CN112862200A - Intelligent feedback optimization method for whole process of highway construction quality - Google Patents
Intelligent feedback optimization method for whole process of highway construction quality Download PDFInfo
- Publication number
- CN112862200A CN112862200A CN202110195411.0A CN202110195411A CN112862200A CN 112862200 A CN112862200 A CN 112862200A CN 202110195411 A CN202110195411 A CN 202110195411A CN 112862200 A CN112862200 A CN 112862200A
- Authority
- CN
- China
- Prior art keywords
- construction
- road
- neural network
- network model
- layer
- 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
- 238000010276 construction Methods 0.000 title claims abstract description 234
- 238000000034 method Methods 0.000 title claims abstract description 107
- 238000005457 optimization Methods 0.000 title claims abstract description 16
- 238000003062 neural network model Methods 0.000 claims abstract description 78
- 239000010410 layer Substances 0.000 claims description 124
- 238000012549 training Methods 0.000 claims description 48
- 210000002569 neuron Anatomy 0.000 claims description 44
- 230000006870 function Effects 0.000 claims description 20
- 238000005096 rolling process Methods 0.000 claims description 20
- 238000012360 testing method Methods 0.000 claims description 17
- 238000013528 artificial neural network Methods 0.000 claims description 16
- 238000004364 calculation method Methods 0.000 claims description 16
- 239000000203 mixture Substances 0.000 claims description 13
- 239000010426 asphalt Substances 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 10
- 230000001133 acceleration Effects 0.000 claims description 7
- 210000004205 output neuron Anatomy 0.000 claims description 6
- 239000002344 surface layer Substances 0.000 claims description 6
- 238000004422 calculation algorithm Methods 0.000 claims description 5
- 230000005284 excitation Effects 0.000 claims description 5
- 238000007781 pre-processing Methods 0.000 claims description 5
- 238000012937 correction Methods 0.000 claims description 3
- 238000005259 measurement Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 2
- 238000012544 monitoring process Methods 0.000 description 33
- 239000000523 sample Substances 0.000 description 21
- 238000005056 compaction Methods 0.000 description 12
- 238000004891 communication Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 238000004088 simulation Methods 0.000 description 5
- 238000013178 mathematical model Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000005303 weighing Methods 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 229910001220 stainless steel Inorganic materials 0.000 description 1
- 239000010935 stainless steel Substances 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Images
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
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- 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
- 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
- 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
- G06Q10/06395—Quality analysis or management
-
- 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
- G06Q50/08—Construction
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- Educational Administration (AREA)
- General Business, Economics & Management (AREA)
- General Health & Medical Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Primary Health Care (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses an intelligent feedback optimization method for the whole process of highway construction quality, which trains a road construction feedback neural network model according to historical road construction parameters and corresponding construction quality indexes; inputting construction parameters of a road which is not constructed in the same area into a road construction feedback neural network model, and obtaining construction quality indexes after the construction of the road is finished through the feedback of the neural network model; and adjusting construction parameters in subsequent unfinished procedures by combining the obtained construction quality indexes, setting multiple groups of adjustable and controllable factors in the subsequent construction process of the road, calculating and feeding back through a neural network model to determine an optimal value, and performing subsequent construction by using the optimal value so as to reduce the influence of uncontrollable factors on the construction quality indexes. The feedback optimization method provides planning reference and decision basis for parameter control in the road construction process, effectively ensures the road construction quality, and reduces the additional cost consumption of road construction.
Description
Technical Field
The invention relates to the technical field of road construction intellectualization, in particular to an intelligent feedback optimization method for the whole process of highway construction quality.
Background
With the progress and development of the technology level, the intelligent construction is widely applied to various fields. In the process of road construction, the intelligent monitoring technology is increasingly used for monitoring and evaluating the road quality, various sensors are mounted to collect data in real time, the collected data are uploaded to a corresponding intelligent monitoring system to be subjected to integrated processing of the data, and meanwhile, a high-precision positioning system is utilized to carry out accurate positioning, so that the wide application of the internet of things technology in the road construction process is promoted, the construction quality of the road is improved, and the construction cost is saved.
However, most of the current researches mainly focus on intelligent monitoring of a single construction process, or a small part of the researches only simply display the data collected from four construction processes in real time although an intelligent monitoring system for the whole construction process exists, and a feedback method for guiding and optimizing the construction process according to the measured data, namely an intelligent feedback optimization method for the whole process of the construction quality of the highway is lacked.
For example, CN 108914751a in the prior art provides a monitoring method for use in a road bed and road surface compaction process, that is, an intelligent compaction monitoring system for use in a road bed and road surface, which includes a model setting module, a control module, a vehicle-mounted module, a communication module and a monitoring center, wherein the model setting module is electrically connected to the control module, the vehicle-mounted module is in communication connection with the control module through the communication module, and the control module is in communication connection with the monitoring center through the communication module. The technology has the advantages that accurate intelligent monitoring of road subgrade and road surface compaction is realized and compaction control precision is improved by comparing multiple data bases with standard compaction parameters. However, the technology only adopts an intelligent monitoring method for a road surface compaction link, does not adopt corresponding monitoring means for a pre-construction process and other construction processes, and does not guide and optimize the construction process according to measured data in the road surface compaction process, so that the method is lack of an intelligent feedback optimization method for the whole process of highway construction quality.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides an intelligent feedback optimization method for the whole process of highway construction quality. The feedback optimization method of the invention trains a road construction feedback neural network model according to historical road construction parameters and corresponding construction quality indexes; the neural network model is used for construction guidance and regulation of a road in construction, so that the road construction quality is effectively guaranteed, and the extra cost consumption of road construction is reduced.
The invention solves the technical problem by designing an intelligent feedback optimization method for the whole process of highway construction quality, which is characterized by comprising the following steps:
the method comprises the following steps: obtaining a raw data set
Acquiring original data of different roads in the same region, wherein the original data comprises construction parameters of the roads and construction quality indexes after the road construction is finished, and the construction parameters comprise mixing temperature and mixing proportion in the asphalt mixing process; the temperature and the transport capacity of the mixture in the transportation process; the temperature and thickness of a paving surface in the paving process; rolling times, surface layer rolling temperature and acceleration of the road roller in the rolling process; the construction quality indexes comprise compactness, deflection value and dry density after the road construction is finished; the construction parameters and the construction quality indexes of one road form an original data set, and the original data of a plurality of roads form an original data set; the construction quality index of each road is obtained by measurement under the same condition;
step two: obtaining a dataset of a network model
Preprocessing the original data sets of different roads in the same region obtained in the step one to obtain a data set D of a network model, dividing the data set D into a training set S and a test set T according to a ratio of 7:3 by a reservation method, wherein D is S and U,
step three: construction of road construction feedback neural network model
Constructing a road construction feedback neural network model based on a BP algorithm, wherein the road construction feedback neural network model is a three-layer perceptron neural network model, an input layer of the road construction feedback neural network model is construction parameters of a road, an output layer of the road construction feedback neural network model is a construction quality index and comprises two hidden layers, wherein the first hidden layer is provided with four hidden nodes, and the second hidden layer is provided with three hidden nodes;
setting an input layer as M, namely, M input signals exist, wherein any one input signal is represented by M; the first hidden layer is I, i.e. there are I neurons, any of which is represented by I; the second hidden layer is J, i.e. there are J neurons, where any neuron is denoted by J; the output layer is P, namely P output neurons, the number of the output neurons is the same as that of the construction quality indexes, and any neuron is represented by P;
the weights of the input layer and the first hidden layer are wmiRepresents; the weight of the first hidden layer and the second hidden layer is wijRepresents; the weight of the second hidden layer and the output layer is wjpRepresents; the input to each neuron is denoted u, the stimulus output is denoted v, the superscripts of u, v denote layers, and the subscripts denote a neuron in a layer, e.g.
An input representing an ith neuron of the first hidden layer; the excitation functions of all the neurons are Sigmoid functions;
taking the construction parameters of the training set S in the step two as the input quantity of an input layer, and taking the construction quality index as the output quantity of an output layer; let X be [ X ] as the input sample set of the training set S1，X2，…，Xk，…，XN]Any corresponding training sample is Xk＝[xk1，xk2，…xkm]K is 1,2, … N, and the actual output is Yk＝[yk1，yk2，…ykP]TThe desired output is dk＝[dk1，dk2，…dkP]T(ii) a N represents the number of roads, m represents the number of construction parameters, k represents the kth road, P represents the number of construction quality indexes, dkCorresponding to the construction quality index of the kth road in the training set S;
setting n as iteration times, wherein the weight and the actual output are both functions of n; input training sample XkThe neural network is obtained by the forward propagation process of the working signal
The error signal of the p-th neuron of the output layer is:
ekp(n)＝dkp(n)-ykp(n)
defining the error energy of the neuron p as
The sum of the errors of all neurons in the output layer is e (n):
transmitting the error sum signal from back to front, and modifying the weight layer by layer in the process of back propagation; when the error sum is smaller than a threshold epsilon, namely | E (n) | < epsilon, the error sum is considered to meet the requirement;
the specific steps of the weight vector training process of the road construction feedback neural network model are as follows:
1) setting variables and parameters
Setting the input quantity of a road construction feedback neural network model as a training sample Xk＝[xk1，xk2，…xkm](k is 1,2, … N), where N is the number of input quantities, that is, the training set S includes the feature values of N roads;
order to
The weight vector between the input layer M and the hidden layer I in the nth iteration is obtained;
order to
The weight vector between the hidden layer I and the hidden layer J in the nth iteration is obtained;
order to
The weight vector between the hidden layer J and the output layer P in the nth iteration is shown;
let Yk(n)＝[yk1(n)，yk2(n)，…ykP(n)]T(k ═ 1,2, … N), which is the actual output value of the neural network at the nth iteration;
dk＝[dk1，dk2，…dkP]T(k ═ 1,2, … N), is the desired output; eta is learning efficiency, and eta is 0.1; n is iteration number, n is 1000, and the threshold value epsilon of the error sum E (n) is 0.0001;
2) initializing weight vector, and assigning weight vector WMI(0)、WIJ(0)、WJP(0) Each element in (1) is a random non-zero value in the range of (-2.4/F ), wherein F is the number of input ends of connected neurons, and n is 0; for weight vector WMI(0) F ═ M; for weight vector WIJ(0) F ═ I; weight vector pair WJP(0)，F＝J；
3) Inputting a random training sample Xk；
4) For input training sample XkForward computing input signal u and output signal v for each layer of neurons in the BP neural network, wherein
P is 1,2, …, P, and the actual output Y is obtainedk(n)；
5) From the desired output dkAnd the actual output Y obtained in the previous stepk(n), calculating the error sum E (n), judging whether the error sum E (n) meets the requirement, and if so, turning to the stepStep 8), if not, turning to the step 6);
6) judging whether n +1 is greater than the maximum iteration times, and if so, turning to the step 8); if not, the input training sample X is processedkCalculating the local gradient delta of each layer of neurons in a reverse mode;
wherein
Is the local gradient between the input layer M and the hidden layer I,
for the local gradient between the hidden layer I and the hidden layer J,
is the local gradient between the hidden layer J and the output layer P; f' () is a derivative function of the Sigmoid function of the excitation function;
7) calculating the weight correction quantity delta w according to the following formula, and correcting the weight:
i＝1，2，…，I；j＝1，2，…，J
p＝1，2，…，P；m＝1，2，…，M
taking the vector formed by the corrected weights as a new weight vector, turning to the step 4), and performing next iteration, wherein the iteration number n is n + 1;
8) the weight vector W in the last iteration according to the previous training sampleMI、WIJ、WJPRepeating the processes from step 2) to step 7) as the initial weight vector of the next training sample which is not repeated randomly, and according to the weight vector W in the last iteration of the last training sampleMI、WIJ、WJPThe weight vector is used as a weight vector of the road construction feedback neural network model, and the trained road construction feedback neural network model is obtained;
step four: verifying the effectiveness of the road construction feedback neural network model trained in the third step
Taking the construction parameters of different roads in the test set T as input quantity of an input layer, and inputting the input quantity into the trained road construction feedback neural network model to obtain an output value of the neural network model; comparing and analyzing the output value of the obtained neural network model with the corresponding construction quality index in the test set T, comparing the output value with the construction quality index in the test set T, calculating a deviation value, and when the deviation value of more than 90% of the construction quality indexes is less than 3%, determining that the road construction feedback neural network model is an effective model;
the calculation method of the deviation value comprises the following steps:
in the formula: pCIs the magnitude of the deviation value, c1For one of the output values obtained via the neural network model, c2Corresponding construction quality indexes in the test set T are obtained;
step five: adjusting construction parameters by applying road construction feedback neural network model and determining optimal value
Inputting construction parameters of a road which is not constructed in the same area into a road construction feedback neural network model, and predicting construction quality indexes after the construction of the road is finished through calculation of the neural network model; and adjusting the construction parameters in unfinished procedures by combining the predicted construction quality indexes, setting multiple groups of adjustable parameters in the subsequent construction process of the road, calculating and feeding back the construction quality indexes corresponding to each group of settings through a neural network model, determining a group of settings corresponding to the construction quality indexes meeting requirements as the optimal values of the adjustable parameters, and performing subsequent construction by using the optimal values.
Compared with the prior art, the invention has the beneficial effects that: the feedback optimization method of the invention trains a road construction feedback neural network model according to historical road construction parameters and corresponding construction quality indexes; inputting construction parameters of a road which is not constructed in the same area into a road construction feedback neural network model, and obtaining construction quality indexes after the construction of the road is finished through the feedback of the neural network model; and adjusting construction parameters in subsequent unfinished procedures by combining the obtained construction quality indexes, setting multiple groups of adjustable and controllable factors in the subsequent construction process of the road, calculating and feeding back through a neural network model to determine an optimal value, and performing subsequent construction by using the optimal value so as to reduce the influence of uncontrollable factors on the construction quality indexes. The feedback optimization method provides planning reference and decision basis for parameter control in the road construction process, effectively ensures the road construction quality, and reduces the additional cost consumption of road construction.
Drawings
Fig. 1 is a schematic structural diagram of a road construction feedback neural network model according to an embodiment of the present invention.
Detailed Description
Specific examples of the present invention are given below. The specific examples are only intended to illustrate the invention in further detail and do not limit the scope of protection of the claims of the present application.
The invention provides an intelligent feedback optimization method for the whole process of highway construction quality, which comprises the following steps:
the method comprises the following steps: obtaining a raw data set
Acquiring original data of different roads in the same region, wherein the original data comprises construction parameters of the roads and construction quality indexes after the road construction is finished, and the construction parameters comprise mixing temperature and mixing proportion in the asphalt mixing process; the temperature and the transport capacity of the mixture in the transportation process; the temperature and thickness of a paving surface in the paving process; rolling times, surface layer rolling temperature and acceleration of the road roller in the rolling process; the construction quality indexes comprise the compaction degree, the deflection value and the dry density after the road construction is finished. The construction parameters and the construction quality indexes of one road form an original data set, and the original data of a plurality of roads form an original data set.
The construction quality index of each road is obtained by measurement under the same condition.
The construction parameters of each road are obtained through monitoring equipment in the respective construction process, the mixing temperature and the mixing proportion in the asphalt mixing process are obtained through monitoring of an intelligent monitoring system of an asphalt mixing station, a plurality of infrared temperature sensors are installed on the inner wall of an asphalt mixture stirrer to collect temperature data in real time, the temperature data are uploaded to the intelligent monitoring system of the asphalt mixing station through a 4G/5G network in real time to be subjected to integrated processing of data, and then the temperature data are input to an artificial neural network model as input characteristic values to be subjected to simulation calculation of compaction quality monitoring indexes. Before mixing construction, the dosage proportion of various aggregates is input into an intelligent monitoring system of an asphalt mixing station, and the proportion change of the various aggregates is collected and displayed in real time in the mixing process by utilizing the difference of the particle sizes of the various aggregates.
The temperature of the mixture and the transport capacity of the mixture in the transportation process are acquired through an intelligent transportation process monitoring system, the weighing instrument is installed below the position where wheels are located when the transport vehicle is loaded, and the weight of the transported asphalt mixture can be calculated by utilizing the difference between the temperature of the mixture and the transport capacity of the mixture through real-time dynamic monitoring of the weight of the vehicle before the transport vehicle enters a field and after the transport vehicle is loaded. The weighing instrument uploads data to an intelligent transportation process monitoring system in real time by using a 4G/5G network for data processing, and then the data is input into an artificial neural network model as an input characteristic value for simulation calculation of compaction quality monitoring indexes. And (3) inserting a temperature probe with a built-in temperature measuring chip and a stainless steel pipe and a diameter of 8mm into the asphalt mixture above the asphalt mixture fully stacked in the carriage of the transport vehicle to monitor the temperature in real time in the transport process. Each temperature probe comprises 4 temperature chips, namely four temperature measuring points, the temperature data of different positions of the mixture in the carriage are monitored in real time, measuring point frequency is set according to the transportation time required by the distance, the temperature data of each measuring point is uploaded to an intelligent transportation process monitoring system in real time for data processing, and then the temperature data are input to an artificial neural network model as input characteristic values for carrying out simulation calculation on compaction quality monitoring indexes.
The temperature and the thickness of a paving surface in the paving process are acquired through an intelligent paving process monitoring system, a plurality of infrared temperature sensors are mounted at the front end of the paver to acquire temperature data of the paving surface and a construction environment in real time, the temperature data are uploaded to the intelligent paving process monitoring system for data processing by utilizing a 4G/5G network in real time, and then the temperature data are input to an artificial neural network as input layer parameters to perform simulation calculation of compaction quality monitoring indexes. The thickness sensors are arranged at the front end and the rear end of the paver, elevation data of the front and the rear of a road surface are acquired through the sensors in the paving process, the data are uploaded to an intelligent monitoring system in the paving process in real time through a 4G/5G network to be subjected to integrated processing of the data, and then the data are input to an artificial neural network as input layer parameters to be subjected to analog calculation of quality monitoring indexes.
The rolling times, the surface layer rolling temperature and the acceleration of the road roller in the rolling process are acquired through an intelligent rolling process monitoring system, an acceleration sensor is mounted on a wheel of the road roller to acquire acceleration data in real time, the data are uploaded to the intelligent rolling process monitoring system through a 4G/5G network in real time to be subjected to integrated processing of the data, and then the data are input to an artificial neural network as input layer parameters to perform simulation calculation of quality monitoring indexes. An infrared temperature sensor is arranged at the front end of the road roller to acquire surface layer temperature data in real time, the temperature data is uploaded to an intelligent monitoring system in the rolling process in real time by using a 4G/5G network for integrated processing of the data, and then the temperature data is input to an artificial neural network as an input layer parameter for analog calculation of a quality monitoring index. The rolling times of the rolling machine are input into the intelligent monitoring system through a constructor.
Step two: obtaining a dataset of a network model
Preprocessing the original data sets of different roads in the same region obtained in the step one to obtain a data set D of a network model, dividing the data set D into a training set S and a test set T according to a ratio of 7:3 by a reservation method, wherein D is S and U,
after training the model at S, T is used to estimate its test error as an estimate of the generalization error. S, T, the division adopts a layered sampling method to avoid the influence on the final result caused by introducing extra deviation in the data division process.
The specific method for preprocessing the raw data sets of different roads is as follows:
taking the maximum value x of each characteristic value (the characteristics comprise construction parameters and construction quality indexes) in the original data sets of different roadsmaxMinimum value xminAnd normalizing each corresponding characteristic value. Order to
In the formula, xmaxAnd xminMinimum and maximum values, x, of the characteristics i of different roads, respectivelyiThe value of the characteristic i in the original data of a certain road, xi' is the value of the characteristic i of the road after normalization; a. b is a parameter, and a is 0.9, and b is (1-a)/2. Since the transfer function of the neuron is a Sigmoid function, the output of the neuron can be prevented from entering a saturation state.
And carrying out normalization processing on the original data of each road according to the mode, taking the normalized data as a data set D of the network model, wherein the number of the roads is the number of data contained in the data set D.
Step three: construction of road construction feedback neural network model
The BP algorithm is used as an example to construct a road construction feedback neural network model, but is not limited to the BP algorithm.
The method comprises the steps of constructing a road construction feedback neural network model based on a BP algorithm, wherein the road construction feedback neural network model is a three-layer perceptron neural network model, an input layer of the road construction feedback neural network model is construction parameters of a road, an output layer of the road construction feedback neural network model is construction quality indexes, the road construction feedback neural network model comprises two hidden layers, the first hidden layer is provided with four hidden nodes, and the second hidden layer is provided with three hidden nodes.
Setting an input layer as M, namely, M input signals exist, wherein any one input signal is represented by M; the first hidden layer is I, i.e. there are I neurons, any of which is represented by I; the second hidden layer is J, i.e. there are J neurons, where any neuron is denoted by J; the output layer is P, namely P output neurons, the number of the output neurons is the same as that of the construction quality indexes, and any neuron is represented by P.
The weights of the input layer and the first hidden layer are wmiRepresents; the weight of the first hidden layer and the second hidden layer is wijRepresents; the weight of the second hidden layer and the output layer is wjpAnd (4) showing. The input to each neuron is denoted u, the stimulus output is denoted v, the superscripts of u, v denote layers, and the subscripts denote a neuron in a layer, e.g.
Representing the input of the ith neuron of the first hidden layer. The stimulus functions of all neurons are Sigmoid functions.
And taking the construction parameters of the training set S in the step two as the input quantity of the input layer, and taking the construction quality index as the output quantity of the output layer. Let X be [ X ] as the input sample set of the training set S1，X2，…，Xk，…，XN]Correspond toAny of the training samples of (1) is Xk＝[xk1，xk2，…xkm]K is 1,2, … N, and the actual output is Yk＝[yk1，yk2，…ykP]TThe desired output is dk＝[dk1，dk2，…dkP]T. N represents the number of roads, m represents the number of construction parameters, k represents the kth road, P represents the number of construction quality indexes, dkAnd corresponding to the construction quality index of the kth road in the training set S.
And setting n as iteration times, wherein the weight and the actual output are both functions of n. Input training sample XkThe neural network is obtained by the forward propagation process of the working signal
The error signal of the p-th neuron of the output layer is:
ekp(n)＝dkp(n)-ykp(n)
defining the error energy of the neuron p as
The sum of the errors of all neurons in the output layer is e (n):
the error sum signal is transmitted from back to front, and the weight values are modified layer by layer in the process of back propagation. And when the error sum is smaller than a threshold epsilon, namely | E (n) | < epsilon, the error sum is considered to meet the requirement.
A mathematical model with a complex nonlinear relation between the mapping construction parameters and the construction quality indexes can be obtained by utilizing the training set S, and the mathematical model is a road construction feedback neural network model.
The specific steps of the weight vector training process of the road construction feedback neural network model are as follows:
1) setting variables and parameters
Construction for setting roadThe input quantity of the feedback neural network model is a training sample Xk＝[xk1，xk2，…xkm]Where N is the number of input quantities, (k is 1,2, … N), that is, the training set S includes the feature values of N roads.
Order to
The weight vector between the input layer M and the hidden layer I in the nth iteration is shown.
Order to
Is the weight vector between the hidden layer I and the hidden layer J at the nth iteration.
Order to
Is the weight vector between the hidden layer J and the output layer P at the nth iteration.
Let Yk(n)＝[yk1(n)，yk2(n)，…ykP(n)]TAnd (k — 1,2, … N) is the actual output value of the neural network at the nth iteration.
dk＝[dk1，dk2，…dkP]TAnd (k ═ 1,2, … N), is the desired output. Eta is learning efficiency, and eta is 0.1; n is the number of iterations, n is 1000, and the threshold epsilon of the error sum e (n) is 0.0001.
2) Initializing weight vector, and assigning weight vector WMI(0)、WIJ(0)、WJP(0) Each element in (1) is a random non-zero value in the range of (-2.4/F ), wherein F is the number of input ends of connected neurons, and n is 0. For weight vector WMI(0) F ═ M; for weight vector WIJ(0) F ═ I; weight vector pair WJP(0)，F＝J。
3) Inputting a random training sample Xk。
4) For input training sample XkForward computing input signal u and output signal v for each layer of neurons in the BP neural network, wherein
P is 1,2, …, P, and the actual output Y is obtainedk(n)。
5) From the desired output dkAnd the actual output Y obtained in the previous stepk(n), calculating the error sum E (n), judging whether the error sum E (n) meets the requirement, if so, turning to the step 8, and if not, turning to the step 6).
6) Judging whether n +1 is greater than the maximum iteration times, and if so, turning to the step 8); if not, the input training sample X is processedkThe local gradient δ for each layer of neurons is calculated in reverse.
Wherein
Is the local gradient between the input layer M and the hidden layer I,
for the local gradient between the hidden layer I and the hidden layer J,
is the local gradient between the hidden layer J and the output layer P. f' () is the derivative function of the Sigmoid function of the excitation function.
7) Calculating the weight correction quantity delta w according to the following formula, and correcting the weight:
i＝1，2，…，I；j＝1，2，…，J
p＝1，2，…，P；m＝1，2，…，M
and taking the vector formed by the corrected weights as a new weight vector, turning to the step 4), and performing the next iteration, wherein the iteration number n is n +1 (the iteration number is added with 1).
8) The weight vector W in the last iteration according to the previous training sampleMI、WIJ、WJPRepeating the processes from step 2) to step 7) as the initial weight vector of the next training sample which is not repeated randomly, and according to the weight vector W in the last iteration of the last training sampleMI、WIJ、WJPAnd (5) obtaining the trained road construction feedback neural network model as the weight vector of the road construction feedback neural network model.
Step four: verifying the effectiveness of the road construction feedback neural network model trained in the third step
And (5) taking the construction parameters of different roads in the test set T as input quantity of an input layer, and inputting the input quantity into the trained road construction feedback neural network model to obtain an output value of the neural network model. And comparing and analyzing the output value of the obtained neural network model with the corresponding construction quality index in the test set T, comparing the output value with the construction quality index in the test set T, calculating a deviation value, and when the deviation value of more than 90 percent of the construction quality indexes is less than 3 percent, determining that the road construction feedback neural network model is an effective model.
The calculation method of the deviation value comprises the following steps:
in the formula: pCIs the magnitude of the deviation value, c1For one of the output values obtained via the neural network model, c2And testing the corresponding construction quality index in the set T.
Step five: adjusting construction parameters by applying road construction feedback neural network model and determining optimal value
Inputting construction parameters of a road which is not constructed in the same area into a road construction feedback neural network model, and predicting construction quality indexes after the construction of the road is finished through calculation of the neural network model; and adjusting the construction parameters in unfinished procedures by combining the predicted construction quality indexes, setting multiple groups of adjustable parameters in the subsequent construction process of the road, calculating and feeding back the construction quality indexes corresponding to each group of settings through a neural network model, determining a group of settings corresponding to the construction quality indexes meeting requirements as the optimal values of the adjustable parameters, and performing subsequent construction by using the optimal values.
In the embodiment, the road construction parameters specifically include the mixing temperature and the mixing proportion in the asphalt mixing process; the temperature and the transport capacity of the mixture in the transportation process; the temperature and thickness of a paving surface in the paving process; rolling times, surface layer rolling temperature and acceleration of the road roller in the rolling process.
And for a road without construction, inputting the set road construction parameters into a road construction feedback neural network model, and obtaining a corresponding construction quality index through calculation of the neural network model. And comparing the obtained construction quality index with a preset road quality requirement, if the construction quality index meets the requirement, constructing according to the road construction parameter, and if the construction quality index does not meet the requirement, adjusting the road construction parameter until the construction quality index obtained by calculation meets the preset road quality requirement, wherein the corresponding construction parameter is the optimal value.
For a road under construction, real-time measured road construction parameters in each process are input into a road construction feedback neural network model in real time, once deviation occurs in measured data, multiple sets of adjustable parameters in the subsequent construction process of the road are set according to the construction parameters of the previous process (the parameters which are constructed are the uncontrollable parameters, and the parameters which are not constructed are the adjustable parameters), the optimal values are determined through calculation and feedback of the neural network model, and subsequent construction is carried out according to the optimal values, so that the influence of the uncontrollable factors on construction quality indexes is reduced.
Nothing in this specification is said to apply to the prior art.
Claims (3)
1. An intelligent feedback optimization method for the whole process of highway construction quality is characterized by comprising the following steps:
the method comprises the following steps: obtaining a raw data set
Acquiring original data of different roads in the same region, wherein the original data comprises construction parameters of the roads and construction quality indexes after the road construction is finished, and the construction parameters comprise mixing temperature and mixing proportion in the asphalt mixing process; the temperature and the transport capacity of the mixture in the transportation process; the temperature and thickness of a paving surface in the paving process; rolling times, surface layer rolling temperature and acceleration of the road roller in the rolling process; the construction quality indexes comprise compactness, deflection value and dry density after the road construction is finished; the construction parameters and the construction quality indexes of one road form an original data set, and the original data of a plurality of roads form an original data set; the construction quality index of each road is obtained by measurement under the same condition;
step two: obtaining a dataset of a network model
Preprocessing the original data sets of different roads in the same region obtained in the step one to obtain a data set D of a network model, dividing the data set D into a training set S and a test set T according to a ratio of 7:3 by a reservation method, wherein D is S U T,
step three: construction of road construction feedback neural network model
Constructing a road construction feedback neural network model based on a BP algorithm, wherein the road construction feedback neural network model is a three-layer perceptron neural network model, an input layer of the road construction feedback neural network model is construction parameters of a road, an output layer of the road construction feedback neural network model is a construction quality index and comprises two hidden layers, wherein the first hidden layer is provided with four hidden nodes, and the second hidden layer is provided with three hidden nodes;
setting an input layer as M, namely, M input signals exist, wherein any one input signal is represented by M; the first hidden layer is I, i.e. there are I neurons, any of which is represented by I; the second hidden layer is J, i.e. there are J neurons, where any neuron is denoted by J; the output layer is P, namely P output neurons, the number of the output neurons is the same as that of the construction quality indexes, and any neuron is represented by P;
the weights of the input layer and the first hidden layer are wmiRepresents; the weight of the first hidden layer and the second hidden layer is wijRepresents; the weight of the second hidden layer and the output layer is wjpRepresents; the input to each neuron is denoted u, the stimulus output is denoted v, the superscripts of u, v denote layers, and the subscripts denote a neuron in a layer, e.g.
An input representing an ith neuron of the first hidden layer; the excitation functions of all the neurons are Sigmoid functions;
taking the construction parameters of the training set S in the step two as the input quantity of an input layer, and taking the construction quality index as the output quantity of an output layer; let X be [ X ] as the input sample set of the training set S1，X2，…，Xk，…，XN]Any corresponding training sample is Xk＝[xk1，xk2，…xkm]K is 1,2, … N, and the actual output is Yk＝[yk1，yk2，…ykp]TThe desired output is dk＝[dk1，dk2，…dkP]T(ii) a N represents the number of roads, m represents the number of construction parameters, k represents the kth road, P represents the number of construction quality indexes, dkCorresponding to the construction quality index of the kth road in the training set S;
setting n as iteration times, wherein the weight and the actual output are both functions of n; input training sample XkThe neural network is obtained by the forward propagation process of the working signal
The error signal of the p-th neuron of the output layer is:
ekp(n)＝dkp(n)-ykp(n)
defining the error energy of the neuron p as
The sum of the errors of all neurons in the output layer is e (n):
transmitting the error sum signal from back to front, and modifying the weight layer by layer in the process of back propagation; when the error sum is smaller than a threshold epsilon, namely | E (n) | < epsilon, the error sum is considered to meet the requirement;
the specific steps of the weight vector training process of the road construction feedback neural network model are as follows:
1) setting variables and parameters
Setting the input quantity of a road construction feedback neural network model as a training sample Xk＝[xk1，xk2，…xkm](k is 1,2, … N), where N is the number of input quantities, that is, the training set S includes the feature values of N roads;
order to
For the nth iterationA weight vector between the entry layer M and the hidden layer I;
order to
The weight vector between the hidden layer I and the hidden layer J in the nth iteration is obtained;
order to
The weight vector between the hidden layer J and the output layer P in the nth iteration is shown;
let Yk(n)＝[yk1(n)，yk2(n)，…ykP(n)]T(k ═ 1,2, … N), which is the actual output value of the neural network at the nth iteration;
dk＝[dk1，dk2，…dkP]T(k ═ 1,2, … N), is the desired output; eta is learning efficiency, and eta is 0.1; n is iteration number, n is 1000, and the threshold value epsilon of the error sum E (n) is 0.0001;
2) initializing weight vector, and assigning weight vector WMI(0)、WIJ(0)、WJP(0) Each element in (1) is a random non-zero value in the range of (-2.4/F ), wherein F is the number of input ends of connected neurons, and n is 0; for weight vector WMI(0) F ═ M; for weight vector WIJ(0) F ═ I; weight vector pair WJP(0)，F＝J；
3) Inputting a random training sample Xk；
4) For input training sample XkForward computing input signal u and output signal v for each layer of neurons in the BP neural network, wherein
Obtain the actual output Yk(n)；
5) From the desired output dkAnd the actual output Y obtained in the previous stepk(n), calculating the error sum E (n), judging whether the error sum E (n) meets the requirement, if so, turning to the step 8, and if not, turning to the stepStep 6);
6) judging whether n +1 is greater than the maximum iteration times, and if so, turning to the step 8); if not, the input training sample X is processedkCalculating the local gradient delta of each layer of neurons in a reverse mode;
wherein
Is the local gradient between the input layer M and the hidden layer I,
for the local gradient between the hidden layer I and the hidden layer J,
is the local gradient between the hidden layer J and the output layer P; f' () is a derivative function of the Sigmoid function of the excitation function;
7) calculating the weight correction quantity delta w according to the following formula, and correcting the weight:
i＝1，2，…，I；j＝1，2，…，J
p＝1，2，…，P；m＝1，2，…，M
taking the vector formed by the corrected weights as a new weight vector, turning to the step 4), and performing next iteration, wherein the iteration number n is n + 1;
8) the weight vector W in the last iteration according to the previous training sampleMI、WIJ、WJPRepeating the processes from step 2) to step 7) as the initial weight vector of the next training sample which is not repeated randomly, and according to the weight vector W in the last iteration of the last training sampleMI、WIJ、WJPThe weight vector is used as a weight vector of the road construction feedback neural network model, and the trained road construction feedback neural network model is obtained;
step four: verifying the effectiveness of the road construction feedback neural network model trained in the third step
Taking the construction parameters of different roads in the test set T as input quantity of an input layer, and inputting the input quantity into the trained road construction feedback neural network model to obtain an output value of the neural network model; comparing and analyzing the output value of the obtained neural network model with the corresponding construction quality index in the test set T, comparing the output value with the construction quality index in the test set T, calculating a deviation value, and when the deviation value of more than 90% of the construction quality indexes is less than 3%, determining that the road construction feedback neural network model is an effective model;
the calculation method of the deviation value comprises the following steps:
in the formula: pc is the magnitude of the deviation, c1For one of the output values obtained via the neural network model, c2Corresponding construction quality indexes in the test set T are obtained;
step five: adjusting construction parameters by applying road construction feedback neural network model and determining optimal value
Inputting construction parameters of a road which is not constructed in the same area into a road construction feedback neural network model, and predicting construction quality indexes after the construction of the road is finished through calculation of the neural network model; and adjusting the construction parameters in unfinished procedures by combining the predicted construction quality indexes, setting multiple groups of adjustable parameters in the subsequent construction process of the road, calculating and feeding back the construction quality indexes corresponding to each group of settings through a neural network model, determining a group of settings corresponding to the construction quality indexes meeting requirements as the optimal values of the adjustable parameters, and performing subsequent construction by using the optimal values.
2. The intelligent feedback optimization method for the whole process of highway construction quality according to claim 1, wherein in the second step, the training set S and the testing set T are divided by a layered sampling leave-out method.
3. The intelligent feedback optimization method for the whole process of highway construction quality according to claim 1, wherein the concrete method for preprocessing the original data sets of different roads in the step two is as follows:
taking the maximum value x of each characteristic value in the original data sets of different roadsmaxMinimum value xminNormalizing each corresponding characteristic value; order to
In the formula, xmaxAnd xminMinimum and maximum values, x, of the characteristics i of different roads, respectivelyiThe value of the characteristic i in the original data of a certain road, xiIs normalizationThen, the value of the characteristic i of the road is obtained; a. b is a parameter, and a is 0.9, and b is (1-a)/2;
and carrying out normalization processing on the original data of each road according to the mode, taking the normalized data as a data set D of the network model, wherein the number of the roads is the number of data contained in the data set D.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110195411.0A CN112862200B (en) | 2021-02-20 | 2021-02-20 | Intelligent feedback optimization method for whole process of highway construction quality |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110195411.0A CN112862200B (en) | 2021-02-20 | 2021-02-20 | Intelligent feedback optimization method for whole process of highway construction quality |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112862200A true CN112862200A (en) | 2021-05-28 |
| CN112862200B CN112862200B (en) | 2022-08-02 |
Family
ID=75988408
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110195411.0A Active CN112862200B (en) | 2021-02-20 | 2021-02-20 | Intelligent feedback optimization method for whole process of highway construction quality |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112862200B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115034697A (en) * | 2022-08-12 | 2022-09-09 | 河北工业大学 | Multi-domain analysis-based multivariate intelligent compaction index grading optimization method and system |
| CN117671396A (en) * | 2024-02-02 | 2024-03-08 | 新疆盛诚工程建设有限责任公司 | Intelligent monitoring and early warning system and method for construction progress |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150030392A1 (en) * | 2012-04-06 | 2015-01-29 | The Board Of Regents Of The University Of Oklahoma | Method and apparatus for determining stiffness of a roadway |
| CN107142810A (en) * | 2017-05-23 | 2017-09-08 | 济南汇通联合市政工程有限责任公司 | A kind of high-viscosity bitumen modifier asphalt paving construction method |
| CN108712809A (en) * | 2018-05-18 | 2018-10-26 | 浙江工业大学 | A kind of luminous environment intelligent control method based on neural network |
| CN109917115A (en) * | 2019-01-23 | 2019-06-21 | 长安大学 | A kind of asphalt comprehensive performance prediction technique |
| CN111105332A (en) * | 2019-12-19 | 2020-05-05 | 河北工业大学 | Highway intelligent pre-maintenance method and system based on artificial neural network |
| CN112252292A (en) * | 2020-10-15 | 2021-01-22 | 河北工业大学 | Real-time highway compactness monitoring method based on artificial neural network |
-
2021
- 2021-02-20 CN CN202110195411.0A patent/CN112862200B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150030392A1 (en) * | 2012-04-06 | 2015-01-29 | The Board Of Regents Of The University Of Oklahoma | Method and apparatus for determining stiffness of a roadway |
| CN107142810A (en) * | 2017-05-23 | 2017-09-08 | 济南汇通联合市政工程有限责任公司 | A kind of high-viscosity bitumen modifier asphalt paving construction method |
| CN108712809A (en) * | 2018-05-18 | 2018-10-26 | 浙江工业大学 | A kind of luminous environment intelligent control method based on neural network |
| CN109917115A (en) * | 2019-01-23 | 2019-06-21 | 长安大学 | A kind of asphalt comprehensive performance prediction technique |
| CN111105332A (en) * | 2019-12-19 | 2020-05-05 | 河北工业大学 | Highway intelligent pre-maintenance method and system based on artificial neural network |
| CN112252292A (en) * | 2020-10-15 | 2021-01-22 | 河北工业大学 | Real-time highway compactness monitoring method based on artificial neural network |
Non-Patent Citations (2)
| Title |
|---|
| 杨学超等: "基于BP人工神经网络的路基压实度预测模型研究", 《甘肃科学学报》 * |
| 王树栋等: "长大隧道软弱围岩施工大变形智能预测方法", 《中国铁道科学》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115034697A (en) * | 2022-08-12 | 2022-09-09 | 河北工业大学 | Multi-domain analysis-based multivariate intelligent compaction index grading optimization method and system |
| CN115034697B (en) * | 2022-08-12 | 2022-11-08 | 河北工业大学 | Multi-domain analysis-based multi-element intelligent compaction index grading optimization method and system |
| CN117671396A (en) * | 2024-02-02 | 2024-03-08 | 新疆盛诚工程建设有限责任公司 | Intelligent monitoring and early warning system and method for construction progress |
| CN117671396B (en) * | 2024-02-02 | 2024-04-26 | 新疆盛诚工程建设有限责任公司 | Intelligent monitoring and early warning system and method for construction progress |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112862200B (en) | 2022-08-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105973444B (en) | A kind of improved truck scale weighing method | |
| CN112862200B (en) | Intelligent feedback optimization method for whole process of highway construction quality | |
| Taghavifar et al. | Application of artificial neural networks for the prediction of traction performance parameters | |
| CN112926265A (en) | Atmospheric porous probe measurement calibration method based on genetic algorithm optimization neural network | |
| CN103577888A (en) | Improved entropy weight AHP and application thereof | |
| CN109917115A (en) | A kind of asphalt comprehensive performance prediction technique | |
| CN109918972A (en) | A kind of driving weight intelligent control method and system | |
| CN104915515A (en) | BP neural network based GFET modeling method | |
| CN111967308A (en) | Online road surface unevenness identification method and system | |
| CN108846200B (en) | Quasi-static bridge influence line identification method based on iteration method | |
| CN106568647A (en) | Nerve network-based concrete strength predication method | |
| CN116778435A (en) | Vehicle axle weight distribution identification method based on computer vision technology | |
| CN112163669A (en) | Pavement subsidence prediction method based on BP neural network | |
| CN110108672B (en) | Aerosol extinction coefficient inversion method based on deep belief network | |
| CN112085348A (en) | Soil fertility assessment method based on fuzzy neural network | |
| CN115948964A (en) | Road flatness prediction method based on GA-BP neural network | |
| CN104899642B (en) | Large deformation flexible body dynamic stress compensation method based on hybrid production style | |
| CN114254498A (en) | Engine dynamic oil consumption prediction method and system based on BP neural network | |
| CN110704789B (en) | Population dynamic measurement and calculation method and system based on 'urban superconcephalon' computing platform | |
| CN105974058B (en) | The method of quick detection preservative water preperation based on electronic nose-artificial neural network | |
| Yang et al. | Automatic measurement of payload for heavy vehicles using strain gages | |
| Liu et al. | Fuzzy optimization BP neural network model for pavement performance assessment | |
| CN104573816B (en) | The neural network clustering method of microwave radiometer remote sensing atmosphere parameter | |
| CN117709203A (en) | Comprehensive error modeling method and system for hydrostatic lead screw feeding system | |
| CN115165770B (en) | Water COD and turbidity simultaneous detection method based on broad spectrum and BPNN |
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 |