CN114035529B - ATL-BMA-based nonlinear industrial process low-cost modeling method - Google Patents
ATL-BMA-based nonlinear industrial process low-cost modeling method Download PDFInfo
- Publication number
- CN114035529B CN114035529B CN202111411517.6A CN202111411517A CN114035529B CN 114035529 B CN114035529 B CN 114035529B CN 202111411517 A CN202111411517 A CN 202111411517A CN 114035529 B CN114035529 B CN 114035529B
- Authority
- CN
- China
- Prior art keywords
- industrial process
- data
- old
- new
- modeling
- 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
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 301
- 238000000034 method Methods 0.000 title claims abstract description 108
- 238000012549 training Methods 0.000 claims abstract description 59
- 230000008569 process Effects 0.000 claims abstract description 54
- 238000013508 migration Methods 0.000 claims abstract description 40
- 230000005012 migration Effects 0.000 claims abstract description 35
- 238000012706 support-vector machine Methods 0.000 claims abstract description 27
- 238000010606 normalization Methods 0.000 claims abstract description 14
- 230000004927 fusion Effects 0.000 claims abstract description 8
- 238000013507 mapping Methods 0.000 claims abstract description 8
- 238000012545 processing Methods 0.000 claims abstract description 4
- 238000012360 testing method Methods 0.000 claims description 29
- 238000006243 chemical reaction Methods 0.000 claims description 8
- 230000006870 function Effects 0.000 claims description 8
- 238000005070 sampling Methods 0.000 claims description 8
- 238000012935 Averaging Methods 0.000 claims description 6
- 238000013526 transfer learning Methods 0.000 claims description 5
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 208000009119 Giant Axonal Neuropathy Diseases 0.000 claims description 3
- 238000009826 distribution Methods 0.000 claims description 3
- 201000003382 giant axonal neuropathy 1 Diseases 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 238000012843 least square support vector machine Methods 0.000 claims description 2
- 230000000694 effects Effects 0.000 description 6
- 239000004973 liquid crystal related substance Substances 0.000 description 4
- 230000007246 mechanism Effects 0.000 description 3
- 238000009776 industrial production Methods 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 230000002411 adverse Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/418—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM]
- G05B19/41885—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM] characterised by modeling, simulation of the manufacturing system
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/30—Nc systems
- G05B2219/32—Operator till task planning
- G05B2219/32339—Object oriented modeling, design, analysis, implementation, simulation language
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Engineering & Computer Science (AREA)
- Manufacturing & Machinery (AREA)
- General Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Feedback Control In General (AREA)
Abstract
The invention provides a nonlinear industrial process low-cost modeling method based on ATL-BMA, which comprises the steps of selecting N groups of similar old process modeling data; collecting a new process modeling initial dataset; dividing new and old process data into two parts respectively, and carrying out normalization processing respectively; converting N groups of old process data into N groups of old process data with new process information, mixing the N groups of old process data with corresponding old process data to obtain N groups of mixed data sets, and training a support vector machine model to obtain N old process basic models with the new process information; mapping the input variables of the new process training set into the operation interval of the input variables of the similar old process, and obtaining the fusion output of the N prediction models; and taking the fusion output of the old process SVM model and the new process input data as the input data of the multi-model migration strategy, and training to obtain a new process model. The method can effectively solve the problems of high modeling cost, limited acquired modeling data and long modeling period of the complex industrial process.
Description
Technical Field
The invention belongs to the technical field of industrial process construction performance prediction models, and particularly relates to a nonlinear industrial process low-cost modeling method based on ATL-BMA.
Background
Modern industrial processes are moving toward upsizing, high efficiency and integration in order to meet the demands of the market for products with multiple specifications, multiple varieties and high quality. On the one hand, with the gradual expansion of the production scale, new industrial production processes are added in the actual production process continuously to meet different product requirements, which leads to the higher and higher complexity of the actual industrial production process. On the other hand, changes in the operating environment and increases in the operating time can cause changes in the characteristics of the actual industrial process. Both of these aspects result in variable characteristics of the process data. Modeling an industrial process with a data-driven approach in this case requires solving a troublesome problem: due to various factors such as cost, modeling data obtained from a new industrial process are seriously insufficient, an accurate process prediction model cannot be established by using a data-driven modeling method under the support of a small amount of modeling data, and meanwhile, the generalization capability of the obtained model is low. In the face of this, it is desirable that existing long run-time industrial process data or knowledge can assist in guiding the creation of predictive models of new industrial processes. Although the characteristics of the operation data of the new and old industrial processes are different to a certain extent, the physical and chemical mechanism which is followed by the inside of the process is unchanged or very similar, so that the new industrial process data and the old industrial process data have the same or similar characteristic space and label space (the dimensions of the input and output data of the new industrial process data are consistent). As shown in fig. 1, the new and old industrial processes can be treated as a target domain and a source domain, respectively, and then the old industrial process data is used to assist in building a new industrial process prediction model through a transfer learning method. However, when the source domain data is far more than the target domain data, the phenomenon of 'negative migration' is easy to occur when the source domain data is used for carrying out supplementary learning on the target domain data under the traditional migration learning structure.
Disclosure of Invention
Aiming at the problems existing in the prior art, the invention provides a nonlinear industrial process low-cost modeling method based on ATL-BMA, which can effectively solve the problems of high modeling cost, limited acquired modeling data and long modeling period of a complex industrial process, and can solve the phenomenon of negative migration which can occur when old process data are far more than new process data in migration learning.
In order to achieve the above object, the present invention provides a method for modeling a nonlinear industrial process at low cost based on an ATL-BMA, comprising the steps of:
step 1: selecting N groups of similar old industrial process modeling data, and determining a stable operation range of an input variable according to information of an actual process to be modeled; meanwhile, a Latin hypercube method is selected for sampling and collecting an initial data set of modeling of a target nonlinear industrial process; the method comprises the following specific steps:
step 1.1: selecting N groups of similar old industrial process modeling data, and recording the data asModeling an ith old industrial process according to equation (1);
where X and X are old industrial process input data, y is old industrial process output data, k i Representing the modeling data volume of the ith old industrial process, wherein n is the dimension of an input variable of the ith old industrial process, and the dimension of the input variable of all industrial processes is consistent and n is equal because of certain similarity of the new industrial process and the old industrial process;
step 1.2: according to the information of the actual process to be modeled, determining the stable operation range of the input variable, selecting discrete sparse data distribution points for sampling and collecting new industrial process modeling data, and obtaining collected new industrial process data D according to a formula (2) new ；
Wherein, l represents the modeling data amount of the new industrial process;
step 2: dividing new industrial process data and old industrial process data into two parts, namely a training data set and a testing data set in a new modeling process and an old modeling process respectively, and carrying out normalization processing on an initial data set of the new industrial process and the modeling data of the old industrial process respectively; wherein, the liquid crystal display device comprises a liquid crystal display device,for new industrial process data, it is divided into new industrial process training data setsAnd new industrial process test data set +.>And mapping the data to [0,1 ] using equation (3)]A section;
wherein z is i Representing the results after normalization of the input or output data of an industrial process, x i Is the data before normalization, x max Is the maximum value before data normalization, x min Is the minimum value;
step 3: converting N groups of old industrial process data into N groups of old industrial process data with new industrial process information by using a new and old industrial process data migration algorithm based on a Cycle GANs; wherein the old industrial process training data set isThe method comprises the following specific steps:
step 3.1: initializing parameters: g parameter θ G ，D o Parameter omega o F parameter θ F ，D n Parameter omega n ，n critic ＝5，α＝0.00005、β 1 ＝0、β 2 ＝0.7，m＝5，λ＝0.5，Epoch＝20000；
Wherein: g represents a generator function of old industrial process to new industrial process data, D o Representing the corresponding discriminators of the old industrial process, F representing the generator function from the new industrial process to the old industrial process, D n Indicating the corresponding discriminant of the new industrial process, n critic Representing the number of training discrimination models, alpha, beta, after training a generator 1 And beta 2 The parameters of the Adam optimizer are that m is the sampling number and Epoch is the model cycle training times;
step 3.2: will be from the first through generator Gi old industrial process dataM samples taken from (1)Conversion to m new industrial process data, denoted X o→n ＝F(X o ) The method comprises the steps of carrying out a first treatment on the surface of the From the new industrial process data +.>M samples collected in->Conversion to m old industrial process data, denoted X n→o ＝F(X n )；
Step 3.3: obtaining a discriminator loss and two forward circulation consistent losses according to a formula (4) and a formula (5);
step 3.4: updating the discriminant D by equation (6) and equation (7) o Parameter omega o And D n Parameter omega n ；
Step 3.5: repeating the steps 3.2 to 3.4n critic Secondary times;
step 3.6: repeating the step 3.2;
step 3.7: calculating two forward loop coincidence losses through a formula (8) and a formula (9);
step 3.8: calculating two generator losses by equation (10) and equation (11);
step 3.9: updating the generator G parameter θ by equation (12) and equation (13) G And F parameter θ F ；
Step 3.10: repeating the steps 3.6-3.9 Epoch times, and converting the new industrial process data into the j old industrial process data by using the trained F, and recording as
Step 3.11: repeating the steps 3.1-3.9 by using each set of old industrial process data, transferring the new industrial process data into the old industrial process domain, obtaining N sets of old industrial process data with new industrial process information by the new industrial process data through anti-transfer learning, and recording as
Step 4: mixing old industrial process data with new industrial process information in the step 3 with corresponding old industrial process data to obtain N groups of mixed data sets;
step 5: dividing a hybrid dataset into hybrid training setsAnd hybrid test dataset->At the same time, N old industrial process training data sets are combined +.>And a new industrial process prediction model y=f (x), training support vector machine SVM models by using N groups of mixed data sets respectively to obtain N old industrial process basic models with new industrial process information, and marking as f 1 (·)-f N (. Cndot.); wherein (1)>k train Is training data set size, +.>k test Is the test data set size, any ith old industrial process, < >>n i Is the ith old industrial process training set size; the method comprises the following specific steps:
step 5.1: initializing parameters;
step 5.2: converting new industrial process data into N groups of old industrial process data carrying new industrial process information through new and old industrial process data migration algorithm based on Cycle GANsMixing D according to equation (14) n→o And D o Obtaining N groups of basic model training data D Basic ；
Step 5.3: by D Basic Training N SVMs to obtain N old industrial process basic models with new industrial process information, denoted as f 1 (·)-f N (·)；
Step 6: mapping the input variables of the new industrial process training set into the operation intervals of the input variables of the similar old industrial process through a model fusion formula (15), and recording the input data of the converted new industrial process training set asFusion output of the N prediction models is obtained through Bayesian model averaging algorithm>
Step 7: fusion output of SVM model of old industrial processAnd new industrial process input data->As input data of a multi-model migration strategy, training a new industrial process model by using a least square support vector machine algorithm to obtain new industrial process model output +.>Completing modeling of a new industrial process;
step 8: model verification, namely evaluating the effectiveness of the SVM model by utilizing root mean square error and a determination coefficient according to a formula (16) and a formula (17), and completing a modeling process if the prediction accuracy of the model obtained in the step 7 on a test data set meets an experimental set threshold value; otherwise, repeating the steps 3 to 7, adding new N groups of old industrial process data samples containing new industrial process information into the mixed samples, and continuing training the new industrial process model until the experimental stop condition is met;
where N is the number of test data, y i Is the output of the predictive model and,is the average value of the predicted output, Y i Is the true output of the new industrial process.
The method comprises the steps of firstly, collecting a small sample data set of nonlinear industrial process modeling by using a Latin hypercube method, combining a plurality of similar old process data, and learning a conversion mapping function between new industrial process data and old industrial process data by using an contrast migration algorithm, so that a small amount of new process data is converted into multiple types of old industrial process data with new industrial process information; then a plurality of 'old process models with new process information' are obtained through a support vector machine regression algorithm, and a foundation is established for modeling of a subsequent new industrial process; and finally, migrating a plurality of trained old industrial process prediction models with new industrial process information by using a multi-model migration strategy and a Bayesian model average theory, and combining a small amount of new industrial process data to obtain a final new industrial process performance prediction model. The invention transfers the useful information of a plurality of existing similar old industrial processes to help to establish a new industrial process performance prediction model, and reduces the modeling cost of the new industrial process; meanwhile, in order to effectively solve the problem of 'negative migration' which can occur when old process data is far more than new process data, a new and old process data migration method based on anti-migration learning is adopted, and migration modeling effect is improved. The method effectively solves the problems of high modeling cost and long modeling period of the complex industrial process, fully utilizes the useful information of the existing similar old industrial process model, simultaneously solves the problem of 'negative migration' possibly occurring when old process data are far more than new process data in migration learning, completes the modeling of the new industrial process, reduces the modeling cost, accelerates the modeling speed and improves the modeling precision.
Drawings
FIG. 1 is a flow chart of migration modeling;
FIG. 2 is a flow chart of a non-linear industrial process low-cost modeling method based on challenge transfer learning and Bayesian model averaging theory;
FIG. 3 is a graph of predicted values of the ATL-BMA model, and SVM model over a compressor A test set;
FIG. 4 is a RMSE histogram of ATL-BMA model, and SVM model predictions versus true values;
FIG. 5 is an R of predicted and actual values of ATL-BMA model, BMA model and SVM model 2 A histogram.
Detailed Description
The invention is further described below with reference to examples and figures.
As shown in fig. 1 to 5, the present invention provides a non-linear industrial process low-cost modeling method based on ATL-BMA (anti-migration learning (Adversarial Transform Learning, ATL) and bayesian model averaging (Bayesian Model Averaging, BMA)), comprising the steps of:
step 1: selecting N groups of similar old industrial process modeling data, and determining a stable operation range of an input variable according to information of an actual process to be modeled; meanwhile, selecting a Latin super-vertical (Latin Hypercube Design, LHD) method to sample and collect a modeling initial data set of a target nonlinear industrial process (new industrial process); the information of the actual process to be modeled comprises parameter rated values, performance curves and the like; the method comprises the following specific steps:
step 1.1: selecting N groups of similar old industrial process modeling data, and recording the data asModeling an ith old industrial process according to equation (1);
where X is the old industrial process input data set, X is the old industrial process input data, y is the old industrial process output data, k i Representing the modeling data volume of the ith old industrial process, wherein n is the dimension of an input variable of the ith old industrial process, and the dimension of the input variable of all industrial processes is consistent and n is equal because of certain similarity of the new industrial process and the old industrial process;
step 1.2: according to the information of the actual process to be modeled, determining the stable operation range of the input variable, selecting discrete sparse data distribution points for sampling and collecting new industrial process modeling data, and obtaining collected new industrial process data D according to a formula (2) new ；
Wherein, l represents the modeling data amount of the new industrial process;
step 2: dividing the new industrial process data and the old industrial process data into two parts, namely a training data set and a testing data set in the new and old modeling processes respectively; in order to avoid adverse effects caused by data dimension differences in the stability of the subsequent training process, it is necessary to ensure that the data are normalized, and respectively performing normalization processing on the new industrial process initial data set and the old industrial process modeling data; wherein, for new industrial process data, it is divided into new industrial process training data setsAnd new industrial process test data set +.>And mapping the data to [0,1 ] using a maximum minimum data normalization method according to equation (3)]A section;
wherein z is i Representing the results after normalization of the input or output data of an industrial process, x i Is the data before normalization, x max Is the maximum value before data normalization, x min Is the minimum value;
step 3: converting N groups of old industrial process data into N groups of old industrial process data with new industrial process information by using a new and old industrial process data migration algorithm based on a Cycle GANs; wherein the old industrial process training data set isThe method comprises the following specific steps:
step 3.1: initializing parameters: g parameter θ G ，D o Parameter omega o F parameter θ F ，D n Parameter omega n ，n critic ＝5，α＝0.00005、β 1 ＝0、β 2 ＝0.7，m＝5，λ＝0.5，Epoch＝20000；
Wherein: g represents a generator function of old industrial process to new industrial process data, D o Representing the corresponding discriminators of the old industrial process, F representing the generator function from the new industrial process to the old industrial process, D n Indicating the corresponding discriminant of the new industrial process, n critic Representing the number of training discrimination models, alpha, beta, after training a generator 1 And beta 2 The parameters of the Adam optimizer are that m is the sampling number and Epoch is the model cycle training times;
step 3.2: from the ith old industrial process data by generator GM samples taken from (1)Conversion to m new industrial process data, denoted X o→n ＝F(X o ) The method comprises the steps of carrying out a first treatment on the surface of the From the new industrial process data +.>M samples collected in->Conversion to m old industrial process data, denoted X n→o ＝F(X n )；
Step 3.3: obtaining a discriminator loss and two forward circulation consistent losses according to a formula (4) and a formula (5);
step 3.4: updating the discriminant D by equation (6) and equation (7) o Parameter omega o And D n Parameter omega n ；
Step 3.5: repeating the steps 3.2 to 3.4n critic Secondary times;
step 3.6: repeating the step 3.2;
step 3.7: calculating two forward loop coincidence losses through a formula (8) and a formula (9);
step 3.8: calculating two generator losses by equation (10) and equation (11);
step 3.9: updating the generator G parameter θ by equation (12) and equation (13) G And F parameter θ F ；
Step 3.10: repeating the steps 3.6-3.9 Epoch times, and converting the new industrial process data into the j old industrial process data by using the trained F, and recording as
Step 3.11: repeating the steps 3.1-3.9 by using each set of old industrial process data, transferring the new industrial process data into the old industrial process domain, obtaining N sets of old industrial process data with new industrial process information by the new industrial process data through anti-transfer learning, and recording as
Step 4: mixing old industrial process data with new industrial process information in the step 3 with corresponding old industrial process data to obtain N groups of mixed data sets;
step 5: dividing a hybrid dataset into hybrid training setsAnd hybrid test dataset->At the same time, N old industrial process training data sets are combined +.>And a new industrial process prediction model y=f (x), training a support vector machine SVM (Support Vector Machine) model by using N groups of mixed data sets respectively to obtain N old industrial process basic models with new industrial process information, and marking as f 1 (·)-f N (. Cndot.); wherein, the liquid crystal display device comprises a liquid crystal display device,
k train is the size of the training data set and,k test is the test data set size, any ith old industrial process,n i is the ith old industrial process training set size; the method comprises the following specific steps:
step 5.1: initializing parameters;
step 5.2: converting new industrial process data into N groups of old industrial process data carrying new industrial process information through new and old industrial process data migration algorithm based on Cycle GANsMixing D according to equation (14) n→o And D o Obtaining N groups of basic model training data D Basic ；
Step 5.3: by D Basic Training N SVMs to obtain N old industrial process basic models with new industrial process information, denoted as f 1 (·)-f N (·)；
Step 6: mapping the input variables of the new industrial process training set into the operation intervals of the input variables of the similar old industrial process through a model fusion formula (15), and recording the input data of the converted new industrial process training set asFusion output of the N prediction models is obtained through Bayesian model averaging algorithm>
Step 7: fusion output of SVM model of old industrial processAnd new industrial process input data->As input data of a multi-model migration strategy, training a new industrial process model by utilizing a least squares support vector machine (Least Squares Support Vector Machine, LSSVM) algorithm to obtain a new industrial process model output ∈>Completing modeling of a new industrial process;
step 8: model verification using root mean Square error (Root Mean Square Error, RMSE) and coefficient determination (R-Square, R) according to equation (16) and equation (17), respectively 2 ) Evaluating the effectiveness of the SVM model, and if the prediction accuracy of the model obtained in the step 7 on the test data set meets the experimental set threshold, completing the modeling process; otherwise, repeating the steps 3 to 7, adding new N groups of old industrial process data samples containing new industrial process information into the mixed samples, and continuing training the new industrial process model until the experimental stop condition is met;
where N is the number of test data, y i Is the output of the predictive model and,is the average value of the predicted output, Y i Is the true output of the new industrial process.
The method comprises the steps of firstly, collecting a small sample data set of nonlinear industrial process modeling by using a Latin hypercube method, combining a plurality of similar old process data, and learning a conversion mapping function between new industrial process data and old industrial process data by using an contrast migration algorithm, so that a small amount of new process data is converted into multiple types of old industrial process data with new industrial process information; then a plurality of 'old process models with new process information' are obtained through a support vector machine regression algorithm, and a foundation is established for modeling of a subsequent new industrial process; and finally, migrating a plurality of trained old industrial process prediction models with new industrial process information by using a multi-model migration strategy and a Bayesian model average theory, and combining a small amount of new industrial process data to obtain a final new industrial process performance prediction model. The invention transfers the useful information of a plurality of existing similar old industrial processes to help to establish a new industrial process performance prediction model, and reduces the modeling cost of the new industrial process; meanwhile, in order to effectively solve the problem of 'negative migration' which can occur when old process data is far more than new process data, a new and old process data migration method based on anti-migration learning is adopted, and migration modeling effect is improved. The method effectively solves the problems of high modeling cost and long modeling period of the complex industrial process, fully utilizes the useful information of the existing similar old industrial process model, simultaneously solves the problem of 'negative migration' possibly occurring when old process data are far more than new process data in migration learning, completes the modeling of the new industrial process, reduces the modeling cost, accelerates the modeling speed and improves the modeling precision.
To verify the effectiveness of the method, experimental data was generated using a laboratory centrifugal compressor mechanism model, and a performance prediction model of the centrifugal compressor was built to verify the effectiveness of the proposed modeling method. Four different but similar compressor models were generated A, B, C, D for simulation experiments by modifying the key geometric parameter simulations of the compressor mechanism model. For A, B, C and D four centrifugal compressors, where compressor a was the new compressor to be modeled, a small amount of new industrial process modeling data was generated, while B, C and D centrifugal compressors were old compressors with long run times, generating a large amount of old industrial process modeling data to assist in the creation of new industrial process prediction models. The stable movement intervals of the new compressor and the old compressor are shown in table 1.
Table 1 centrifugal compressor A, B, C, D steady operation interval and corresponding One-Hot encoding
And comparing the prediction effect of the built model with the prediction effects of two groups of comparison experiment models, and further displaying the superiority of the method. The three sets of comparison methods were specifically as follows:
method 1: a small amount of new industrial process data is converted into old industrial process data through anti-migration learning, the old industrial process data are mixed with each group of old industrial process data, a plurality of SVM models are obtained through training, then a new compressor prediction model is built through a multi-model migration strategy, and finally model accuracy is tested by using new compressor test data. The ATL-BMA method was designated in the analysis of experimental results.
Method 2: and training a plurality of old compressor SVM models by using a plurality of sets of old compressor data, then establishing a new compressor prediction model by combining a few new compressor training data through a multi-model migration strategy, and finally testing model accuracy by using new compressor test data. The BMA method was recorded in the analysis of experimental results.
Method 3: and only a small amount of new compressor training data is used for establishing a new compressor SVM model, the new compressor SVM model is used as a new compressor prediction model, and finally, the new compressor test data is used for testing the model precision. The SVM method is marked in the analysis of experimental results.
Figure 3 shows the predicted values of the model on the compressor a test set for three methods. The graph shows that the predicted value of the model built by the ATL-BMA method is the highest in matching degree with the test set, which proves that the ATL-BMA method can effectively utilize the useful information of similar old industrial processes to help the establishment of a new industrial process model, and simultaneously, the ATL-BMA method can effectively utilize the information between the new and old industrial processes more than a simple multi-model migration method.
To further compare the accuracy of the three models, FIGS. 4 and 5 illustrate the RMSE and R of the predicted and actual values of the three models 2 As can be seen from the figure, the method provided by the chapter can fully utilize a large amount of old industrial process data and a small amount of new industrial process data, effectively improve the prediction precision of the model and reduce the cost of establishing the model.
According to the analysis, the performance prediction model is built for the new industrial process by adopting the anti-migration learning method and the Bayesian model average theory, the existing similar old industrial process performance prediction model in the industry is fully utilized, old and new industrial process data are migrated, a plurality of old industrial process prediction models containing new industrial process information are built by using a support vector machine, and finally the old industrial process model is trained by utilizing the Bayesian model average theory, so that the modeling speed of the new industrial process is accelerated, the modeling cost is reduced, the negative migration effect caused by more old industrial process data than the new industrial process data in the old and new industrial process migration modeling is solved, and the prediction model meeting the precision requirement is obtained. Meanwhile, the method can effectively utilize the information between new and old industrial processes compared with a simple multi-model migration method. Closer to the actual output, a significant cost reduction for modeling industrial processes.
Claims (1)
1. A method for modeling a nonlinear industrial process at low cost based on ATL-BMA, comprising the steps of:
step 1: selecting N groups of similar old industrial process modeling data, and determining a stable operation range of an input variable according to information of an actual process to be modeled; meanwhile, a Latin hypercube method is selected for sampling and collecting an initial data set of modeling of a target nonlinear industrial process; the method comprises the following specific steps:
step 1.1: selecting N groups of similar old industrial process modelsData, noted asModeling an ith old industrial process according to equation (1);
where X and X are old industrial process input data, y is old industrial process output data, k i Representing the modeling data volume of the ith old industrial process, wherein n is the dimension of an input variable of the ith old industrial process, and the dimension of the input variable of all industrial processes is consistent and n is equal because of certain similarity of the new industrial process and the old industrial process;
step 1.2: according to the information of the actual process to be modeled, determining the stable operation range of the input variable, selecting discrete sparse data distribution points for sampling and collecting new industrial process modeling data, and obtaining collected new industrial process data D according to a formula (2) new ；
Wherein, l represents the modeling data amount of the new industrial process;
step 2: dividing new industrial process data and old industrial process data into two parts, namely a training data set and a testing data set in a new modeling process and an old modeling process respectively, and carrying out normalization processing on an initial data set of the new industrial process and the modeling data of the old industrial process respectively; wherein, for new industrial process data, it is divided into new industrial process training data setsAnd new industrial process test data set +.>And utilizeEquation (3) maps data to [0,1 ]]A section;
wherein z is i Representing the results after normalization of the input or output data of an industrial process, x i Is the data before normalization, x max Is the maximum value before data normalization, x min Is the minimum value;
step 3: converting N groups of old industrial process data into N groups of old industrial process data with new industrial process information by using a new and old industrial process data migration algorithm based on a Cycle GANs; wherein the old industrial process training data set isThe method comprises the following specific steps:
step 3.1: initializing parameters: g parameter θ G ，D o Parameter omega o F parameter θ F ，D n Parameter omega n ，n critic ＝5，α＝0.00005、β 1 ＝0、β 2 ＝0.7，m＝5，λ＝0.5，Epoch＝20000；
Wherein: g represents a generator function of old industrial process to new industrial process data, D o Representing the corresponding discriminators of the old industrial process, F representing the generator function from the new industrial process to the old industrial process, D n Indicating the corresponding discriminant of the new industrial process, n critic Representing the number of training discrimination models, alpha, beta, after training a generator 1 And beta 2 The parameters of the Adam optimizer are that m is the sampling number and Epoch is the model cycle training times;
step 3.2: from the ith old industrial process data by generator GM samples collected in->Conversion to m new industrial process data, denoted X o→n ＝F(X o ) The method comprises the steps of carrying out a first treatment on the surface of the From the new industrial process data +.>M samples collected in->Conversion to m old industrial process data, denoted X n→o ＝F(X n )；
Step 3.3: obtaining a discriminator loss and two forward circulation consistent losses according to a formula (4) and a formula (5);
step 3.4: updating the discriminant D by equation (6) and equation (7) o Parameter omega o And D n Parameter omega n ；
Step 3.5: repeating the steps 3.2 to 3.4n critic Secondary times;
step 3.6: repeating the step 3.2;
step 3.7: calculating two forward loop coincidence losses through a formula (8) and a formula (9);
step 3.8: calculating two generator losses by equation (10) and equation (11);
step 3.9: updating the generator G parameter θ by equation (12) and equation (13) G And F parameter θ F ；
Step 3.10: repeating the steps 3.6-3.9 Epoch times, and converting the new industrial process data into the j old industrial process data by using the trained F, and recording as
Step 3.11: repeating the steps 3.1-3.9 by using each set of old industrial process data, transferring the new industrial process data into the old industrial process domain, obtaining N sets of old industrial process data with new industrial process information by the new industrial process data through anti-transfer learning, and recording as
Step 4: mixing old industrial process data with new industrial process information in the step 3 with corresponding old industrial process data to obtain N groups of mixed data sets;
step 5: dividing a hybrid dataset into hybrid training setsAnd hybrid test dataset->At the same time, N old industrial process training data sets are combined +.>And a new industrial process prediction model y=f (x), training support vector machine SVM models by using N groups of mixed data sets respectively to obtain N old industrial process basic models with new industrial process information, and marking as f 1 (·)-f N (. Cndot.); wherein (1)>k train Is training data set size, +.>k test Is the test data set size, any ith old industrial process, < >>n i Is the ith old industrial process training set size; the method comprises the following specific steps:
step 5.1: initializing parameters;
step 5.2: converting new industrial process data into N groups of old industrial process data carrying new industrial process information through new and old industrial process data migration algorithm based on Cycle GANsMixing D according to equation (14) n→o And D o Obtaining N groups of basic model training data D Basic ；
Step 5.3: by D Basic Training N SVMs to obtain N old industrial process basic models with new industrial process information, denoted as f 1 (·)-f N (·)；
Step 6: mapping the input variables of the new industrial process training set into the operation intervals of the input variables of the similar old industrial process through a model fusion formula (15), and recording the input data of the converted new industrial process training set asFusion output of the N prediction models is obtained through Bayesian model averaging algorithm>
Step 7: fusion output of SVM model of old industrial processAnd new industrial process input data->As input data of a multi-model migration strategy, training a new industrial process model by using a least square support vector machine algorithm to obtain new industrial process model output +.>Completing modeling of a new industrial process;
step 8: model verification, namely evaluating the effectiveness of the SVM model by utilizing root mean square error and a determination coefficient according to a formula (16) and a formula (17), and completing a modeling process if the prediction accuracy of the model obtained in the step 7 on a test data set meets an experimental set threshold value; otherwise, repeating the steps 3 to 7, adding new N groups of old industrial process data samples containing new industrial process information into the mixed samples, and continuing training the new industrial process model until the experimental stop condition is met;
where N is the number of test data, y i Is the output of the predictive model and,is the average value of the predicted output, Y i Is the true output of the new industrial process.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111411517.6A CN114035529B (en) | 2021-11-25 | 2021-11-25 | ATL-BMA-based nonlinear industrial process low-cost modeling method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111411517.6A CN114035529B (en) | 2021-11-25 | 2021-11-25 | ATL-BMA-based nonlinear industrial process low-cost modeling method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114035529A CN114035529A (en) | 2022-02-11 |
| CN114035529B true CN114035529B (en) | 2023-09-08 |
Family
ID=80138766
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111411517.6A Active CN114035529B (en) | 2021-11-25 | 2021-11-25 | ATL-BMA-based nonlinear industrial process low-cost modeling method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114035529B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103500281A (en) * | 2013-10-08 | 2014-01-08 | 广西大学 | Nonlinear system modeling method of process of sugar boiling crystallization |
| CN106156434A (en) * | 2016-07-11 | 2016-11-23 | 江南大学 | Sliding window time difference Gaussian process regression modeling method based on the low and deep structure of local time |
| CN109902378A (en) * | 2019-02-25 | 2019-06-18 | 中国矿业大学 | Complex industrial process low cost modeling method based on multi-model migration and BMA theory |
| CN112750277A (en) * | 2021-01-05 | 2021-05-04 | 武汉大学 | Indoor falling detection system and method fusing track data and sensor posture |
-
2021
- 2021-11-25 CN CN202111411517.6A patent/CN114035529B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103500281A (en) * | 2013-10-08 | 2014-01-08 | 广西大学 | Nonlinear system modeling method of process of sugar boiling crystallization |
| CN106156434A (en) * | 2016-07-11 | 2016-11-23 | 江南大学 | Sliding window time difference Gaussian process regression modeling method based on the low and deep structure of local time |
| CN109902378A (en) * | 2019-02-25 | 2019-06-18 | 中国矿业大学 | Complex industrial process low cost modeling method based on multi-model migration and BMA theory |
| CN112750277A (en) * | 2021-01-05 | 2021-05-04 | 武汉大学 | Indoor falling detection system and method fusing track data and sensor posture |
Non-Patent Citations (1)
| Title |
|---|
| A Minimum-Cost Modeling Method for Nonlinear Industrial Process Based on Multimodel Migration and Bayesian Model Averaging Method;Fei Chu;《IEEE Transactions on Automation Science and Engineering 》;第17卷(第2期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114035529A (en) | 2022-02-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108647272B (en) | Method for predicting concentration of butane at bottom of debutanizer by expanding small samples based on data distribution | |
| CN109635245A (en) | A kind of robust width learning system | |
| CN114169442B (en) | Remote sensing image small sample scene classification method based on double prototype network | |
| CN104504475A (en) | AR*-SVM (support vector machine) hybrid modeling based haze time series prediction method | |
| CN107798383B (en) | Improved positioning method of nuclear extreme learning machine | |
| CN116448419A (en) | Zero sample bearing fault diagnosis method based on depth model high-dimensional parameter multi-target efficient optimization | |
| CN106022954A (en) | Multiple BP neural network load prediction method based on grey correlation degree | |
| CN104462850A (en) | Multi-stage batch process soft measurement method based on fuzzy gauss hybrid model | |
| CN111325264A (en) | Multi-label data classification method based on entropy | |
| CN110046377B (en) | Selective integration instant learning soft measurement modeling method based on heterogeneous similarity | |
| CN104504288A (en) | Method for non-linear multistage intermittent process soft measurement based on multi-directional support vector cluster | |
| CN112884012A (en) | Building energy consumption prediction method based on support vector machine principle | |
| CN111079856A (en) | CSJITL-RVM-based multi-period intermittent process soft measurement modeling method | |
| CN116187835A (en) | Data-driven-based method and system for estimating theoretical line loss interval of transformer area | |
| CN114360652B (en) | Cell strain similarity evaluation method and similar cell strain culture medium formula recommendation method | |
| CN114035529B (en) | ATL-BMA-based nonlinear industrial process low-cost modeling method | |
| CN114239397A (en) | Soft measurement modeling method based on dynamic feature extraction and local weighted deep learning | |
| CN112947080A (en) | Scene parameter transformation-based intelligent decision model performance evaluation system | |
| CN111967187A (en) | Modeling method and device for predicting output data of gas turbine in starting process | |
| Liu et al. | Fuzzy c-mean algorithm based on Mahalanobis distances and better initial values | |
| CN112348194A (en) | Model stability evaluation method based on precision error | |
| Yun et al. | [Retracted] Quality Evaluation and Satisfaction Analysis of Online Learning of College Students Based on Artificial Intelligence | |
| CN112925202B (en) | Fermentation process stage division method based on dynamic feature extraction | |
| CN116776230B (en) | Method and system for identifying signal based on feature imprinting and feature migration | |
| CN110942149B (en) | Feature variable selection method based on information change rate and condition mutual information |
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 |