CN109840362B - Multi-objective optimization-based integrated just-in-time learning industrial process soft measurement modeling method - Google Patents
Multi-objective optimization-based integrated just-in-time learning industrial process soft measurement modeling method Download PDFInfo
- Publication number
- CN109840362B CN109840362B CN201910039438.3A CN201910039438A CN109840362B CN 109840362 B CN109840362 B CN 109840362B CN 201910039438 A CN201910039438 A CN 201910039438A CN 109840362 B CN109840362 B CN 109840362B
- Authority
- CN
- China
- Prior art keywords
- sample
- model
- samples
- learning machine
- extreme learning
- 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
Images
Abstract
The invention discloses an integrated instant learning industrial process soft measurement modeling method based on multi-objective optimization, and belongs to the field of industrial process soft measurement modeling. Aiming at the problems of redundancy and non-linearity of the process commonly existing in industrial data, the method adopts an evolutionary multi-objective optimization method to optimize the input variable and the model structure in the historical sample database, eliminates the variable irrelevant or weakly relevant to the quality variable, improves the quality of the database sample, and effectively balances the relation between the complexity of the model and the prediction precision. In addition, partial samples similar to the query samples are selected from the historical sample library obtained through optimization to construct a local extreme learning machine model, and the Pareto optimal solution obtained through multi-objective optimization is integrated through a selective integration strategy, so that the nonlinear problem of the industrial process can be effectively treated. According to the method, the prediction precision and the calculation efficiency of the industrial process soft measurement modeling are improved by optimizing the modeling data structure and the extreme learning machine model structure.
Description
Technical Field
The invention belongs to the field of industrial process soft measurement modeling, and particularly relates to an integrated instant learning soft measurement modeling method based on multi-objective optimization.
Background
In the modern industrial production process, in order to ensure that the product quality meets increasingly severe production indexes, some key process variables or quality variables need to be detected on line in real time, so that the real-time control of the production process is realized. However, in an actual industrial production process, the sensor performance is very strict because the actual industrial production process is sometimes in a severe environment such as high temperature, high pressure, strong corrosiveness and the like, and the high-performance sensor often has the problems of high cost, difficulty in maintenance and the like, and in addition, the long time of off-line analysis is another important factor restricting the real-time control of the production process. The advent of soft-sensing provides an effective way to monitor such difficult-to-measure parameters on-line.
The core of the soft measurement technology is to construct a functional relation between an auxiliary variable (an easily measured variable) and a dominant variable (a difficultly measured variable) through a certain optimal criterion, and realize online estimation of the dominant variable through computer software. Due to the development of technologies such as distributed control systems and databases, data-driven soft measurement modeling technologies have received much attention. In data-driven modeling, a local model represented by instant learning adopts the concept of 'divide-and-conquer', so that the local process characteristics can be accurately described, the computational complexity of the model is remarkably reduced, the characteristics of nonlinearity, time-varying property and the like of the process are effectively processed, and the method has great attention in the field of soft measurement modeling.
However, the learning-on-demand technique often relies on samples in the database, and the quality of the samples in the database plays a crucial role in the model prediction accuracy. In the actual production process, due to factors such as sensor redundancy and inconsistent sampling frequency, the input variables in the database have the defects of redundancy, weak correlation between the input variables and the output variables, and the like, so that the complexity of model calculation is increased, and problems such as overfitting are easy to occur. Therefore, reasonably improving the quality of the historical sample database is crucial to improving the computational efficiency and prediction accuracy of the soft measurement model. In addition, because a positive correlation often exists between the prediction accuracy of the model and the complexity of the model, how to reduce the complexity of the model as much as possible under the condition of ensuring the prediction accuracy of the model to be constant is another problem to be solved by the invention. Meanwhile, considering that an industrial process often presents complex process characteristics such as strong nonlinearity, time-varying property and the like, only a series of suboptimal models can be obtained by adopting an instant learning soft measurement modeling technology, and the suboptimal models are difficult to fully consider different characteristics of the process.
Disclosure of Invention
The invention aims to solve the technical problems of how to improve the quality of a historical sample database in the soft measurement modeling technology and how to reduce the complexity of a model while improving the prediction precision of the model. For the purpose, the invention provides an integrated instant learning industrial process soft measurement modeling method based on multi-objective optimization, which comprises the following steps:
(1) collecting industrial process data D through a distributed control system or an off-line detection method, constructing a database for soft measurement modeling, and determining an auxiliary variable X related to a predicted variable y through mechanism analysis of the industrial process, wherein the auxiliary variable X is an input variable, and X is { X ═ X { (X })1,x2,...,xM}；
(2) All samples in the database are normalized and divided into a training set DtrainVerification set DvalidateAnd test set DtestWherein the training set DtrainTraining, validation set D for modelsvalidateOptimization for model parameters, test set DtestFor evaluation of model performance;
(3) training set D by adopting evolutionary multi-objective optimization methodtrainOptimizing input variables and the number of nodes of a hidden layer of the extreme learning machine model, eliminating redundant or input variables weakly related to output variables, obtaining S Pareto optimal solutions according to optimization, and sequentially updating a training set D according to the Pareto optimal solutions trainObtaining S new training sets and using the S new training sets as a new modeling sample database;
the multi-objective optimization problem is described as follows:
min[f1(x),f2(x)]
f1(x) And f2(x) For two objective functions to be optimized
f2(x)＝Nhidden×M*
f1(x) To verify the prediction error of the sample, f2(x) Is the complexity of the model, NvalTo verify the set DvalidateThe number of samples of (a) to (b),
to verify the set DvalidatePredicted value of the ith sample, yval,iTo verify the set DvalidateTrue value of the ith sample, M*In order to optimize the number of input variables,
the specific calculation process of (a) is to verify the set DvalidateThe samples in (1) are taken in turn as query samples xq,iAccording to M*Updating the input variables of the query sample, and selecting the training set D according to the Euclidean distance similaritytrainThe first P similar samples with the query sample are selected to construct an extreme learning machine model, and the predicted output is obtained
x is the decision variable to be optimized, x ═ x1,x2,...,xm...,xM,NhiddenIn which x ismM represents the M-th input variable of the sample, M represents the number of input variables, NhiddenHiding the number of nodes of a layer for the extreme learning machine model; lb and ub are respectively the lower limit and upper limit of x, A and b are inequality constraint terms of x;
(4) updating the test set D according to S new modeling sample databasestestThe input variables are corresponding to obtain S test sample sets, samples in each test sample set are sequentially taken as query samples, P similar samples are selected from a corresponding new modeling sample database according to the Euclidean distance similarity to construct a local extreme learning machine model, the S local extreme learning machine models are obtained, and the predicted output of the test sample sets is obtained 
According to S new modeling sample databases, sequentially comparing the test set DtestThe input variables in the process are updated, irrelevant or redundant variables are removed, and the obtained variables are used as a new input variable set Xtest,newObtaining S test sample sets, sequentially taking samples in each test sample set as query samples, selecting the previous P similar samples from a corresponding new modeling sample database according to the Euclidean distance similarity to construct a local extreme learning machine model, obtaining S local extreme learning machine models, and obtaining the predicted output of the test sample sets
Representing the S-th prediction output.
(5) Updating the verification set D according to S new modeling sample databasesvalidateThe input variables are correspondingly obtained S verification sample sets, samples in each verification sample set are sequentially taken as query samples, P similar samples are selected from a corresponding new modeling sample database according to the Euclidean distance similarity to construct a local extreme learning machine model, and the S local extreme learning machine models are obtained and integrated; updating the verification set D according to S new modeling sample databasesvalidateThe input variables are correspondingly obtained S verification sample sets, samples in each verification sample set are sequentially taken as query samples, the previous P similar samples are selected from a corresponding new modeling sample database according to the Euclidean distance similarity to construct a local extreme learning machine model, and then the S local extreme learning machine models { ELM (extreme learning machine model) } are obtained 1,ELM2,...,ELMSGet predicted outputs of S validation samples, respectively
Output prediction of validation samples
And (3) as an input, and the real output is used as an output, a PLS model is constructed, and regression coefficients of the PLS model are stored.
(6) And (4) pruning the S local extreme learning machine models obtained in the step (5), and selecting a sub-model with higher prediction precision to construct a final integrated model for the test sample set in the step (4).
(a) The absolute values of the PLS model regression coefficients of the verification samples are arranged in a descending order according to the magnitude to obtain { | [ beta ]1|,|β2|...,|βS|}；
(b) Selecting front S with better prediction effect by calculating the contribution ratio CP of the local extreme learning machine model*The sub-models construct an integrated model, and the calculation formula of the contribution rate CP is as follows:
(c) selecting S from the S local extreme learning machine models in the step (4) according to the pruned verification sample sub-model index*A local extreme learning machine model performs PLS integration into an integrated model.
The invention has the advantages of
According to the method, the input variables and the number of nodes of the hidden layer of the extreme learning machine model are optimized through a multi-objective evolutionary optimization method, samples with high similarity to the query sample are selected according to the similarity of Euclidean distances to construct the extreme learning machine model, submodels obtained by different Pareto solutions are integrated through an integrated learning method, and therefore the high-performance integrated instant learning soft measurement model is obtained. Compared with other methods at present, the method can well process the redundant variables or the input variables weakly related to the output variables in the database, reduce the complexity of the model while ensuring the prediction precision of the model, effectively process the nonlinearity of the industrial process, and improve the calculation efficiency of modeling and the prediction precision of the model.
Drawings
FIG. 1 is a flow chart of the integrated just-in-time learning soft measurement modeling based on multi-objective optimization in the method of the present invention.
Detailed Description
The invention is further described with reference to the following figures and specific examples.
Example 1: as shown in FIG. 1, the integrated just-in-time learning industrial process soft measurement modeling method based on multi-objective optimization comprises the following steps: industrial process data are collected through a distributed control system or an off-line detection method, and a database for soft measurement modeling is constructed. Auxiliary variables related to the predictive variables are determined by mechanistic analysis of the industrial process.
The second step is that: the samples in the database are normalized and divided into training sets (D)train∈RJ×Q) Verification set (D)validate∈RK×Q) And test set (D)test∈RT×Q). The training set is used for training the model, the verification set is used for optimizing the model parameters, and the test set is used for evaluating the model performance.
The third step: optimizing the input variables in the training set by adopting a multi-objective evolutionary optimization method (NSGA-II algorithm),
the multi-objective optimization problem is described as follows:
min[f1(x),f2(x)]
f1(x) And f2(x) For two objective functions to be optimized, since the objective function f1(x) And f2(x) Contradictory, i.e., one solution is best at one goal and possibly worst at another, so the objective of the multi-objective optimization algorithm is to find a set of homogeneous solutions The balance is achieved, so that each sub-target is optimized as much as possible. Generally speaking, the multi-objective optimization problem does not have only one optimal solution, but obtains a group of optimal solution sets consisting of a plurality of Pareto optimal solutions.
f2(x)＝Nhidden×M*
f1(x) To verify the prediction error of the sample, f2(x) Is the complexity of the model, NvalTo verify the set DvalidateThe number of samples of (a) to (b),
to verify the set DvalidatePredicted value of the ith sample, yval,iTo verify the set DvalidateTrue value of the ith sample, M*In order to optimize the number of input variables,
the specific calculation process of (a) is to verify the set DvalidateThe samples in (1) are taken in turn as query samples xq,iAccording to M*Updating the input variables of the query sample, and selecting the training set D according to the Euclidean distance similaritytrainThe first P similar samples with the query sample are selected to construct an extreme learning machine model, and the predicted output is obtained
x is the decision variable to be optimized, x ═ x1,x2,...,xm...,xM,NhiddenIn which xmM represents the M-th input variable of the sample, M represents the number of input variables, NhiddenHiding the number of nodes of a layer for the extreme learning machine model; lb and ub are respectively lower limit and upper limit of x, A and b are inequality constraint terms of x;
the fourth step: updating the test set according to the S new modeling sample databasesDtestThe input variables are corresponding to obtain S test sample sets, samples in each test sample set are sequentially taken as query samples, P similar samples are selected from a corresponding new modeling sample database according to the Euclidean distance similarity to construct a local extreme learning machine model, the S local extreme learning machine models are obtained, and the predicted output of the test sample sets is obtained 
According to S new modeling sample databases, sequentially comparing the test set DtestThe input variables in the process are updated, irrelevant or redundant variables are removed, and the obtained variables are used as a new input variable set Xtest,newS test sample sets are obtained, samples in each test sample set are sequentially taken as query samples, the previous P similar samples are selected from a corresponding new modeling sample database according to the Euclidean distance similarity to construct a local extreme learning machine model, the S local extreme learning machine models are obtained, and the predicted output of the test sample sets is obtained
Representing the S-th prediction output.
The fifth step: updating the verification set D according to S new modeling sample databasesvalidateThe input variables are correspondingly obtained S verification sample sets, samples in each verification sample set are sequentially taken as query samples, P similar samples are selected from a corresponding new modeling sample database according to the Euclidean distance similarity to construct a local extreme learning machine model, and the S local extreme learning machine models are obtained and integrated; updating the verification set D according to S new modeling sample databasesvalidateObtaining S verification sample sets corresponding to the input variables, sequentially taking samples in each verification sample set as query samples according to Europe Selecting the front P similar samples from the corresponding new modeling sample database according to the similarity of the distance between the two adjacent local extreme learning machines to construct a local extreme learning machine model, and obtaining S local extreme learning machine models { ELM1,ELM2,...,ELMSGet predicted outputs of S validation samples, respectively
Output prediction of validation samples
And (3) as an input, and the real output is used as an output, a PLS model is constructed, and regression coefficients of the PLS model are stored.
And a sixth step: and (4) pruning the S local extreme learning machine models obtained in the step (5), and selecting a sub-model with higher prediction precision to construct a final integrated model for the test sample set in the step (4). The method specifically comprises the following steps:
(a) the absolute values of the PLS model regression coefficients of the verification samples are arranged in a descending order according to the magnitude to obtain { | [ beta ]1|,|β2|...,|βS|}；
(b) Selecting front S with better prediction effect by calculating the contribution ratio CP of the local extreme learning machine model*The sub-models construct an integrated model, and the calculation formula of the contribution rate CP is as follows:
wherein, | betaSI represents the absolute value of the s-th regression coefficient of the PLS model, and satisfies | beta1|≥|β2|≥…≥|βSIf the CP is more than or equal to 95 percent, stopping model pruning, and storing the indexes of the sub-models and the corresponding regression coefficients;
(c) selecting S from the S local extreme learning machine models in the step (4) according to the pruned verification sample sub-model index *A local extreme learning machine model performs PLS integration into an integrated model.
Similarity of Euclidean distance in the above steps
Wherein d isnRepresenting the weighted euclidean distance between the query sample and the similar samples,
xnrepresenting similar samples, xqRepresents the query sample, n ∈ 1, …, P), q ∈ 1, …, P), σ ∈ 1, …nIs that
The standard deviation of (a) is determined,
is a localization parameter.
Example 2: as will be further described in connection with the debutanizer industrial process, the debutanizer is a part of the unit for desulfurization and naphtha separation in an industrial refinery process with the goal of minimizing the concentration of butane in the bottoms. However, the butane concentration is difficult to realize real-time online detection at present. The soft measurement method is adopted to predict the butane concentration on line, so that the desulfurization efficiency of the debutanizer can be effectively improved. According to the mechanism analysis, x is1The overhead temperature; x is the number of2Overhead pressure; x is the number of3Overhead reflux amount; x is the number of4The product outflow at the top of the column; x is the number of5The temperature of the tower plate of the sixth layer; x is the number of6The tower low temperature is 1; x is the number of7The 7 monitored variables of the tower bottom temperature 2 are used as auxiliary variables for constructing a soft measurement model, and the output variable is the butane concentration. The effectiveness of the method is verified by constructing a dynamic process soft measurement model, the dynamic soft measurement model is constructed by adopting a moving average model structure, 2388 groups of sample data are obtained totally, the input variables are 49, 1194 samples are used as training samples, 597 samples are used as verification samples, and 597 samples are used as test samples.
The following detailed description of the implementation steps is made in conjunction with this specific process:
1. collecting industrial process data of the debutanizer, preprocessing the data, and removing abnormal values and missing values.
2. The 2388 sets of data were normalized. 1194 samples of the samples form a training sample set, 597 samples form a verification sample set, and 597 samples form a test sample set.
3. Input variables in the training sample set are optimized offline. And coding 49 input variables and the number of nodes of a hidden layer of the extreme learning machine model to obtain decision variables for optimization, wherein the number of the decision variables is 50, and optimizing the decision variables by adopting an NSGA-II algorithm.
4. And updating the training sample set and taking the training sample set as a historical sample database for soft measurement modeling, updating the input variables in the verification sample set and the test sample set according to the optimized input variable index, and removing redundant variables.
5. And for the verification sample, calculating the similarity between the verification sample and the historical sample in the database according to the Euclidean distance similarity, and selecting 30 similar samples to construct an extreme learning machine soft measurement model to obtain a local prediction value of the butane concentration. And (3) taking the local predicted values as input variables, verifying the real output of the sample as an output value, constructing a PLS model, and obtaining a regression coefficient of the model. And then, the model regression coefficients are arranged in a descending order according to the magnitude, the contribution ratio CP of the submodel is continuously calculated in a superposition mode, when the CP is more than or equal to 95%, the calculation is stopped, and the index of the current submodel and the corresponding regression coefficient are stored.
6. And for the test sample, calculating the similarity between the test sample and the historical sample in the database according to the Euclidean distance similarity, and selecting 30 similar samples to construct an extreme learning machine soft measurement model so as to obtain a local prediction value of the butane concentration.
7. And selecting the sub-model corresponding to the test sample according to the sub-model index and the model regression coefficient obtained by verifying the sample, and performing PLS integration on the sub-model to obtain the butane concentration predicted value of the test sample.
8. And (4) comparing the prediction accuracy of the butane concentration by different soft measurement models. And comparing the prediction errors of the butane concentration under 2 conditions, namely comparing the traditional instantaneous learning extreme learning machine soft measurement model and the integrated instantaneous learning soft measurement modeling method for optimizing the input variables in the historical sample database and the hidden layer node number of the extreme learning machine. The prediction error results of the different methods are shown in table 1, wherein smaller prediction errors indicate higher prediction accuracy. As can be seen from the table 1, the integrated just-in-time learning industrial process soft measurement modeling method based on multi-objective optimization improves the prediction accuracy of the soft measurement model.
TABLE 1 Root Mean Square Error (RMSE) in debutanizer for different processes
| Method | RMSE |
| Instant learning extreme learning machine soft measurement modeling | 0.0844 |
| Integrated just-in-time learning industrial process soft measurement modeling based on multi-objective optimization | 0.0500 |
The above-described embodiments are intended to illustrate rather than limit the invention, and any modifications and variations of the present invention are within the spirit and scope of the appended claims.
Claims (4)
1. A multi-objective optimization-based integrated instant learning industrial process soft measurement modeling method is characterized by comprising the following steps:
(1) collecting industrial process data D, constructing a database for soft measurement modeling, and determining an auxiliary variable X related to a predicted variable y through mechanism analysis of the industrial process, wherein the auxiliary variable X is an input variable, and X is { X ═ X }1,x2,...,xM}；
(2) All samples in the database are normalized and divided into a training set DtrainVerification set DvalidateAnd test set DtestWherein the training set DtrainTraining, validation set D for modelsvalidateOptimization for model parameters, test set DtestFor evaluation of model performance;
(3) training set D by adopting evolutionary multi-objective optimization methodtrainOptimizing input variables and the number of nodes of a hidden layer of the extreme learning machine model, eliminating redundant or input variables weakly related to output variables, obtaining S Pareto optimal solutions according to optimization, and sequentially updating a training set D according to the Pareto optimal solutions trainObtaining S new training sets and using the S new training sets as a new modeling sample database;
the multi-objective optimization problem is described as follows:
min[f1(x),f2(x)]
f1(x) And f2(x) For two objective functions to be optimized
f2(x)＝Nhidden×M*
f1(x) To verify the prediction error of the sample, f2(x) Is the complexity of the model, NvalTo verify the set DvalidateThe number of samples of (a) to (b),
to verify the set DvalidatePredicted value of the ith sample, yval,iTo verify the set DvalidateTrue value of the ith sample, M*In order to optimize the number of input variables,
the specific calculation process of (a) is to verify the set DvalidateThe samples in (1) are taken in turn as query samples xq,iAccording to M*Updating the input variables of the query sample, and selecting the training set D according to the Euclidean distance similaritytrainThe first P similar samples with the query sample are selected to construct an extreme learning machine model, and the predicted output is obtained
x is the decision variable to be optimized, x ═ x1,x2,...,xm...,xM,NhiddenIn which xmM represents the M-th input variable of the sample, M represents the number of input variables, NhiddenHiding the number of nodes of a layer for the extreme learning machine model; lb and ub are respectively the lower limit and upper limit of x, A and b are inequality constraint terms of x;
(4) updating the test set D according to S new modeling sample databasestestThe input variables are corresponding to obtain S test sample sets, samples in each test sample set are sequentially taken as query samples, P similar samples are selected from a corresponding new modeling sample database according to the Euclidean distance similarity to construct a local extreme learning machine model, the S local extreme learning machine models are obtained, and the predicted output of the test sample sets is obtained 
(5) Updating the verification set D according to S new modeling sample databasesvalidateThe input variables are correspondingly obtained S verification sample sets, samples in each verification sample set are sequentially taken as query samples, P similar samples are selected from a corresponding new modeling sample database according to the Euclidean distance similarity to construct a local extreme learning machine model, and the S local extreme learning machine models are obtained and integrated;
(6) pruning the S local extreme learning machine models obtained in the step (5), and selecting a sub-model with higher prediction precision to construct a final integrated model for the test sample set in the step (4);
the specific process of the step (6) is as follows:
(a) the absolute values of the PLS model regression coefficients of the verification samples are arranged in a descending order according to the magnitude to obtain { | [ beta ]1|,|β2|...,|βS|}；
(b) Selecting front S with better prediction effect by calculating the contribution ratio CP of the local extreme learning machine model*The sub-models construct an integrated model, and the calculation formula of the contribution rate CP is as follows:
wherein, | betaSI represents the absolute value of the s-th regression coefficient of the PLS model, and satisfies | beta1|≥|β2|≥…≥|βSIf the CP is more than or equal to 95 percent, stopping model pruning, and storing the indexes of the sub-models and the corresponding regression coefficients;
(c) selecting S from the S local extreme learning machine models in the step (4) according to the pruned verification sample sub-model index *And carrying out PLS integration on the local extreme learning machine models to obtain an integrated model.
2. The integrated just-in-time learning industrial process soft measurement modeling method based on multi-objective optimization as claimed in claim 1, wherein the step (4) is implemented by the following specific processes:
according to S new modeling sample databases, sequentially comparing the test set DtestThe input variables in the process are updated, irrelevant or redundant variables are removed, and the obtained variables are used as a new input variable set Xtest,newS test sample sets are obtained, samples in each test sample set are sequentially taken as query samples, the previous P similar samples are selected from a corresponding new modeling sample database according to the Euclidean distance similarity to construct a local extreme learning machine model, the S local extreme learning machine models are obtained, and the test sample sets are obtainedPredicted output of (2)
Representing the S-th prediction output.
3. The integrated just-in-time learning industrial process soft measurement modeling method based on multi-objective optimization as claimed in claim 1, wherein the specific process of the step (5) is as follows:
updating the verification set D according to S new modeling sample databasesvalidateThe input variables are correspondingly obtained S verification sample sets, samples in each verification sample set are sequentially taken as query samples, P previous similar samples are selected from a corresponding new modeling sample database according to the Euclidean distance similarity to construct a local extreme learning machine model, and then S local extreme learning machine models { ELM (empirical mode model) are obtained 1,ELM2,...,ELMSGet the predicted output of S validation samples accordingly
The prediction of the verification sample is output
And (3) as an input, and the real output is used as an output, a PLS model is constructed, and regression coefficients of the PLS model are stored.
4. The integrated just-in-time learning industrial process soft measurement modeling method based on multi-objective optimization according to any one of claims 1 or 2, characterized in that the Euclidean distance similarity
Wherein d isnRepresenting the weighted euclidean distance between the query sample and the similar samples,
xnrepresenting similar samples, xqRepresents the query sample, n ∈ 1, …, P, q ∈ 1, …, P, σnIs that
The standard deviation of (a) is determined,
is a localization parameter.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910039438.3A CN109840362B (en) | 2019-01-16 | 2019-01-16 | Multi-objective optimization-based integrated just-in-time learning industrial process soft measurement modeling method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910039438.3A CN109840362B (en) | 2019-01-16 | 2019-01-16 | Multi-objective optimization-based integrated just-in-time learning industrial process soft measurement modeling method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109840362A CN109840362A (en) | 2019-06-04 |
| CN109840362B true CN109840362B (en) | 2022-06-14 |
Family
ID=66883886
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910039438.3A Active CN109840362B (en) | 2019-01-16 | 2019-01-16 | Multi-objective optimization-based integrated just-in-time learning industrial process soft measurement modeling method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109840362B (en) |
Families Citing this family (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112446520A (en) * | 2019-08-30 | 2021-03-05 | 京东数字科技控股有限公司 | Method and device for predicting product scheduling time |
| CN111914466B (en) * | 2019-09-07 | 2023-10-24 | 宁波大学 | Chemical process monitoring method based on related variable distributed modeling |
| CN110751987A (en) * | 2019-10-17 | 2020-02-04 | 北京化工大学 | Method for establishing soft measurement model of monitoring variable for ethylene preparation through naphtha cracking |
| CN111914492B (en) * | 2020-04-28 | 2022-09-13 | 昆明理工大学 | Evolution optimization-based semi-supervised learning industrial process soft measurement modeling method |
| CN111768000A (en) * | 2020-06-23 | 2020-10-13 | 中南大学 | Industrial process data modeling method for online adaptive fine-tuning deep learning |
| CN112016241B (en) * | 2020-07-23 | 2023-09-29 | 武汉数字化设计与制造创新中心有限公司 | Soft measurement modeling method based on mLASSO-MLP model |
| CN112596391B (en) * | 2020-12-24 | 2022-08-02 | 中国兵器装备集团自动化研究所 | Deep neural network large time lag system dynamic modeling method based on data driving |
| CN112989635B (en) * | 2021-04-22 | 2022-05-06 | 昆明理工大学 | Integrated learning soft measurement modeling method based on self-encoder diversity generation mechanism |
| CN113377072B (en) * | 2021-06-24 | 2023-02-07 | 华东交通大学 | Rare earth extraction process dosage optimization setting method based on instant learning |
| CN114117919B (en) * | 2021-11-29 | 2022-08-23 | 中国石油大学(华东) | Instant learning soft measurement modeling method based on sample collaborative representation |
| CN114626304B (en) * | 2022-03-21 | 2024-02-27 | 齐鲁工业大学 | Online prediction soft measurement modeling method for ore pulp copper grade |
| CN115094193A (en) * | 2022-06-27 | 2022-09-23 | 中冶华天南京工程技术有限公司 | Intelligent molten iron pretreatment desulfurization system based on data mining |
| CN116738866B (en) * | 2023-08-11 | 2023-10-27 | 中国石油大学(华东) | Instant learning soft measurement modeling method based on time sequence feature extraction |
| CN116821695B (en) * | 2023-08-30 | 2023-11-03 | 中国石油大学(华东) | Semi-supervised neural network soft measurement modeling method |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106682312A (en) * | 2016-12-28 | 2017-05-17 | 浙江大学 | Industrial process soft-measurement modeling method of local weighing extreme learning machine model |
| CN106951995A (en) * | 2017-03-22 | 2017-07-14 | 华东交通大学 | A kind of EHV transmission electric field extreme learning machine predicts multiple-objection optimization screen method |
| CN107274387A (en) * | 2017-05-19 | 2017-10-20 | 西安电子科技大学 | The end member extraction method of target in hyperspectral remotely sensed image based on Evolutionary multiobjective optimization |
| CN108133286A (en) * | 2017-12-21 | 2018-06-08 | 河海大学 | A kind of underground water multi-objective calculation method based on surface subsidence alternative model |
| US10061300B1 (en) * | 2017-09-29 | 2018-08-28 | Xometry, Inc. | Methods and apparatus for machine learning predictions and multi-objective optimization of manufacturing processes |
| CN108549234A (en) * | 2018-05-11 | 2018-09-18 | 江南大学 | A kind of multiobjective optimization control method based on dynamic variate |
-
2019
- 2019-01-16 CN CN201910039438.3A patent/CN109840362B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106682312A (en) * | 2016-12-28 | 2017-05-17 | 浙江大学 | Industrial process soft-measurement modeling method of local weighing extreme learning machine model |
| CN106951995A (en) * | 2017-03-22 | 2017-07-14 | 华东交通大学 | A kind of EHV transmission electric field extreme learning machine predicts multiple-objection optimization screen method |
| CN107274387A (en) * | 2017-05-19 | 2017-10-20 | 西安电子科技大学 | The end member extraction method of target in hyperspectral remotely sensed image based on Evolutionary multiobjective optimization |
| US10061300B1 (en) * | 2017-09-29 | 2018-08-28 | Xometry, Inc. | Methods and apparatus for machine learning predictions and multi-objective optimization of manufacturing processes |
| CN108133286A (en) * | 2017-12-21 | 2018-06-08 | 河海大学 | A kind of underground water multi-objective calculation method based on surface subsidence alternative model |
| CN108549234A (en) * | 2018-05-11 | 2018-09-18 | 江南大学 | A kind of multiobjective optimization control method based on dynamic variate |
Non-Patent Citations (4)
| Title |
|---|
| Huaiping Jin 等."Ensemble just-in-time learning framework through evolutionary multi-objective optimization for soft sensor development of nonlinear industrial processes".《Chemometrics and Intelligent Laboratory Systems》.2019,第184卷 * |
| Mingyang Zhang 等."Hyperspectral band selection based on multi-objective optimization with high information and low redundancy".《Applied Soft Computing》.2018,第70卷 * |
| 王礼云 等."极限学习机多目标模型选择研究".《计算机仿真》.2014,第31卷(第8期), * |
| 赵佳琦."进化多目标优化学习算法及其应用".《中国优秀博硕士学位论文全文数据库(博士)信息科技辑》.2019, * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109840362A (en) | 2019-06-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109840362B (en) | Multi-objective optimization-based integrated just-in-time learning industrial process soft measurement modeling method | |
| CN111222549B (en) | Unmanned aerial vehicle fault prediction method based on deep neural network | |
| CN105117602B (en) | A kind of metering device running status method for early warning | |
| CN101899563B (en) | PCA (Principle Component Analysis) model based furnace temperature and tension monitoring and fault tracing method of continuous annealing unit | |
| EP3289420A1 (en) | Computer system and method for causality analysis using hybrid first-principles and inferential model | |
| TW201615844A (en) | Method and system of cause analysis and correction for manufacturing data | |
| CN103927412A (en) | Real-time learning debutanizer soft measurement modeling method on basis of Gaussian mixture models | |
| CN112284440B (en) | Sensor data deviation self-adaptive correction method | |
| JP2009516303A (en) | Apparatus and method for identifying useful data in a process control environment | |
| Pani et al. | A survey of data treatment techniques for soft sensor design | |
| CN105468850A (en) | Multi-residual error regression prediction algorithm based electronic product degradation trend prediction method | |
| TWI738016B (en) | Intelligent learning mechanism and adaptive regulation system for liquid gas filling | |
| CN112904810B (en) | Process industry nonlinear process monitoring method based on effective feature selection | |
| CN113723010A (en) | Bridge damage early warning method based on LSTM temperature-displacement correlation model | |
| CN105868164A (en) | Soft measurement modeling method based on monitored linear dynamic system model | |
| CN109376892A (en) | A kind of equipment state prediction method based on life cycle phase locating for equipment | |
| CN116859843A (en) | Production equipment fault monitoring method and system based on industrial big data | |
| Wang et al. | A novel sliding window PCA-IPF based steady-state detection framework and its industrial application | |
| CN116380445A (en) | Equipment state diagnosis method and related device based on vibration waveform | |
| CN112418682A (en) | Security assessment method fusing multi-source information | |
| CN110737975B (en) | Wind power plant wind speed and power prediction and abnormality correction method based on empirical mode decomposition and autoregressive model | |
| CN115238573A (en) | Hydroelectric generating set performance degradation trend prediction method and system considering working condition parameters | |
| CN116662925A (en) | Industrial process soft measurement method based on weighted sparse neural network | |
| CN112257893A (en) | Complex electromechanical system health state prediction method considering monitoring error | |
| CN111898673A (en) | Dissolved oxygen content prediction method based on EMD and LSTM |
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 |