics and Transportation Review,114, 292-311.) developed a simulated optimization method in combination with a scatter search algorithm to optimize a multi-level distribution network with defined delivery times, delayed delivery, and order satisfaction rate constraints. Frazzon et al (E.M.Frazzon, A.Albrecht, M.Pires, E.Israel, M.K uck, m.freitag.2018.hybrid approach for the integrated scheduling of production and transport processes along supply chain. International Journal of Production Research,56 (5), 2019-2035.) propose a supply chain scheduling method integrating simulation and genetic algorithms in an attempt to reduce the number of overdue orders with a controllable cost. Taghdisian et al (H.Taghdisian, M.R.Pishvaie, F.Farhadi.2015.Multi-objective optimization approach for green design of methanol plant based on CO 2-efficiency indicator. Journal of Cleaner Production,103 (15), 640-650.) optimized methanol production using genetic algorithms to maximize yield and minimize carbon emissions. The optimization problem in these studies is multi-objective in nature, but often is converted to a single-objective problem by means of weighting preferences (H.Taghdisian, M.R.Pishvaie, F.Farhadi.2015.Multi-objective optimization approach for green design of methanol plant based on CO 2-efficiency indicator. Journal of Cleaner Production,103 (15), 640-650.) or converting part of the objective to constraints (H.Ch vez, K.casting-Villar, L.Herrera, A.Bustos.2017.Simulation-based multi-objective model for supply chains with disruptions in transport. Robotics and Computer-Integrated Manufacturing,43,39-49;A.Pan,S.Y.S.Leung,K.L.Moon,K.W.Yeung.2009.Optimal reorder decision-making in the agent-based apparel supply chain. Exert Systems with Applications,36 (4), 8571-8581;J.F.Robles,M.Chica,O.Cordon.2020.Evolutionary multiobjective optimization to target social network influentials in viral marketing.Expert Systems with Applications,147,113183.). The resulting solution in this way is highly sensitive to weight vectors, or performs well on some targets but not others (N.Srinivas, K.Deb.1994.Multi-objective optimization using nondominated sorting in genetic algorithms. Evolutionary Computation,2 (3), 221-248.). These solutions are known as pareto optimal solutions or non-dominant solutions (N.Srinivas, K.Deb.1994.Multi-objective optimization using nondominated sorting in genetic algorithms.evolutionary Computation,2 (3), 221-248; V.Chankong, Y.Y.Haimes.1983.Multi-objective decision making theory and methodology (North-Holland, new York); A.E.hans.1988.multicriteria optimization for highly accurate systems.multicriteria Optimization in Engineering and Sciences,19, 309-352.). NSGA-II (K.Deb, A.Pratap, S.Agarwal, T.Meyarivan.2002.A fast and elitist multiobjective genetic algorithm: NSGA-II. IEEE Transactions on Evolutionary Computation,6 (2), 182-197.) is one of the most popular pareto optimal algorithms at present because of its good diversity of distribution of solutions and better convergence near the true pareto front. Robles et al (J.F.Robles, M.Chica, O.Cordon.2020.Evolutionary multiobjective optimization to target social network influentials in viral marking.expert Systems with Applications,147,113183.) have used a simulation optimization method in combination with NSGA-II to solve the problem of maximizing the impact of social networks, found that multi-objective algorithms perform better than single-objective algorithms, and that NSGA-II can get more solutions than decomposition-based multi-objective evolutionary algorithms (MOEA/D) in the number of pareto front schemes. Avci et al (M.G.Avci, H.Selim.2017.A Multi-objective, formulation-based optimization framework for supply chains with premium freight. Expert Systems with Applications,67,95-106.) and Avci et al (M.G.Avci, H.Selim.2018.A Multi-objective simulation-based optimization approach for inventory replenishment problem with premium freights in convergent supply chain. Omega,80, 153-165.) discuss optimal inventory strategies under different inventory principles, finding NSGA-II to be superior to MOEA/D in finding optimal solutions. Combining simulation with the multi-objective evolutionary algorithm NSGA-II is an effective way to deal with the multi-objective normative decision problem.
Since the calculation experiments are a further development of simulation, the multi-objective decision method supported by the calculation experiments combined with the multi-objective evolutionary algorithm NSGA-II should perform well. While few papers have proposed computational experiment-supported multi-objective decision methods for use in the field of complex supply chain evolution, some researchers have used single-objective evolutionary algorithms in computational experiments to predict and analyze the outcome of a solution, such as optimizing the supply chain path in the experiment (X.Xue, Y.M.Kou, S.F.Wang, Z.Z.Liu.2018.Computational experiment research on the equalization-oriented service strategy in collaborative modeling.IEEE Transactions on Services Computing,11 (2), 369-383.) or using single-objective evolutionary algorithms in computational experiments to characterize the evolution process of supply chain individuals (X.Xue, S.F.Wang, L.J.Zhang, Z.Y.Feng, Y.D.Guo.2019.Social learning evolution (SLE): computational experiment-based modeling framework of social modeling.IEEE Transactions on Industrial Informatics,15 (6), 3343-3355.). Xue et al (X.Xue, Y.M.Kou, S.F.Wang, Z.Z.Liu.2018.Computational experiment research on the equalization-oriented service strategy in collaborative manufacturing.IEEE Transactions on Services Computing,11 (2), 369-383) use computational experiments to evaluate different service policies and use genetic algorithms in experiments to deal with manufacturing service composition problems. The social manufacturing framework is modeled from the perspective of social learning evolution using computational experiments, and the evolution of individual enterprises is described using genetic algorithms (X.Xue, S.F.Wang, L.J.Zhang, Z.Y.Feng, Y.D.Guo.2019.Social learning evolution (SLE): computational experiment-based modeling framework of social manufacturing. IEEE Transactions on Industrial Informatics,15 (6), 3343-3355.). Both papers demonstrate the effectiveness of the combination of computational experiments with evolutionary algorithms. However, there is still room for further exploration:
(1) Since both articles focus on multi-objective predictive or analytical decisions in a complex supply chain, the problem of how to extend predictive or analytical decisions to multi-objective normative decisions is not solved;
(2) Because the single-objective optimization algorithm cannot realize multi-objective optimization of the scheme, the supply chain performance objective and the additional stability objective cannot realize simultaneous optimization, which may lead to poor performance of the actual optimization scheme, it is necessary to optimize the scheme using the multi-objective evolutionary algorithm NSGA-II.
In the multi-objective normative decision method, after the manager adopts a multi-objective evolutionary algorithm NSGA-II to realize multi-objective optimization, the manager adopts a multi-objective decision analysis method to sort and screen the pareto optimal scheme. As an extension of the classical TOPSIS approach in terms of objective preferences, TOPSIS-to-entropy fusion (i.e., entropy weight-TOPSIS approach) has been successfully used to solve the problems of green vendor selection (b.m. dos Santos, L.P.Godoy, L.M.S.Campos.2019.Performance evaluation of green suppliers using Entropy-TOPSIS-f.journal of Cleaner Production,207, 498-509.) and public evaluation blockchain (H.M.Tang, Y.Shi, P.W.Dong.2019.Public blockchain evaluation using entropy and topsis.experet Systems with Applications,117, 204-210.). These studies indicate that the entropy weight-TOPSIS approach can rank pareto optimal solutions obtained after optimization of the solution in frame to support multi-objective decisions. However, after the scheme evolution optimization supported by the calculation experiment is finished, fewer articles are used for sequencing the obtained pareto optimal scheme by adopting the entropy weight-TOPSIS method.
From the above-mentioned researches, it can be found that the multi-objective evolutionary algorithm NSGA-II is a popular optimization method at present and is widely applied to the optimization problem of the supply chain, and some articles integrate simulation and the multi-objective algorithm NSGA-II to support multi-objective decision of the supply chain. In these studies, successful application of simulation optimization shows that the multi-objective decision method supported by calculation experiments can help solve complex decision problems in the view of complex system evolution, however, related studies are less.
In addition, in order to make up for the shortcomings of subjective preferences of the manager, the selection of the pareto optimal scheme based on objective preferences is needed to further refine the multi-objective decision-making method. However, few articles combine the entropy weight-TOPSIS method with a multi-objective decision method supported by simulation optimization or computational experiments.
Disclosure of Invention
The invention aims to provide a complex supply chain multi-objective decision method supported by calculation experiments aiming at the defects of the prior art. The method can explore and predict the multi-element emergence of the scheme in the complex supply chain, perform multi-objective optimization on the scheme based on the emergence result, and sort the optimized scheme under objective preference to obtain decision suggestions.
The aim of the invention is realized by the following technical scheme: a complex supply chain multi-objective decision method supported by calculation experiments comprises the following steps:
(1) Designing a computational experiment to build a supply chain model: and designing a calculation experiment model to describe a supply chain, and establishing a mapping relation between a model decision variable and a decision target through calculation experiment development and parameter verification to realize multi-element evaluation of a scheme. The description of the supply chain includes: the supply chain structure and internal entities, the decision mechanism of the supply chain entities, the interaction mode between the supply chain entities, and the decision variables and decision targets of the model.
(2) Supply chain multi-objective optimization: after a calculation experiment model is established based on the step (1), a scheme is generated through a multi-objective evolutionary algorithm NSGA-II, and a scheme is estimated from the view angles of a supply chain performance target and a stability target through deployment calculation experiments, wherein the result of scheme estimation is used as a basis for carrying out elite reservation on the NSGA-II and generating a new scheme to realize further evolutionary optimization. The scheme evaluation based on the calculation experiment comprises three steps of situation establishment, exploration and prediction of the phenomenon of multiple appearances and scheme evaluation. This step is repeated until the termination condition is satisfied. And the scheme set obtained after the termination condition is met is the pareto optimal scheme set.
(3) Pareto optimization scheme ordering: after the multi-objective optimization of the scheme in the step (2) is completed, taking the decision selection difficulty of a manager caused by the non-uniqueness of the scheme in the obtained pareto optimization scheme set into consideration, calculating multi-objective weights through an entropy weight method, and calculating TOPSIS scores of the pareto optimization scheme based on the entropy weights for sorting so as to realize the sorting selection of the pareto optimization scheme under objective preference.
Further, step (1) comprises the following sub-steps:
(1.1) defining a supply chain boundary and a body: as part of the socioeconomic system, the real supply chain often exists in the form of a "supply network" which is a complex macro system, so that the real supply chain needs to be moderately abstract according to the situation, and the boundaries and the internal composition of the supply chain are defined. The intrinsic composition includes the supply chain entities and the entity decision mechanism, as well as the interaction mechanism between the supply chain entities.
(1.2) development of a calculation model: based on the abstract real supply chain SC, the calculation model development is to build a supply chain model ASC through an abstract mathematical language, and design a program module to realize the construction of the supply chain model. Supply chain modelAs an approximation of the real supply chain SC (i.e. asc≡sc), it is mainly composed ofStatic directed graph reflecting supply chain structure>And a function DF reflecting the evolution trend of the supply chain. Structural directed graph in supply chain +.>V represents a node in the supply chain, i.e. a supply chain body, such as an enterprise or an individual, which often has its individual attribute properties C, its decision mechanism D and constraints R, which are different in terms of identity characteristics, decision mechanism and resource constraints, and which need to be described by parameter settings, algorithm models, etc. E represents the interaction and association between the supply chain entities. Based on this directed graph->A static supply chain model may be built. The evolution trend DF of the supply chain model can be expressed as a directed graph +.>I.e. the evolution trend of the node-to-node relationship in the directed graph. After the quantitative description of the supply chain model is implemented by mathematical language, a computational experiment system needs to be developed to implement the supply chain system evolution in a computer program. The system mainly comprises an experiment parameter module, an experiment execution module and a result evaluation module.
(1.3) computing model implementation: after the development of the computational experiment model is completed, the computational experiment model needs to be realized by a computer program. The experimental parameter module mainly comprises the settings of system environment variable World and decision variable Solution (one group of decision variables is recorded as Solution) i I=1, 2, …, I. I is the total number of schemes), the setting of parameters often comes from realistic scenarios, expert knowledge, and proven axiom. Inputting parameters (including system environment variables and decision variables) and completing initialization of the system and establishment of the situation when each experiment is executed, deploying calculation experiments and performing the experiments to obtain the solutionn i The result of the next macroscopic appearance (noted as the jth time, j=1, 2, …, J.J as the number of protocol repetitions)Can use vector +.>Describing macroscopic appearance phenomenon from M index dimensions (namely M decision targets), and determining variable Solution in environment variable World i The value of the mth decision goal of the lower supply chain in a single trial is abbreviated as +.>Considering the diversity of macroscopic appearance, experiments need to be repeated a plurality of times to realize the appearance as full as possible, and the Solution in the proposal is obtained i Experimental results below->Based on the obtained macroscopic appearance, performance values for M performance indicators of the supply chain can be calculated, typically expressed as the mean of the performance values in the multiple macroscopic appearance, +.>
(1.4) parameter sensitivity verification: considering the influence of partial parameter (including system environment variable World and decision variable Solution) setting on the experimental result in the calculation experimental model, for the parameter setting of the system environment variable, the value of the experimental parameter (non-decision variable) and the value range of the decision variable should be further adjusted and confirmed through actual data, expert knowledge and the like. And performing sensitivity verification on parameters in the model after confirmation.
(1.5) evaluation of experimental results: after verifying the rationality of the calculated model parameters, the model parameters and the experimental results thereof are evaluated mainly from two aspects of M performance indexes and 1 stability target of the supply chain. Each performance indicator KPI m The value of (2) is the level of the corresponding target in repeated experimentsAnd (5) an average value. The stability objective reflects the fluctuations of all performance indicators in the repeated experiments, thus using the standard deviation function
Representation scheme Solution i Stability of the results. The smaller the value of the standard deviation, the higher the stability of the protocol.
Further, step (2) comprises the following sub-steps:
(2.1) initializing: initializing decision variable information and setting the evolution algebra to be 1.
(2.2) generating a new generation population: within the variable value range, a population is randomly generated, and the evolution algebra is increased by one generation.
(2.3) selecting one member of the population: and sequentially selecting one member in the population, and carrying out the next scheme evaluation.
(2.4) protocol evaluation: for the member selected in step (2.3), i.e. a set of decision variables, it is input into a computational experiment model, a context is established, and a computational experiment is deployed and implemented. And (3) repeating the experiment for multiple times to explore and predict the multiple occurrences of the scheme in the supply chain, calculating an experiment result after reaching the requirement of the repeated times (the multiple occurrences are explored, obtaining a target value of the performance and a target value of the stability of the supply chain, completing multi-target evaluation, and taking the target value as the basis for sequencing the scheme in the step (2.5). This step is repeated until all members (protocols) in the population are evaluated.
(2.5) scheme ordering: after all members of the population, i.e. all protocols, have been evaluated, non-dominant ranking is performed according to the evaluation results of each protocol in step (2.4). And if the evolution algebra is 1 at the moment, sequencing the whole population, and if the evolution algebra is more than 1 at the moment, merging the parent population and the newly generated child population, and sequencing. In the sorting process, non-dominant sorting is performed and the degree of congestion of the scheme at each layer is calculated.
(2.6) elite retention: according to the order of the non-dominant layers, adding the whole non-dominant layers into the next generation parent population in turn until the situation that all non-dominant layers cannot be added into the next generation population occurs (if the dominant layers are added into the next generation parent population, the overflow of the quantity of the next generation parent population can be caused). For schemes at this non-dominant layer, they are ranked based on the degree of congestion, the greater the degree of congestion, the higher the likelihood of entering the next generation parent population.
And (2.7) if the exit condition is met, the scheme optimization is finished to obtain a pareto optimal scheme set, otherwise, the selection of the mosaic system, the crossing and the variation of the population are carried out to generate a new generation of population, and the steps (2.2) - (2.6) are repeated.
Further, step (3) comprises the following sub-steps:
(3.1) target value normalization: target values of the schemes under multiple targets can be obtained for the pareto optimal scheme set obtained after the multi-element evolution and optimization, and the target values of N groups of schemes on M+1 targets form a score matrix. First, the score matrix is normalized in each target dimension, namely: dividing the score of a solution in a certain target dimension by the sum of all solution scores in that target dimension
(3.2) calculating information entropy: by the formula:and calculating information entropy under different target dimensions. In order to make the entropy of information in all target dimensions within the interval [0,1 ], let k here be k=1/lnN.
(3.3) entropy value reverse processing: the distribution conditions of the target values are different in different target dimensions, the target value distribution is concentrated in some target dimensions, and the target value distribution is dispersed in some target dimensions. The target dimension in which the target value distribution is more dispersed should be more focused, and therefore, the entropy value obtained in step (3.2) is subjected to the inverse process: e (E) m′ ′＝1-E m′ 。
(3.4) weight normalization: based on E determined in step (3.3) m′ ' can be used to describe the forward relationship between the degree of dispersion of the target value distribution and the degree of attention, considering that in the subsequent step, the entropy of information will be the weight of the multiple targets, the E obtained is thus referred to herein m′ ' further normalization:
(3.5) score matrix normalization: taking into account the difference in the dimensions of different target dimensions, normalizing the score matrix in each target dimension by a normalization formula:
(3.6) determining ideal and non-ideal conditions: after normalization operation is completed, the maximum value under different target dimensions is calculatedMinimum value z' m′ (M' =1, 2, …, m+1), determining ideal and irrational conditions in each target dimension:
(3.7) calculating the distances of the scheme to ideal and non-ideal cases: after determining ideal and non-ideal conditions in each target dimension, the Euclidean distance between the scheme and the two is calculated:
(3.8) calculate TOPSIS score: calculating a TOPSIS score based on the distances of the solution found in step (3.7) to the corresponding ideal and non-ideal cases at different target dimensions Wherein w is m′ And the weight corresponding to different target dimensions. After obtaining the TOPSIS score, the schemes are ranked based on the score, and the higher the score, the better the scheme performance.
The beneficial effects of the invention are as follows: the invention can support macroscopic appearance and multiple evolution of a complex supply chain, predict possible results of a scheme in reality, and give decision suggestions according to the emerging results, namely, the decisions are expanded from predictability to standardability. The invention can make up the decision defect of the manager based on subjective preference, increase the decision alternative space of the manager and improve the decision effect.
Drawings
FIG. 1 is a block diagram of the method of the present invention;
fig. 2 is a specific flow chart of the method of the present invention.
Detailed Description
The complex supply chain multi-objective decision method supported by the calculation experiment provides a comprehensive, effective and reliable decision analysis tool for a manager.
As shown in fig. 1 and 2, an embodiment of a complex supply chain multi-objective decision method for multi-generation smart phone marketing of the present invention includes the following steps:
(1) Designing a computational experiment to build a supply chain model: and designing a calculation experiment model to describe a supply chain, and establishing a mapping relation between a model decision variable and a decision target through calculation experiment development and verification to realize multi-element evaluation of a scheme. The description of the supply chain includes: the supply chain structure and internal entities, the decision mechanism of the supply chain entities, the interaction mode between the supply chain entities, and the decision variables and decision targets of the model.
(1.1) defining a supply chain boundary and a body: the present embodiment is a two-level supply chain with one enterprise and a large number of consumers. Businesses offer third generation smart phone products to the marketplace that maximize the revenue, acceptance of innovative technology, consumer utility, and supply chain stability of the business by taking into account the pricing, production, and advertising decisions of these products. All consumers have differences in price sensitivity and technical sensitivity, and in the decision process, the consumers can obtain preliminary consumer utility according to the product price and the technical attribute and the differences in self attribute (price sensitivity and technical sensitivity), and then consider migration phenomenon in the consumer decision process to obtain a final decision.
(1.2) development of a calculation model: based on the abstract supply chain SC, the calculation model development is to build a supply chain model ASC through an abstract mathematical language, and design a program module to realize the construction of the supply chain model. Supply chain modelAs an approximation of the real supply chain SC, asc≡sc, the diagram is mainly composed of a static directed graph reflecting the supply chain structure +.>And a function DF reflecting the evolution trend of the supply chain.
Structure directed graph in supply chainV denotes a node in the supply chain, i.e. the supply chain body, which in this embodiment consists of enterprise manufacturers and a large number of consumers Consumers; one Consumer is denoted Consumer s S=1, 2, …, S, i.e. v= { manager, consumer 1 ,Consumer 2 ,…,Consumer S And S is the number of consumers.
These nodes V differ in identity, decision mechanism and resource constraints, in this embodiment price sensitive Price Sensitivity s Sensitivity to technology Technology Sensitivity s To describe the individual consumer attribute trait C, i.e., c= { Price Sensitivity s ,Technology Sensitivity s S=1, 2, …, S. The consumer's decision mechanism D comprises three steps: (1) calculating a consumer utility function according to the consumer discrete selection model; (2) calculating the utility of the consumer under the influence of the advertisement; (3) taking randomness in consumer decisions into consideration, a migration matrix is used for representing the change of the consumer decisions, and a final decision is obtained. In the consumer decision process, constraint R is: utility needs to be greater than or equal to 0 and the product is once sold out and no longer available to the consumer. The connection mode E is that a manufacturer provides products to a consumer, and the consumer selects to purchase one product (of three products) or not to purchase the product.
Based on this directed graphA static supply chain may be established. Whereas the evolution trend DF of the supply chain can be represented as a directed graph +.>Is a trend of evolution of (a).
After the quantitative description of the supply chain is implemented by a mathematical language, a computational experiment system needs to be developed to implement the supply chain system evolution in a computer program. The calculation experiment system mainly comprises an experiment parameter module, an experiment execution module and a result evaluation module.
(1.3) computing model implementation: after the development of the calculation experiment model in the step (1.2) is completed, the calculation experiment model needs to be realized through a computer program.
The experiment parameter module mainly comprises the settings of a system environment variable World and a decision variable Solution; a set of decision variables is denoted as Solution i I=1, 2, …, I; i is the total number of schemes, taking i= 19920. In the present embodimentThe system environment variable World includes consumer price sensitive and technology sensitive distributions (mean and variance), vendor's historical sales, and the like. The decision variable is price of three products pi And yield quality pi Pi=1, 2,3, advertisement investment ads for the latest products, i.e.
The experiment execution module comprises: inputting parameters including system environment variables and decision variables at the beginning of each experiment execution; and the initialization of the system and the establishment of the situation are completed, the calculation experiment is deployed and the experiment is carried out, and the Solution in the scheme is obtained i Results of the next (single) macroscopic appearanceJ is the number of protocol repetitions, taking j=50. Can use vector +.>Describing macroscopic appearance phenomena from M index dimensions (decision targets), m=3; in the environment variable World and decision variable Solution i The value of the mth decision goal of the lower supply chain in the jth trial is recorded as +.>
Considering the diversity of macroscopic appearance, experiments need to be repeated a plurality of times to realize the appearance as full as possible, and the Solution in the proposal is obtained i The following experimental resultsBased on the obtained macroscopic appearance, performance values for M performance indicators of the supply chain can be calculated, typically expressed as the mean of the performance values in the multiple macroscopic appearance, +.>
(1.4) parameter sensitivity verification: the influence of partial parameter (including system environment variable World and decision variable Solution) setting on the experimental result in the calculation experimental model is considered.
For parameter setting of system environment variables, the values of experimental parameters (non-decision variables) and the value ranges of decision variables are confirmed through further adjustment of actual data, expert knowledge and the like; and sensitivity verification is carried out on relevant parameters in the model.
(1.5) evaluation of experimental results: the result evaluation module is used for evaluating the model and the experimental result thereof mainly from two aspects of M performance indexes and 1 stability target (total M+1=4) of the supply chain after verifying the rationality of the calculated model parameters.
Each performance indicator KPI m The value of (2) is the average value of the corresponding targets in repeated experiments. The Stability objective reflects all performance indicator fluctuations in repeated experiments, thus using the standard deviation function Stability i Representation scheme Solution i The smaller the value of the standard deviation, the higher the stability of the protocol.
(2) Supply chain multiplex evolution and optimization: after a calculation experiment model is established based on the step (1), a scheme is generated through a multi-objective evolution algorithm NSGA-II, a calculation experiment is implemented to evaluate the scheme from the view angle of a supply chain performance target and a stability target, and the result of scheme evaluation is used as NSGA-II to carry out elite reservation, and a new scheme is generated to realize the basis of further evolution optimization.
The scheme evaluation based on the calculation experiment comprises three steps of situation establishment, exploration and prediction of the multi-element emerging phenomenon and scheme evaluation, and the three steps are repeated until the termination condition is met, and the scheme set obtained after the termination condition is met is a pareto optimal scheme set. The method specifically comprises the following steps:
(2.1) initializing: and initializing decision variable information, setting the evolution algebra to be 1, and setting the number of individuals in the population to be 120.
(2.2) generating a new generation population: within the variable value range, a population is randomly generated, and the evolution algebra is increased by one generation.
(2.3) selecting one member of the population: and sequentially selecting one member in the population, and carrying out the next scheme evaluation.
(2.4) protocol evaluation: for the member selected in step (2.3), i.e. a set of decision variables, it is input into a computational experiment model, a context is established, and a computational experiment is deployed and implemented. Repeating the experiment for multiple times to explore and predict the multiple occurrences of the scheme in the supply chain, after the requirement of the repeated times is met, completing the exploration of the multiple occurrences, calculating the experimental result, obtaining the performance target value and the stability target value of the supply chain, completing the multi-objective evaluation, and taking the multi-objective evaluation as the basis of the sequencing of the scheme in the step (2.5). Repeating the step (2.4) until the evaluation of all members (schemes) in the population is completed.
(2.5) scheme ordering: after all the members of the population (i.e. all the protocols) have been evaluated, the non-dominant ranking is performed according to the evaluation results of each protocol in step (2.4). If the evolution algebra is 1 at the moment, sequencing the whole population; if the evolution algebra is larger than 1 at this time, merging the parent population and the newly generated offspring population, and then sequencing. In the sorting process, non-dominant sorting is performed and the degree of congestion of the scheme at each layer is calculated.
(2.6) elite retention: and adding the whole non-dominant layers into the next-generation parent population in turn according to the order of the non-dominant layers until the situation that all non-dominant layers cannot be added into the next-generation parent population appears, wherein the situation is shown that if the dominant layers are added into the next-generation parent population, the overflow of the quantity of the next-generation parent population is caused. For schemes at this non-dominant layer, they are ranked based on the degree of congestion, the greater the degree of congestion, the higher the likelihood of entering the next generation parent population.
(2.7) if the exit condition is met (generating 200 generations of offspring), the scheme optimization is finished, and a pareto optimization scheme set is obtained; and (2) if the exit condition is not met, selecting a mosaic system, crossing and mutating the population to generate a new generation population, and repeating the steps (2.2) - (2.6).
(3) Pareto optimization scheme ordering selection: after the multi-element evolution and optimization of the scheme in the step (2) are completed, the decision selection difficulty of a manager caused by the non-uniqueness of the scheme is considered in the obtained pareto optimization scheme set, the multi-objective weight is calculated through an entropy weight method, the TOPSIS scores of the pareto optimization scheme are calculated based on the entropy weight, and the sorting is carried out, so that the sorting selection of the pareto optimization scheme under objective preference is realized.
(3.1) target value normalization p n,m′ : target values of the schemes under multiple targets can be obtained for the pareto optimal scheme set obtained after the multi-element evolution and optimization, the target values of the N groups of schemes on M+1 targets form a score matrix, and the score matrix is normalized under each target dimension, namely: score u of a solution in a certain target dimension n,m′ Divided by the sum of all solution scores in that target dimension:
wherein M' =1 to m+1.
(3.2) calculating the information entropy E under different target dimensions m′ ：
In order to make the entropy of information in all target dimensions within the interval [0,1 ], let k=1/lnN here.
(3.3) entropy value inverse Process E m′ ': under different target dimensions, the distribution conditions of the target values are different: under some target dimensions, the target value distribution is concentrated; in some target dimensions, the target value distribution is dispersed. The target dimension in which the target value distribution is more dispersed should be more focused, and therefore, the entropy value obtained in step (3.2) is subjected to the inverse process:
E m′ ′＝1-E m′ 。
(3.4) weight normalization w m′ : based on E determined in step (3.3) m′ ' may be used to describe a forward relationship between the degree of dispersion of the target value distribution and the degree of attention; the obtained E is treated here, considering that in the subsequent step the entropy of the information will be the weight of the multi-object m′ ' further normalization:
(3.5) score matrix normalization z n,m′ : taking into account the difference in the dimensions of different target dimensions, normalizing the score matrix in each target dimension by a normalization formula:
(3.6) determining ideal and non-ideal conditions: after normalization operation is completed, the maximum value under different target dimensions is calculatedMinimum value z' m′ (M' =1, 2, …, m+1), determining the ideal condition a in each target dimension * And non-rational condition a':
(3.7) calculating the distance of the plan to the ideal caseAnd solution to non-ideal distance S' n 。
After determining ideal and non-ideal conditions in each target dimension, the Euclidean distance between the scheme and the two is calculated:
(3.8) calculate TOPSIS score: calculating a TOPSIS score C based on the distances of the solution obtained in the step (3.7) to the corresponding ideal and non-ideal conditions under different target dimensions n The method comprises the steps of carrying out a first treatment on the surface of the After the TOPSIS score is obtained, the schemes are ordered based on the score, and the higher the score is, the better the scheme is represented; the decision result is the highest scoring scheme.
Wherein w is m′ And the weight corresponding to different target dimensions.
The present invention is not limited to the above-described embodiments, and all other examples obtained by a person of ordinary skill in the art without making any inventive effort are within the scope of the present invention.
Claims (4)
1. The complex supply chain multi-objective decision method supported by the calculation experiment is characterized by comprising the following steps of:
(1) Designing a computational experiment to build a supply chain model: designing a calculation experiment model to describe a supply chain, and establishing a mapping relation between a model decision variable and a decision target through calculation experiment development and parameter verification to realize multi-element evaluation of a scheme; the description of the supply chain includes: the supply chain structure and internal entities, the decision mechanism of the supply chain entities, the interaction mode between the supply chain entities, and the decision variables and decision targets of the model;
the step (1) comprises the following sub-steps:
(1.1) defining a supply chain boundary and a body: defining boundaries and intrinsic composition of the supply chain; the internal components comprise a supply chain main body and a main body decision mechanism, and an interaction mechanism among the main bodies of the supply chain;
(1.2) development of a calculation model: based on the real supply chain SC, a supply chain model ASC is established, and a program module is designed to realize the establishment of the supply chain model; supply chain modelAs an approximation of the real supply chain SC, asc≡sc, the diagram is mainly composed of a static directed graph reflecting the supply chain structure +.>And a function DF reflecting the evolution trend of the supply chain; structural directed graph in supply chain +.>Wherein V represents a node in the supply chain, node V having its individual attribute trait C, its decision mechanism D and constraint R; e represents interaction modes and links between supply chain entities; based on directed graph->Establishing a static supply chain model; whereas the evolution trend DF of the supply chain model is expressed as directed graph +.>After quantitatively describing the supply chain model, the evolution trend of the connection between the middle nodes needs to develop a calculation experiment system to realize the evolution of the supply chain system in a computer program; node V in the supply chain is the supply chain body, including enterprises, individuals;
(1.3) calculation experiment model implementation: in the calculationAfter the experimental model is developed, the experimental model needs to be realized by a computer program; the experiment parameter module mainly comprises the settings of system environment variable World and decision variable Solution, and one group of decision variables are recorded as Solution i I=1, 2, …, I; i is the total number of schemes; when each experiment is executed, inputting parameters, completing initialization of the system and establishment of a situation, deploying calculation experiments and performing the experiments to obtain a Solution in the scheme i Results of the jth macroscopic appearanceJ is the scheme repetition number; the input parameters comprise system environment variables and decision variables; by vectors->Describing macroscopic appearance phenomenon from M index dimensions, and determining variable Solution in environment variable World i The value of the mth decision goal of the lower supply chain in a single trial is recorded as +.>Repeating the experiment for several times to obtain Solution i Experimental results below->Calculating, based on the obtained macroscopic appearance, performance values for the M performance indicators of the supply chain, expressed as an average of the performance values in the plurality of macroscopic appearances,
(1.4) parameter sensitivity verification: sensitivity verification is carried out on relevant parameters in a calculation experiment model, wherein the sensitivity verification comprises a system environment variable World and a decision variable Solution;
(1.5) evaluation of experimental results: after verifying and calculating the rationality of experimental model parameters, the model and experimental structure are mainly implemented on two aspects of the aspects of M performance indexes and 1 stability target of the supply chainEvaluating fruits; each performance indicator KPI m The value of (2) is the average value of the corresponding targets in repeated experiments; the stability objective reflects the fluctuation of all performance indexes in repeated experiments, so that the Solution is expressed by a standard deviation function i Stability of the results; the smaller the value of the standard deviation, the higher the stability of the protocol;
(2) Supply chain multi-objective optimization: after a calculation experiment model is established based on the step (1), generating a scheme through a multi-objective evolutionary algorithm NSGA-II, deploying a calculation experiment to evaluate the scheme from the view angle of a supply chain performance target and a stability target, and taking the result of scheme evaluation as the basis of NSGA-II to reserve elite and generate a new scheme so as to realize further evolutionary optimization; the scheme evaluation based on the calculation experiment comprises three parts, namely, situation establishment, exploration and prediction of the multi-element emerging phenomenon and scheme evaluation; repeating the step (2) until the termination condition is met; the scheme set obtained after the termination condition is met is the pareto optimization scheme set;
(3) Pareto optimization scheme ordering: after the multi-objective optimization of the scheme in the step (2) is completed, calculating multi-objective weights through an entropy weight method, and sorting TOPSIS scores of the pareto optimization scheme based on the entropy weights so as to realize sorting selection of the pareto optimization scheme.
2. The complex supply chain multi-objective decision method supported by computational experiments of claim 1, wherein step (2) comprises the sub-steps of:
(2.1) initializing: initializing decision variable information, and setting the evolution algebra to be 1;
(2.2) generating a new generation population: randomly generating a population in the variable value range, and increasing the evolution algebra by one generation;
(2.3) selecting one member of the population: sequentially selecting one member in the population, and performing next scheme evaluation;
(2.4) protocol evaluation: inputting the member selected in the step (2.3), namely a group of decision variables, into a calculation experiment model, establishing a situation, deploying and implementing a calculation experiment; repeating the experiment for multiple times to explore and predict the multiple occurrences of the scheme in the supply chain, after the requirement of the repeated times is met, completing the exploration of the multiple occurrences, calculating the experimental result to obtain the performance target value and the stability target value of the supply chain, completing the multi-objective evaluation, and taking the multi-objective evaluation as the basis of the sequencing of the scheme in the step (2.5); repeating the step until the evaluation of all members in the population is completed;
(2.5) scheme ordering: after all the members in the population are evaluated, non-dominant ranking is performed according to the evaluation results of each scheme in the step (2.4); if the evolution algebra is 1 at this time, sequencing the whole population, and if the evolution algebra is more than 1 at this time, merging the parent population and the newly generated child population, and sequencing; in the sorting process, non-dominant sorting is carried out, and the congestion degree of the scheme in each layer is calculated;
(2.6) elite retention: adding the whole non-dominant layer into the next generation parent population in turn according to the non-dominant layer order until the situation that all non-dominant layers cannot be added into the next generation population occurs; for schemes at the non-dominant layer, sorting them based on congestion level, the greater the congestion level, the higher the likelihood of entering the next generation parent population;
(2.7) if the exit condition is met, the scheme optimization is finished, and a pareto optimization scheme set is obtained; otherwise, selecting a mosaic system, crossing and mutating the population, generating a new generation population, and repeating the steps (2.2) - (2.6).
3. The complex supply chain multi-objective decision method supported by computational experiments of claim 2, wherein in step (2.6), the inability to add all non-dominant layers to the next generation population means: adding the dominant layer to the next generation parent population will cause overflow of the next generation parent population number.
4. The complex supply chain multi-objective decision method supported by computational experiments of claim 1, wherein step (3) comprises the sub-steps of:
(3.1) target value normalization: obtaining target values of the pareto optimal scheme sets obtained after the multi-element evolution and optimization under multiple targets, wherein the target values of N groups of schemes on M+1 targets form a score matrix; first, the score matrix is normalized in each target dimension, i.e., the score of a solution in a certain target dimension divided by the sum of all the solution scores in that target dimension
(3.2) calculating information entropy: by passing throughCalculating information entropy under different target dimensions; let k=1/ln N here;
(3.3) entropy value reverse processing: performing inverse processing on the entropy value obtained in the step (3.2): e (E) m′ ′＝1-E m′ ；
(3.4) weight normalization: e obtained in the step (3.3) m′ ' further normalization:
(3.5) score matrix normalization: normalizing the score matrix at each target dimension:
(3.6) determining ideal and non-ideal conditions: after normalization operation is completed, the method calculates the dimensions of different targetsMaximum value of (2)Minimum value z' m′ Determining ideal and irrational conditions in each target dimension:
(3.7) calculating the distances of the scheme to ideal and non-ideal cases: after determining ideal and non-ideal conditions in each target dimension, the Euclidean distance between the scheme and the two is calculated:
(3.8) calculate TOPSIS score: calculating a TOPSIS score based on the distances of the solution found in step (3.7) to the corresponding ideal and non-ideal cases at different target dimensionsWherein w is m′ Weights corresponding to different target dimensions; after obtaining the TOPSIS score, the schemes are ranked based on the score, and the higher the score, the better the scheme performance.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210751540.8A CN115146455B (en) | 2022-06-28 | 2022-06-28 | Complex supply chain multi-objective decision method supported by calculation experiment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210751540.8A CN115146455B (en) | 2022-06-28 | 2022-06-28 | Complex supply chain multi-objective decision method supported by calculation experiment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115146455A CN115146455A (en) | 2022-10-04 |
| CN115146455B true CN115146455B (en) | 2024-04-05 |
Family
ID=83410147
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210751540.8A Active CN115146455B (en) | 2022-06-28 | 2022-06-28 | Complex supply chain multi-objective decision method supported by calculation experiment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115146455B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117132165B (en) * | 2023-09-13 | 2024-07-12 | 深圳达普信科技有限公司 | Dynamic performance evaluation method and system based on data mining |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113991751A (en) * | 2021-10-27 | 2022-01-28 | 云南电网有限责任公司电力科学研究院 | Automatic power generation control scheduling method based on hybrid algorithm |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8015127B2 (en) * | 2006-09-12 | 2011-09-06 | New York University | System, method, and computer-accessible medium for providing a multi-objective evolutionary optimization of agent-based models |
-
2022
- 2022-06-28 CN CN202210751540.8A patent/CN115146455B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113991751A (en) * | 2021-10-27 | 2022-01-28 | 云南电网有限责任公司电力科学研究院 | Automatic power generation control scheduling method based on hybrid algorithm |
Non-Patent Citations (2)
| Title |
|---|
| 分布式电源选址定容的多目标优化算法;夏澍;周明;李庚银;;电网技术;20110905(第09期);全文 * |
| 基于决策偏好的多目标粒子群算法及其应用;王丽萍;江波;邱飞岳;;计算机集成制造系统;20100115(第01期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115146455A (en) | 2022-10-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Xu et al. | A multi-objective decision-making model with fuzzy rough coefficients and its application to the inventory problem | |
| JP2004529406A (en) | Method and apparatus for dynamic real-time market segmentation | |
| WO2003056406A2 (en) | Methods and systems for generating business models | |
| WO2004090692A2 (en) | Methods and systems for interactive evolutionary computing (iec) | |
| Dogan et al. | A reinforcement learning approach to competitive ordering and pricing problem | |
| CN109241399A (en) | A kind of personalized recommendation method combined based on trust and influence power | |
| Rhim et al. | Assessing potential threats to incumbent brands: New product positioning under price competition in a multisegmented market | |
| Navgaran et al. | Evolutionary based matrix factorization method for collaborative filtering systems | |
| CN115146455B (en) | Complex supply chain multi-objective decision method supported by calculation experiment | |
| Werawithayaset et al. | Stock closing price prediction using machine learning | |
| Yue et al. | Product life cycle based demand forecasting by using artificial bee colony algorithm optimized two-stage polynomial fitting | |
| Chien et al. | An integrated approach for IC design R&D portfolio decision and project scheduling and a case study | |
| CN115965203A (en) | User-side virtual power plant flexible adjustment service transaction strategy optimization implementation method | |
| Sampath et al. | A generalized decision support framework for large‐scale project portfolio decisions | |
| CN115689443A (en) | Supply chain management method and system based on volatility clustering and block chain | |
| CN115600009A (en) | Deep reinforcement learning-based recommendation method considering future preference of user | |
| Nilashi et al. | A comparative study of adaptive neuro fuzzy inferences system (ANFIS) and fuzzy inference system (FIS) approach for trust in B2C electronic commerce websites | |
| CN109190820B (en) | Electric power market electricity selling quantity depth prediction method considering user loss rate | |
| Kılıç | Supplier selection application based on a fuzzy multiple criteria decision making methodology | |
| Birtolo et al. | Searching optimal product bundles by means of GA-based Engine and Market Basket Analysis | |
| Kaplarević-Mališić et al. | Identifying optimal architectures of physics-informed neural networks by evolutionary strategy | |
| Junyou | Stock Price forecasting using PSO-trained neural networks | |
| Rohitratana et al. | Agent-based simulations of the software market under different pricing schemes for software-as-a-service and perpetual software | |
| CN116304289A (en) | Information chain recommendation method and device for supply chain based on graphic neural network | |
| US8090835B1 (en) | Optimizing website traffic among content sources |
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 |