Key process index soft measurement method based on dynamic mechanism model parameter evolution

Abstract

The invention relates to the technical field of industrial process key index monitoring, and particularly discloses a key process index soft measurement method based on dynamic mechanism model parameter evolution, which ensures the precision of an embedded hybrid model by establishing the embedded hybrid model based on a mechanism model and a GraphTrans model and estimating the parameters of the mechanism model through the GraphTrans model, and improves the accuracy of the embedded hybrid model. Meanwhile, specific physical significance is provided for dynamic mechanism model parameters output by the GraphTrans model through the mechanism model, prediction of key indexes of the industrial process is completed through the embedded hybrid model, and the problem that a traditional soft measurement method with a comprehensive state space is introduced, and the calculation efficiency is greatly improved in the calculation process of the weighted sum is solved. And conflicts between static parameters and dynamic characteristics restrict the application effect of the integrated mechanism model.

Description

Technical Field

[0001] The present application relates to the technical field of industrial process key indicator monitoring, and specifically discloses a key process indicator soft measurement method based on dynamic mechanism model parameter evolution. Background Art

[0002] In the process of achieving high-level manufacturing in modern industry, key process indicators have become an important foothold for regulating the production process. Various optimization and control methods are carried out around the key process indicators of different industrial objects. In the process of obtaining key process indicators of some industrial processes, the acquisition of key process indicators through soft measurement technology can make up for the problem of low detection frequency of key process indicators.

[0003] Traditional soft sensing methods, by establishing mechanism models, use knowledge from fields such as thermodynamics, electrochemistry, and fluid mechanics to summarize and generalize system phenomena, and describe the reaction details of the modeling object in the form of partial differential equations. However, the relationship between industrial process parameters and mechanism model parameters is complex. At the same time, in complex industrial modeling tasks, there is a conflict between dynamic industrial reaction characteristics and static parameters of the mechanism model, which limits their accuracy.

[0004] Therefore, in the prior art, a comprehensive state space is introduced to describe the highly complex dynamics of industrial processes and the additional requirements imposed by intelligent and optimal manufacturing systems. In the application process, the comprehensive state space is divided into subspaces indicating different operating conditions in view of the multimodal and time-varying characteristics of the reaction process dynamics. The mechanism model is trained for each different operating subspace to obtain the corresponding mechanism model parameters. Afterwards, the system dynamics is characterized as the weighted sum of the submodels, where the weight is the probability that the current operating point belongs to different operating conditions, and the weight will be updated as the operating point moves in the comprehensive state space. This integrated modeling method describes the dynamic reaction characteristics of industrial objects with the weighted sum of mechanism models, which overcomes the accuracy defects of the mechanism model to a certain extent and performs well in the prediction task of key process indicators in the hydrometallurgical zinc purification process.

[0005] However, the introduced comprehensive state space describes the dynamic response characteristics of industrial objects in the form of a weighted sum of multiple mechanism models. The characterization ability of a single mechanism model is limited. The contradiction between its static mechanism parameters and the dynamic response characteristics of the modeled object will lead to insufficient modeling accuracy of the mechanism model, restricting its application effect. The weighted sum of the mechanism model does not fundamentally overcome the contradiction between the static parameters of the mechanism model and the dynamic response characteristics of the modeled object. A single static mechanism model is difficult to describe the dynamic response characteristics of a class of sub-conditions, and the weighted combination of a series of static mechanism models is also difficult to describe the complete reaction process. The conflict between static parameters and dynamic characteristics will restrict the application effect of the integrated mechanism model.

[0006] The present invention provides a key process index soft measurement method based on dynamic mechanism model parameter evolution to solve the above problems. Summary of the invention

[0007] The purpose of the present invention is to solve the problem that in the traditional soft measurement method with integrated state space, the conflict between static parameters and dynamic characteristics in the weighted sum calculation process will restrict the application effect of the integrated mechanism model.

[0008] In order to achieve the above object, the basic scheme of the present invention provides a soft measurement method for key process indicators based on the evolution of dynamic mechanism model parameters, including constructing an embedded hybrid model, wherein the embedded hybrid model includes: A mechanism model, wherein the mechanism model is constructed based on the knowledge of the field to which the process belongs; A GraphTrans model embedded in the mechanism model, wherein the GraphTrans model includes S layers of coding blocks stacked in sequence and a fully connected layer connected to the coding block of the Sth layer, each coding block including a graph convolutional neural network module and a Transformer module; According to the established embedded hybrid model, soft measurement of key process indicators is performed, including the following steps: Data extraction: extract the process parameters of the process being carried out, and establish a correlation network between the process parameters based on the knowledge of the field to which the process belongs, and obtain the graph structure data between the process parameters; The input of data, the mechanism model synchronously receives the extracted process parameters and dynamic mechanism model parameters, and the dynamic mechanism model parameters are calculated by the GraphTrans model based on the graph structure data established according to the association network; For data processing, the mechanism model receives process parameters and dynamic mechanism model parameters, makes real-time estimation of key process indicators, and completes soft measurement of key process indicators.

[0009] Furthermore, the graph structure data is an adjacency matrix of the graph, and an association network between process parameters is established according to the domain knowledge to which the process belongs, and a graph mask matrix between process parameters is also obtained.

[0010] Furthermore, after obtaining the graph mask matrix, the graph mask matrix is ​​supplemented into each encoding block in the GraphTrans model to supplement the graph mask matrix into the Transformer module.

[0011] Furthermore, the graph mask matrix is ​​expressed as follows: ; ; Where M represents the image mask matrix, m ijis the element of the graph mask matrix, M is a square matrix of size N×N, N is the number of variable nodes; when node i can affect the change of node j, m ij =1, when node i cannot affect the change of node j, m ij =0.

[0012] Furthermore, the mechanism model is constructed based on the first-order reaction kinetics equation, the material conservation equation, the Arrhenius equation and the continuous stirred tank reactor equation.

[0013] Furthermore, after the mechanism model synchronously receives the extracted process parameters and the dynamic mechanism model parameters, the calculation formula of the prediction process is as follows: Y=f mech (X;θ); In the formula, f mech (•) represents the mechanism model, θ is the dynamic parameter output by the GraphTrans model, X is the process parameter, and Y is the predicted key process indicator.

[0014] Furthermore, the steps of obtaining dynamic mechanism model parameters of the GraphTrans model are as follows: The encoding block of the first layer receives all the initial node features, and aggregates the features from neighboring nodes through the graph convolutional neural network module in the encoding block of the first layer; The features of the aggregated neighbor nodes are input into the Transformer module in the encoding block of the same layer. By using the Transformer module with a graph mask matrix and combining the multi-head self-attention mechanism inside the Transformer module, different variable correlation patterns are extracted to obtain the attention features between the associated variables. The obtained attention features are input into the encoding block of the next layer and feature extraction is implemented layer by layer. Finally, the encoding block of the last layer inputs the results into the fully connected layer to obtain the dynamic mechanism model parameters.

[0015] Furthermore, the process of aggregating features from neighboring nodes by the graph convolutional neural network module in the encoding block of the first layer is as follows: ; In the formula, H l is the node feature matrix of the lth layer, is the adjacency matrix with self-connection added, and , initially H 0 =X; In the formula, A represents the adjacency matrix, , I is the identity matrix, yes The degree matrix of , W lis the weight matrix of the lth layer, and , σ(•) is a nonlinear activation function; Where R is a real number, N is the number of variables, which is also the number of nodes in the graph, and F is the characteristic length.

[0016] Furthermore, the process of extracting the attention features between the associated variables by using the Transformer module with a graph mask matrix is ​​as follows: Calculate the attention weights: ; Where Q, K and V are the query matrix, key matrix and value matrix respectively, and Q = XW q , K = XW k , V = XW v , W q ,W k ,W v denotes a learnable weight matrix, and , is the scaling factor used to stabilize the gradient, and T is the transposed sign of the matrix; Calculate the output A of each head in the multi-head self-attention mechanism h : ; In the formula, Q h ,K h ,V h are the query matrix, key matrix and value matrix of the h-th head respectively, and T is the transpose symbol of the matrix; The outputs of all heads are concatenated and transformed linearly to generate the final output, which is summarized in the calculation process. The calculation formula is as follows: ; Where ⊙ represents element-by-element multiplication, and M represents the graph mask matrix; The Replace function ensures zero attention weight by setting the mask value to a negative number. The expression is as follows: Replace(i)={i,i≠0;-1e 9 ,i=0}.

[0017] Furthermore, the process of calculating the dynamic mechanism model parameters by the fully connected layer is as follows: θ = W out H S ; In the formula, H S is the attention feature output by the last layer encoding block, W out is a learnable weight matrix, representing the learned mechanism model parameters, and , P is the mechanism parameter dimension.

[0018] The principles and effects of this basic solution are: 1. Compared with the prior art, the present invention establishes an embedded hybrid model based on the mechanism model and the GraphTrans model, estimates the parameters of the mechanism model through the GraphTrans model, and ensures the accuracy of the embedded hybrid model. At the same time, the mechanism model provides specific physical meanings for the dynamic mechanism model parameters output by the GraphTrans model, enhances the interpretability of the embedded hybrid model, and completes the prediction of key indicators of the industrial process through the embedded hybrid model, solving the problem that the traditional soft measurement method with integrated state space is introduced, and the conflict between static parameters and dynamic characteristics in the weighted sum calculation process will restrict the application effect of the integrated mechanism model.

[0019] 2. Compared with the prior art, the present invention introduces a graphical mask matrix into the Transformer module and applies it in the calculation process of the self-attention mechanism, introduces the correlation information of the matrix variables into the Transformer module, and clears the influence of irrelevant variables on the attention weight to prevent the generation of useless information, thereby better modeling industrial processes with complex variable relationships. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 A schematic diagram of the hydrometallurgical zinc purification process proposed in the embodiment of the present application is shown; Figure 2 A flow chart of a key process indicator soft-sensing method based on dynamic mechanism model parameter evolution proposed in an embodiment of the present application is shown; Figure 3 A schematic diagram showing the difference between the soft measurement prediction results and the true values ​​of the key process indicators of the cobalt removal reactor No. 1 using the key process indicator soft measurement method based on the dynamic mechanism model parameter evolution proposed in the embodiment of the present application; Figure 4 The graph shows the change curve of the parameters output by the GraphTrans model proposed in the embodiment of the present application, where (a) is the parameter A β (b) is the change curve of parameter E e (c) is the change curve of parameter α, and (d) is the change curve of parameter e eq 's change curve graph. DETAILED DESCRIPTION

[0022] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation mode, structure, characteristics and effects of the present invention are described in detail below in combination with the accompanying drawings and preferred embodiments.

[0023] The key process indicator soft measurement method based on the dynamic mechanism model parameter evolution is shown in the following embodiment, which is an example of applying the key process indicator soft measurement method based on the dynamic mechanism model parameter evolution provided by the present invention to the soft measurement of cobalt ion concentration in the cobalt removal reactor No. 1 in the purification process of a wet zinc smelting process.

[0024] In the hydrometallurgical zinc purification process, Figure 1 As shown, there are two steps of copper removal sub-processes and five steps of cobalt removal sub-processes in sequence. In this embodiment, only the cobalt removal sub-process is shown. In the cobalt removal sub-process, five reactors are built in series in sequence, from the cobalt removal reactor No. 1 to the cobalt removal reactor No. 5. Reactor No. 1 receives the solution overflowed from the copper removal, and completes the cobalt removal process through a cascade reaction. In the purification process of the wet zinc smelting process, the cobalt ion concentration is used as a key process indicator. In this embodiment, the outlet cobalt ion concentration of the cobalt removal reactor No. 1 is used as a key process indicator for soft measurement. The specific implementation is as follows: First, an embedded hybrid model is constructed. The constructed embedded hybrid model includes: The mechanism model is constructed based on the domain knowledge of the process, and is specifically constructed based on the domain knowledge and process characteristics of the cobalt removal process in the field of hydrometallurgy.

[0025] The mechanism model is an important source of interpretability for the embedded hybrid model, and provides specific physical meanings for the parameters ultimately output by the GraphTrans model. The mechanism model describes the inherent physical and chemical reaction laws through its internal structure, thus providing structural interpretability for the model. The parameters output by the GraphTrans model are directly used as the mechanism parameters of the mechanism model. Although these parameters were originally dimensionless parameters, under the prediction framework of the mechanism model, through the training process of prediction error back propagation, the parameters output by the GraphTrans model gradually approach the true mechanism parameters. In other words, these parameters play the role that true mechanism parameters should play within the prediction framework of the mechanism model, and can accurately describe production phenomena, so they can be equivalently regarded as true mechanism parameters.

[0026] In practical application, the mechanism model of the cobalt removal reactor f mech(•)The establishment is based on the first-order reaction kinetics equation, material conservation equation, Arrhenius equation and continuous stirred tank reactor equation. The partial differential equation of the mechanism model is as follows: ; Among them, the process parameter information in the established mechanism model is shown in Table 1 below: Table 1

[0027] Moreover, in this embodiment, the parameters that need to be identified in the mechanism model include the reaction coefficient A β , ion precipitation activation energy E e , cathode transfer coefficient α and cathode reaction equilibrium potential e eq .

[0028] The embedded hybrid model also includes a GraphTrans model embedded in the mechanism model, wherein the GraphTrans model includes S layers of coding blocks stacked in sequence and a fully connected layer connected to the coding block of the Sth layer, each coding block includes a graph convolutional neural network (GCN) module and a Transformer module, and the output of the GraphTrans model is used as the input of the mechanism model.

[0029] The GraphTrans model is a data model that integrates the graph convolutional neural network (GCN) and the Transformer model. Figure 2 As shown, Figure 2 The components of the GraphTrans model are shown in Figure 2.

[0030] The mechanism model in the embedded hybrid model is used as a prediction model to estimate key process indicators and complete soft measurement of key process indicators.

[0031] Specifically, according to the established embedded hybrid model, soft measurement of key process indicators is performed, including the following steps: Data extraction: extract the process parameters of the process being carried out, and establish a correlation network between the process parameters based on the knowledge of the field to which the process belongs, and obtain the graph structure data between the process parameters; Data input, the mechanism model synchronously receives the extracted process parameters and dynamic mechanism model parameters, and the dynamic mechanism model parameters are calculated by the GraphTrans model receiving the graph structure data; Data processing, after the mechanism model receives the process parameters and dynamic mechanism model parameters, it makes real-time estimation of key process indicators and completes soft measurement of key process indicators.

[0032] The established graph structure data includes the adjacency matrix of the graph, which is expressed as follows: ; In the formula, A represents the adjacency matrix, a ij is the element of the adjacency matrix, A is a square matrix of size N×N, N is the number of variable nodes, v ij is the directed edge connecting node i and node j, E is the set of all directed edges, when v ij ∈E, it means there is a directed edge from node i to node j. At this time, a ij =1, otherwise a ij =0, there are 10 nodes in the association graph obtained in this embodiment, N=10.

[0033] The adjacency matrix A represents the direction of information transmission between node variables and can guide the information aggregation process of the GCN module.

[0034] According to the association network between the process parameters, a graph mask matrix is ​​also established, and the graph mask matrix is ​​added to each encoding block in the GraphTrans model. Specifically, the graph mask matrix is ​​added to the Transformer module.

[0035] The image mask matrix created is as follows: ; ; Where M represents the image mask matrix, m ij is the element of the graph mask matrix, M is a square matrix of size N×N, N is the number of variable nodes, M can represent the causal relationship between nodes, when node i can affect the change of node j, m ij =1, when node i cannot affect the change of node j, m ij =0.

[0036] In this embodiment, the graph mask matrix is ​​introduced into the calculation mechanism of the Transformer model, so that the Transformer can be equipped with the ability to process graph structured data, thereby more efficiently extracting variable association features in process parameters.

[0037] In the process of data input, after the adjacency matrix of the graph is input into the GraphTrans model, the steps for the GraphTrans model to obtain the dynamic mechanism model parameters are as follows: The encoding block of the first layer receives all the initial node features, aggregates the features from neighboring nodes through the GCN module in the encoding block of the first layer, and captures the local structure information. The specific calculation process is as follows: ; In the formula, H lis the node feature matrix of the lth layer, is the adjacency matrix with self-connection added, and , initially H 0 =X, X is the standardized variable matrix, and , R is a real number, N is the number of variables, which is also the number of nodes in the graph, and F is the feature length. In the original data, it is the time step length. One time step obtains one sample value, and F time steps represent the sample values ​​of a period of time, and , I is the identity matrix, yes The degree matrix of , W l is the weight matrix of the lth layer, and , σ(•) is a nonlinear activation function, such as ReLU.

[0038] The features of the aggregated neighbor nodes are input into the Transformer module in the encoding block of the same layer. By using the Transformer module with a graph mask matrix and combining the multi-head self-attention mechanism inside the Transformer module, different variable correlation patterns are extracted to obtain the attention features between the associated variables. The specific calculation is as follows: Calculate the attention weights: ; Where Q, K and V are the query matrix, key matrix and value matrix respectively, and Q = XW q , K = XW k , V = XW v , W q ,W k ,W v denotes a learnable weight matrix, and , is the scaling factor used to stabilize the gradient, and T is the transposed sign of the matrix.

[0039] Calculate the output A of each head in the multi-head self-attention mechanism h : ; In the formula, Q h ,K h ,V h are the query matrix, key matrix and value matrix of the h-th head respectively, and T is the transpose symbol of the matrix.

[0040] The outputs of all heads are concatenated and linearly transformed to generate the final output, and are aggregated during the calculation process. The mask matrix is ​​applied to the attention mechanism to eliminate the attention weights of irrelevant node pairs. The calculation formula is as follows: ; Where ⊙ represents element-by-element multiplication, M represents the graph mask matrix, T is the transpose sign of the matrix, and the Replace function ensures zero attention weight by setting the mask value to a sufficiently small negative number: Replace(i)={i,i≠0;-1e 9 ,i=0}.

[0041] By applying a mask matrix, the attention mechanism is focused only on causally related pairs of variables; The obtained attention features are input into the encoding block of the next layer and feature extraction is realized layer by layer. Finally, the encoding block of the last layer converts the feature H S Input to the fully connected layer to obtain dynamic mechanism model parameters: θ = W out H S ; In the formula, H S is the attention feature output by the last layer encoding block, W out is a learnable weight matrix, representing the learned mechanism model parameters, and , P is the dimension of the mechanism parameters; each element in θ will be used as a parameter inside the mechanism model, so its output dimension needs to be specially set to match the number of parameters required by the mechanism model.

[0042] After that, the data is processed, and the mechanism model receives the process parameters and dynamic mechanism model parameters. The key process indicators are estimated in real time through the partial differential equations of the established mechanism model to complete the soft measurement of the key process indicators. The difference between the predicted results and the true values ​​is as follows: Figure 3 shown.

[0043] The prediction process of the mechanism model is as follows:

[0044] Y=f mech (X;θ); In the formula, f mech (•) represents the mechanism model, which is expressed as a set of partial differential equations derived from domain knowledge, θ is the dynamic parameter output by the GraphTrans model, X is the process parameter, and Y is the predicted key process indicator, such as ion concentration.

[0045] When the mechanism model receives the dynamic mechanism model parameters, the mechanism model gives specific physical meanings to the parameter values ​​output by the GraphTrans model. β 、E e , α and e eq The change curves are as follows Figure 4As shown in Figure (a), Figure (b), Figure (c), and Figure (d), it can be seen that each parameter variable fluctuates around a specific value, but the fluctuation range is different. It is consistent with the actual production conditions. After the process design is completed and production begins, the internal reaction state of the reactor should be kept as stable as possible to ensure the production of qualified products. The fluctuation of the parameter value of the mechanism model directly reflects the parameter mismatch phenomenon.

[0046] In this embodiment, five other models are introduced to predict the cobalt ion concentration at the outlet of the No. 1 cobalt removal reactor. Combined with the embedded hybrid model, a total of six models are shown in Table 2 below: Table 2

[0047] The present invention uses mean absolute error (MAE), root mean square error (RMSE), mean absolute percentage error (MAPE) and R 2 To measure the accuracy of the model, the specific calculation formula is as follows: ; ; ; ;

[0048] Where N is the number of samples in the test data set, R n is the actual value of the nth sample, P n is the predicted value of the nth sample, is the mean of the true values ​​of the samples, is the mean of the predicted values. The performance of the six models shown in Table 2 on the four indicators is shown in Table 3: Table 3

[0049] It can be seen that the GraphTrans model proposed in the present invention is superior to a series of classic models such as GCN and Transformer, which verifies the excellent dynamic feature extraction ability of the model and can detect subtle differences in reactor operation modes from process parameters. At the same time, it can be found that GraphTrans-CM is significantly superior to GraphTrans, proving the effectiveness of the embedded hybrid model proposed in the present invention. The effective combination of data model and mechanism model can further improve the prediction accuracy of process indicators.

[0050] The present invention establishes an embedded hybrid model based on a mechanism model and a GraphTrans model, estimates the parameters of the mechanism model in real time through the GraphTrans model, thereby ensuring the accuracy of the embedded hybrid model. At the same time, the mechanism model provides specific physical meanings for the dynamic mechanism model parameters output by the GraphTrans model, thereby enhancing the interpretability of the embedded hybrid model. The embedded hybrid model is used to predict key indicators of the industrial process, thereby solving the problem that in the traditional soft measurement method with an integrated state space, the conflict between static parameters and dynamic characteristics in the weighted sum calculation process will restrict the application effect of the integrated mechanism model.

[0051] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modifications to equivalent embodiments of equivalent changes by using the technical contents disclosed above without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A soft-sensing method for key process indicators based on the evolution of dynamic mechanism model parameters, characterized in that: The method comprises building an embedded hybrid model, wherein the embedded hybrid model comprises: A mechanism model, wherein the mechanism model is constructed based on the knowledge of the field to which the process belongs; A GraphTrans model embedded in the mechanism model, wherein the GraphTrans model includes S layers of coding blocks stacked in sequence and a fully connected layer connected to the coding block of the Sth layer, each coding block including a graph convolutional neural network module and a Transformer module; According to the established embedded hybrid model, soft measurement of key process indicators is performed, including the following steps: Data extraction: extract the process parameters of the process being carried out, and establish a correlation network between the process parameters based on the knowledge of the field to which the process belongs, and obtain the graph structure data between the process parameters; The input of data, the mechanism model synchronously receives the extracted process parameters and dynamic mechanism model parameters, and the dynamic mechanism model parameters are calculated by the GraphTrans model based on the graph structure data established according to the association network; For data processing, the mechanism model receives process parameters and dynamic mechanism model parameters, makes real-time estimation of key process indicators, and completes soft measurement of key process indicators.

2. The key process indicator soft-sensing method based on the dynamic mechanism model parameter evolution according to claim 1 is characterized in that: The graph structure data is an adjacency matrix of the graph, and an association network between process parameters is established according to the domain knowledge to which the process belongs, and a graph mask matrix between the process parameters is also obtained.

3. The key process indicator soft-sensing method based on the dynamic mechanism model parameter evolution according to claim 2 is characterized in that: After obtaining the graph mask matrix, each encoding block in the GraphTrans model is supplemented with the graph mask matrix to supplement the graph mask matrix into the Transformer module.

4. The key process indicator soft-sensing method based on the dynamic mechanism model parameter evolution according to claim 2 is characterized in that: The graph mask matrix is ​​represented as follows: ； ； Where M represents the image mask matrix, m ij is the element of the graph mask matrix, M is a square matrix of size N×N, where N is the number of variable nodes; When node i can affect the change of node j, m ij =1, when node i cannot affect the change of node j, m ij =0.

5. The key process indicator soft-sensing method based on dynamic mechanism model parameter evolution according to claim 1 is characterized in that: The mechanism model is constructed based on the first-order reaction kinetics equation, the material conservation equation, the Arrhenius equation and the continuous stirred tank reactor equation.

6. The key process indicator soft-sensing method based on dynamic mechanism model parameter evolution according to claim 1 is characterized in that: After the mechanism model synchronously receives the extracted process parameters and the dynamic mechanism model parameters, the calculation formula of the prediction process is as follows: Y=f mech (X;θ); In the formula, f mech (•) represents the mechanism model, θ is the dynamic parameter output by the GraphTrans model, X is the process parameter, and Y is the predicted key process indicator.

7. The key process indicator soft-sensing method based on dynamic mechanism model parameter evolution according to claim 2 is characterized in that: The steps of obtaining dynamic mechanism model parameters of the GraphTrans model are as follows: The encoding block of the first layer receives all the initial node features, and aggregates the features from neighboring nodes through the graph convolutional neural network module in the encoding block of the first layer; The features of the aggregated neighbor nodes are input into the Transformer module in the encoding block of the same layer. By using the Transformer module with a graph mask matrix and combining the multi-head self-attention mechanism inside the Transformer module, different variable correlation patterns are extracted to obtain the attention features between the associated variables. The obtained attention features are input into the encoding block of the next layer and feature extraction is implemented layer by layer. Finally, the encoding block of the last layer inputs the results into the fully connected layer to obtain the dynamic mechanism model parameters.

8. The key process indicator soft-sensing method based on dynamic mechanism model parameter evolution according to claim 7 is characterized in that: The process of aggregating features from neighboring nodes by the graph convolutional neural network module in the encoding block of the first layer is as follows: ； In the formula, H l is the node feature matrix of the lth layer, is the adjacency matrix with self-connection added, and , initially H 0 =X; In the formula, A represents the adjacency matrix, , I is the identity matrix, yes The degree matrix of , W l is the weight matrix of the lth layer, and , σ(•) is a nonlinear activation function; Where R is a real number, N is the number of variables, which is also the number of nodes in the graph, and F is the characteristic length.

9. The key process indicator soft-sensing method based on dynamic mechanism model parameter evolution according to claim 8, characterized in that: The process of extracting the attention features between associated variables by using the Transformer module with a graph mask matrix is ​​as follows: Calculate the attention weights: ； Where Q, K and V are the query matrix, key matrix and value matrix respectively, and Q = XW q , K = XW k , V = XW v , W q ,W k ,W v denotes a learnable weight matrix, and , is the scaling factor used to stabilize the gradient, and T is the transposed sign of the matrix; Calculate the output A of each head in the multi-head self-attention mechanism h : ； In the formula, Q h ,K h ,V h are the query matrix, key matrix and value matrix of the h-th head respectively, and T is the transpose symbol of the matrix; The outputs of all heads are concatenated and transformed linearly to generate the final output, which is summarized in the calculation process. The calculation formula is as follows: ； Where ⊙ represents element-by-element multiplication, and M represents the graph mask matrix; The Replace function ensures zero attention weight by setting the mask value to a negative number. The expression is as follows: Replace(i)={i,i≠0;-1e 9 ,i=0}。 10. The key process indicator soft-sensing method based on dynamic mechanism model parameter evolution according to claim 8, characterized in that: The process of calculating the dynamic mechanism model parameters by the fully connected layer is as follows: θ=W out H S ; In the formula, H S is the attention feature output by the last layer encoding block, W out is a learnable weight matrix, representing the learned mechanism model parameters, and , P is the mechanism parameter dimension.

Images