CN114036832A - Batch process modeling and final product quality prediction method based on big data - Google Patents
Batch process modeling and final product quality prediction method based on big data Download PDFInfo
- Publication number
- CN114036832A CN114036832A CN202111302383.4A CN202111302383A CN114036832A CN 114036832 A CN114036832 A CN 114036832A CN 202111302383 A CN202111302383 A CN 202111302383A CN 114036832 A CN114036832 A CN 114036832A
- Authority
- CN
- China
- Prior art keywords
- final product
- product quality
- variable
- batch process
- state variable
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 52
- 239000012467 final product Substances 0.000 title claims abstract description 42
- 238000010923 batch production Methods 0.000 title claims abstract description 40
- 238000013507 mapping Methods 0.000 claims abstract description 33
- 238000005259 measurement Methods 0.000 claims abstract description 21
- 230000008569 process Effects 0.000 claims abstract description 19
- 238000013528 artificial neural network Methods 0.000 claims abstract description 18
- 230000007246 mechanism Effects 0.000 claims abstract description 10
- 239000013598 vector Substances 0.000 claims description 29
- 238000012549 training Methods 0.000 claims description 15
- 238000005070 sampling Methods 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 6
- 230000000306 recurrent effect Effects 0.000 claims description 6
- 238000012795 verification Methods 0.000 claims description 5
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 239000007795 chemical reaction product Substances 0.000 claims description 2
- 238000003756 stirring Methods 0.000 claims description 2
- 125000004122 cyclic group Chemical group 0.000 abstract description 7
- 239000000047 product Substances 0.000 description 9
- 239000011159 matrix material Substances 0.000 description 5
- 239000000126 substance Substances 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 3
- 239000012847 fine chemical Substances 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000011217 control strategy Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000010438 heat treatment Methods 0.000 description 2
- 238000003908 quality control method Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000003889 chemical engineering Methods 0.000 description 1
- 238000001816 cooling Methods 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000009533 lab test Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005312 nonlinear dynamic Methods 0.000 description 1
- 238000000513 principal component analysis Methods 0.000 description 1
- 238000004886 process control Methods 0.000 description 1
- 238000011112 process operation Methods 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06395—Quality analysis or management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
-
- 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
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Evolutionary Computation (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- General Engineering & Computer Science (AREA)
- Marketing (AREA)
- Health & Medical Sciences (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Quality & Reliability (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Educational Administration (AREA)
- Operations Research (AREA)
- Life Sciences & Earth Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computer Hardware Design (AREA)
- Medical Informatics (AREA)
- Geometry (AREA)
- Manufacturing & Machinery (AREA)
- Primary Health Care (AREA)
- Feedback Control In General (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a batch process modeling and final product quality prediction method based on big data, which belongs to the technical field of batch process modeling and final product quality prediction in fine chemistry and pharmacy, and specifically comprises the steps of using an encoder to nonlinearly map a process measurement variable and a control input variable into a higher-dimensional state variable, and using a cyclic neural network to construct a state variable dynamic model; mapping the state variable back to the measurement variable through a decoder to form a nonlinear mapping parameter; learning nonlinear mapping parameters and dynamic models by using batch process historical operation big data; and establishing a final product quality prediction model in the batch process by combining an attention mechanism. The invention avoids modeling and learning in stages, the final product quality prediction model explicitly considers the influence of the state variable at the historical moment on the final product quality, and the problem of the quality performance reduction of the final product predicted by the cyclic neural network along with the increase of the input length is solved.
Description
Technical Field
The invention belongs to the technical field of batch process modeling and final product quality prediction in fine chemistry and pharmacy, and particularly relates to a batch process modeling and final product quality prediction method based on big data.
Background
The existing process modeling, control and final product quality prediction methods based on batch operation data facing the fine chemical and pharmaceutical fields are roughly classified into the following three types:
the first type is to build a batch process model based on the first principle, and identify model parameters using batch run data. The model established by the method has strong interpretability, but some parameters and equations of the established model are estimated according to the operation result of a laboratory small-batch experiment, the actual large-scale batch process has a lot of time-varying and nonlinear factors, the identified model is not accurate enough, and some batch processes use open-loop control, so the control precision is not high, the quality prediction precision of the final product is low, and the quality result is often obtained by testing after the batch process is finished.
The second type is a method based on statistical batch control, which obtains latent variables from batch operation data by methods such as principal component analysis, partial least squares and typical correlation analysis learning, or directly learns a linear state space model by a subspace identification method. The method learns the linear correlation between batch operation data and the quality of the final product, and predicts the quality of the final product by using the data of past batches. The method has the advantages that a system dynamic linear model is established, the controller is convenient to design, and the dependence of the control algorithm on the model accuracy can be reduced by using the iterative learning control algorithm; and the product quality control can be integrated into a dynamic optimization problem by combining with model predictive control. However, the model established by the method is a time-invariant linear dynamic model and is a linear approximation of a nonlinear system, and the quality of products fluctuates greatly in the fine chemical engineering and pharmaceutical processes due to the uncertainty of raw material components, various product types, complex process and the like, so that the control accuracy of the statistical batch control method is still to be improved.
And the third type is that a nonlinear model is established between process variable data and product quality based on nonlinear modeling methods such as a feedforward neural network, a cyclic neural network and a support vector machine, and the methods have good prediction capability and can provide an empirical model for long-time prediction. The method utilizes the prediction state of the last time point and the quality of the final product to establish a nonlinear mapping model, the prediction result is very dependent on the state prediction quality of the last time point, and the dynamic influence of the process operation at the same batch of historical time on the quality of the final product is not considered. Meanwhile, the method still does not solve the problem of product quality control in batch process, namely, the operation given track is difficult to set for obtaining ideal final product quality, and even if the nonlinear models are used for designing a control algorithm to enable the reactor to track the operation track at the given temperature, the ideal product quality cannot be achieved necessarily.
Therefore, the three methods have certain defects in the quality prediction of the final product.
Disclosure of Invention
In order to overcome the defects of the existing batch process modeling and final product quality prediction of the fine chemical industry and the pharmacy based on big data, the invention provides a batch process modeling and final product quality prediction method based on big data.
In order to achieve the purpose, the invention adopts the technical scheme that: the method for batch process modeling and final product quality prediction based on big data comprises the following steps:
s1: nonlinearly mapping a process measurement variable and a control input variable into a higher-dimensional state variable by using an encoder, and constructing a state variable dynamic model by using a recurrent neural network;
s2: mapping the state variable back to the measurement variable through a decoder to form a nonlinear mapping parameter;
s3: learning nonlinear mapping parameters and dynamic models by using batch process historical operation big data;
s4: and establishing a final product quality prediction model in the batch process by using the state variable obtained by learning inference and combining an attention mechanism.
In one possible implementation, in steps S1 and S2, an encoder f is usedcAnd feAnd nonlinearly mapping the process measurement variable x (t) and the control input variable u (t) at the time t into a higher-dimensional state variable, and constructing a state variable dynamic model by using a Recurrent Neural Network (RNN):
the state variable h (t) at the time t is processed by a decoder fdNonlinear mapping back to the measured variable x (t), forming a nonlinear mapping parameter:
x(t)=fd(h(t));
in step S4, a batch process end product quality index variable q (t) is established in conjunction with Attention mechanism Attentionf) And (3) a prediction model, namely fitting the quality prediction model by using the historical batch quality analysis result:
q(tf)=Attention(h1,…,ht,…hf)。
in one possible implementation manner, in step S1, the measurement vectors of the sampling batches and the sampling times are selected to constitute the process measurement variable, the control vectors of the sampling batches and the sampling times are selected to constitute the control input variable, and the measurement vectors establish the training database by using the duration running data of the PI control or the open loop control.
In one possible implementation mode, the measurement vector comprises temperature, pressure, density and a viscosity value obtained by conversion of the torque of a stirring motor of the reaction kettle; the control vectors include steam jacket pressure and vent rate.
In one possible implementation, in step S3, the sequence data of multiple batches is slid step by step, and a training set of sequence data building blocks with a specific window length is taken; and dividing the sequence data sets into a training set and a verification set, optimizing the minimum objective function to obtain optimized mapping parameters, and selecting an optimal dynamic model and optimized mapping parameters by using the verification set.
The batch process modeling and final product quality prediction method based on big data has the advantages that: compared with the prior art, the method uses an encoder to nonlinearly map the process measurement variable and the control input variable into a higher-dimensional state variable, and uses a cyclic neural network to construct a state variable dynamic model; mapping the state variable back to the measurement variable through a decoder to form a nonlinear mapping parameter; learning nonlinear mapping parameters and dynamic models by using batch process historical operation big data; and establishing a final product quality prediction model in the batch process by using the state variable obtained by learning inference and combining an attention mechanism. The invention combines the nonlinear mapping and the dynamic process in a learning training frame by batch process modeling, thereby avoiding modeling and learning in stages; the final product quality prediction model explicitly considers the influence of the state variable at the historical moment on the final product quality, and the problem that the performance of the cyclic neural network for predicting the final product quality is reduced along with the increase of the input length is solved.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed for the embodiments or the prior art descriptions will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
FIG. 1 is a flowchart of a method for batch process modeling and final product quality prediction based on big data according to an embodiment of the present invention;
Detailed Description
In order to make the technical problems, technical solutions and advantageous effects to be solved by the present invention more clearly apparent, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Referring to fig. 1, a batch process modeling and final product quality prediction method based on big data according to the present invention will now be described. The batch process modeling and final product quality prediction method based on big data specifically comprises the following steps:
s1: nonlinearly mapping a process measurement variable and a control input variable into a higher-dimensional state variable by using an encoder, and constructing a state variable dynamic model by using a recurrent neural network;
s2: mapping the state variable back to the measurement variable through a decoder to form a nonlinear mapping parameter;
s3: learning nonlinear mapping parameters and dynamic models by using batch process historical operation big data;
s4: and establishing a final product quality prediction model in the batch process by using the state variable obtained by learning inference and combining an attention mechanism.
Compared with the prior art, the batch process modeling and final product quality prediction method based on big data combines the nonlinear mapping and the dynamic process in a learning training frame, so that staged modeling and learning are avoided; the final product quality prediction model explicitly considers the influence of the state variable at the historical moment on the final product quality, and the problem that the performance of the cyclic neural network for predicting the final product quality is reduced along with the increase of the input length is solved.
In a specific embodiment of the present invention, the method comprises the following steps:
(1) training a batch process nonlinear dynamic model based on a cyclic neural network of an encoder-decoder structure, wherein the process is as follows:
is provided withFor the process measurement vector of the kth sampling time of the b-th batch, setFor the control vector of the kth sampling time of the b-th batch, let q(b)And the final quality index vector of the product of the batch b, such as the number average molecular weight, the weight average molecular weight, the polydispersity index and the like, is obtained by laboratory test analysis after the batch process operation is finished. Let the number of sampling instants per batch be NbAnd collecting historical operation data of PI control or open loop control in B batches to establish a training database.
(1.a) establishing the following batch process nonlinear discrete dynamic model based on the encoder-decoder recurrent neural network:
wherein the intermediate vector zkThe sum of the nonlinear mapping of the state vector and the nonlinear mapping of the control vector;
vk=σ(Vhhk+Vzzk+bv);
wherein, Vh、VzFor calculating a weighting vector vkWeight matrix of bvIs its offset vector;
wherein the state vector hk+1From hkAnd the intermediate vector zkMeridian vkThe weight is obtained by the weighting of the received signal,adding elements in the two vectors respectively, wherein 1 represents a vector with all the elements being 1; setting up a suitable neural network hierarchy such thatAnd hkHas consistent dimension, and because the number of observable variables in the chemical and pharmaceutical processes is small, the state variable h in the model iskDimension N ofhShould be greater than the observation vector xkDimension (d) of (a).
Encoder fe(. The) is composed of an L-layer neural network, and the expression is as follows:
wherein the content of the first and second substances,is a nonlinear activation function tanh, WlAs a weight matrix of the neural network, clIs a bias vector;
decoder fd(. The) also comprises an L-layer neural network, and the expression is as follows:
wherein the content of the first and second substances,is a matrix WlTranspose of (d)lIs a bias vector.
(1.b) learning and training the dynamic model by using the batch process operation historical data. Gradually sliding the sequence data of each batch b for s steps, and taking the window length as NwTo construct a training set of sequences. Let the total sequence number be NsLet the j-th sequence of data beWill NsThe sequence data sets are divided into training and validation sets.
Let the number of training sequence sets be NtrCombining the parameter set to be optimized of the model into theta, and setting the first vector of the jth sequence asArranged in a modelOptimizing a minimization objective function:
the optimized parameter theta can be obtained. And selecting an optimal model structure and optimal parameters by using the verification set.
(2) And (3) constructing a final product quality prediction model by using the state vector by using an attention mechanism, wherein the process is as follows:
(2.a) establishing a final product quality prediction model based on an attention mechanism:
is provided withIs the state variable at the last moment of the batch b,is the state variable of the kth time of the b-th batch, and is calculated firstlyFor each state beforeThe weight value of (2):
the score function here uses the following calculation method:
Will next beAndspliced into a vector and multiplied by a full-connection matrix W of appropriate dimensionscObtaining the state of adding attention mechanism
wherein the matrix Wa、Wh0And an offset vector bho(set as parameter set θ) will be obtained by learning training.
(2.b) the states of all the lots learned by the step (1)Product quality index q obtained by actual test(b)Learning a final product quality prediction model, wherein an optimization objective function is as follows:
in addition, the quality prediction model is applied to implementation of batch control. As the early stage of batch processes in many chemical and pharmaceutical fields is generally a heating strategy, a middle-stage cooling strategy and a later-stage readjusting heating strategy, a simple open-loop control strategy or a temperature closed-loop control strategy can be adopted in the first half of process control, the model is applied to carry out multi-step prediction in the second half, and the control is carried out by combining a rolling optimization method in model prediction control.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.
Claims (5)
1.A batch process modeling and final product quality prediction method based on big data is characterized by comprising the following steps:
s1: nonlinearly mapping a process measurement variable and a control input variable into a higher-dimensional state variable by using an encoder, and constructing a state variable dynamic model by using a recurrent neural network;
s2: mapping the state variable back to the measurement variable through a decoder to form a nonlinear mapping parameter;
s3: learning nonlinear mapping parameters and dynamic models by using batch process historical operation big data;
s4: and establishing a final product quality prediction model in the batch process by using the state variable obtained by learning inference and combining an attention mechanism.
2. The big-data based batch process modeling and final product quality prediction method of claim 1, wherein:
in steps S1 and S2, the encoder f is usedcAnd feAnd nonlinearly mapping the process measurement variable x (t) and the control input variable u (t) at the time t into a higher-dimensional state variable, and constructing a state variable dynamic model by using a Recurrent Neural Network (RNN):
the state variable h (t) at the time t is processed by a decoder fdNonlinear mapping back to the measured variable x (t), forming a nonlinear mapping parameter:
x(t)=fd(h(t));
in step S4, a batch process end product quality index variable q (t) is established in conjunction with Attention mechanism Attentionf) And (3) a prediction model, namely fitting the quality prediction model by using the historical batch quality analysis result:
q(tf)=Attention(h1,…,ht,…hf)。
3. the method of claim 1, wherein in step S1, the measurement vectors of the sampling batches and sampling times are selected to form process measurement variables, the control vectors of the sampling batches and sampling times are selected to form control input variables, and the measurement vectors are used to build a training database using the time-lapse running data of PI control or open-loop control.
4. The big-data based batch process modeling and final product quality prediction method as claimed in claim 3, wherein the measurement vectors include temperature, pressure, density and viscosity values converted by the torque of the stirring motor of the reaction kettle; the control vectors include steam jacket pressure and vent rate.
5. The method of claim 1, wherein in step S3, the sequence data of multiple batches is slid step by step, and a training set of sequence data building blocks with a specific window length is obtained; and dividing the sequence data sets into a training set and a verification set, optimizing the minimum objective function to obtain optimized mapping parameters, and selecting an optimal dynamic model and optimized mapping parameters by using the verification set.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111302383.4A CN114036832A (en) | 2021-11-04 | 2021-11-04 | Batch process modeling and final product quality prediction method based on big data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111302383.4A CN114036832A (en) | 2021-11-04 | 2021-11-04 | Batch process modeling and final product quality prediction method based on big data |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114036832A true CN114036832A (en) | 2022-02-11 |
Family
ID=80136367
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111302383.4A Pending CN114036832A (en) | 2021-11-04 | 2021-11-04 | Batch process modeling and final product quality prediction method based on big data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114036832A (en) |
-
2021
- 2021-11-04 CN CN202111302383.4A patent/CN114036832A/en active Pending
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109060001B (en) | Multi-working-condition process soft measurement modeling method based on feature transfer learning | |
CN109581870B (en) | Dynamic matrix control method for temperature in energetic material reaction kettle | |
CN108920888B (en) | Continuous stirred tank type reaction process identification method based on deep neural network | |
CN108920863B (en) | Method for establishing energy consumption estimation model of robot servo system | |
CN111079836A (en) | Process data fault classification method based on pseudo label method and weak supervised learning | |
CN110609476B (en) | Multivariate nonlinear dynamic system model prediction control method based on Gaussian process model | |
CN113012766B (en) | Self-adaptive soft measurement modeling method based on online selective integration | |
CN105116726A (en) | Parameter design method for nonlinear predictive controller based on mechanism model | |
CN104504288A (en) | Method for non-linear multistage intermittent process soft measurement based on multi-directional support vector cluster | |
CN113031553A (en) | Intermittent process quality prediction method based on automatic clustering combined with partial least squares | |
CN110245398B (en) | Soft measurement deep learning method for thermal deformation of air preheater rotor | |
CN117455320B (en) | QGMVPN model-based multi-kettle polymerization process melt quality index online prediction method | |
CN110221540B (en) | Continuous stirred reactor system control method based on Hammerstein model | |
CN113011102B (en) | Multi-time-sequence-based Attention-LSTM penicillin fermentation process fault prediction method | |
CN114239397A (en) | Soft measurement modeling method based on dynamic feature extraction and local weighted deep learning | |
CN114004346A (en) | Soft measurement modeling method based on gating stacking isomorphic self-encoder and storage medium | |
JPH06337852A (en) | Time series prediction method by neural network | |
Babuška et al. | Fuzzy modeling of enzymatic penicillin-G conversion | |
CN110794676A (en) | CSTR process nonlinear control method based on Hammerstein-Wiener model | |
CN114036832A (en) | Batch process modeling and final product quality prediction method based on big data | |
CN110826184B (en) | Variational Bayesian identification method for NARX model structure and parameters under time-varying lag | |
Zhang et al. | Active learning strategy for online prediction of particle size distribution in cobalt oxalate synthesis process | |
CN110879873B (en) | Soft measurement method and system for vine copula correlation description based on Hamilton Monte Carlo sampling | |
Zhang et al. | Reliable multi-objective on-line re-optimisation control of a fed-batch fermentation process using bootstrap aggregated neural networks | |
Guo et al. | Modelling for multi-phase batch processes using steady state identification and deep recurrent neural network |
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 |