Process quality prediction method and device, coefficient determination method and device and electronic equipment

Abstract

The invention provides a process quality prediction method and device, a coefficient determination method and device and electronic equipment, and relates to the technical field of production. According to the method, the process quality of a target process is predicted according to process parameters through an uncertainty agent model, the uncertainty agent model is a model based on chaos polynomial expansion, and a polynomial coefficient in the uncertainty agent model is obtained according to a process quality prediction result obtained by a machine learning model based on process parameter prediction. Due to the fact that the machine learning model can rapidly achieve prediction and is higher in timeliness, online deployment is more facilitated, real-time prediction of the process quality is achieved, prediction is achieved through the uncertainty agent model, uncertainty in the process parameters can be fully considered, and prediction accuracy can be improved.

Description

Technical Field

[0001] The present application relates to the field of production technology, and in particular to a process quality prediction method, coefficient determination method, device and electronic equipment. Background Art

[0002] In the process of process production, the quality of the product mainly depends on the process parameters involved in the product processing. The unreasonable design of process parameters has a great impact on the quality of the product. There are many factors that affect the process quality in the production process, and there is a strong uncertainty and coupling in the process parameters during the control process. In order to predict the process quality, the relevant technology generally inputs the process-related variables into the simulation model for simulation to predict the process quality. However, the simulation model has a long running process and poor timeliness, and cannot meet the requirements of real-time online prediction. Summary of the invention

[0003] The purpose of the embodiments of the present application is to provide a process quality prediction method, coefficient determination method, device and electronic equipment to improve the problem that the existing prediction methods have poor timeliness and cannot meet the requirements of real-time online prediction.

[0004] In a first aspect, an embodiment of the present application provides a process quality prediction method, the method comprising:

[0005] Obtaining process parameters involved in target process processing;

[0006] The process quality of the target process is predicted according to the process parameters through an uncertainty proxy model, wherein the uncertainty proxy model is a model based on chaotic polynomial expansion, and the polynomial coefficients in the uncertainty proxy model are obtained by the process quality prediction results obtained based on the process parameters predicted by a machine learning model.

[0007] In the above implementation process, the uncertainty proxy model is used to predict the process quality of the target process according to the process parameters. The uncertainty proxy model is a model based on chaotic polynomial expansion. The polynomial coefficients in the uncertainty proxy model are obtained by the process quality prediction results obtained by the machine learning model based on the process parameters. Since the machine learning model can quickly realize prediction and has higher timeliness, it is more conducive to online deployment to realize real-time prediction of process quality. Moreover, by realizing prediction through the uncertainty proxy model, the uncertainty in the process parameters can be fully considered, which can improve the accuracy of the prediction.

[0008] Optionally, before predicting the process quality of the target process according to the process parameters using the uncertainty proxy model, the method further includes:

[0009] Inputting the process parameters into the machine learning model to predict the process quality and obtain a process quality prediction result;

[0010] The polynomial coefficients of the uncertainty proxy model are determined according to the process quality prediction result.

[0011] In the above implementation process, since the machine learning model is a pre-trained model, it can obtain more accurate process quality prediction results. Therefore, more accurate polynomial coefficients can be obtained based on the process quality prediction results, which is beneficial to improving the prediction accuracy of the uncertainty model. The machine learning model can quickly obtain results and is more efficient in determining the polynomial coefficients, which is beneficial to online real-time prediction.

[0012] Optionally, inputting the process parameters into the machine learning model to predict the process quality and obtain a process quality prediction result includes:

[0013] Sampling the process parameters to obtain sampling point data;

[0014] The sampling point data is input into the machine learning model to predict the process quality, and the process quality prediction results corresponding to each sampling point data are obtained.

[0015] In the above implementation process, the process parameters are sampled and then input into the machine learning model for prediction, so that the process quality prediction results corresponding to the data at different sampling points can be predicted, which can provide more data for the subsequent determination of the polynomial coefficients.

[0016] Optionally, sampling the process parameters to obtain sampling point data includes:

[0017] The process parameters are sampled using Latin hypercube sampling to obtain sampling point data, so that fewer samples can be used when the same threshold is reached, thereby reducing the complexity of the calculation.

[0018] Optionally, constructing the polynomial coefficients of the uncertainty proxy model according to the process quality prediction result includes:

[0019] The least square method is used to fit the process quality prediction results corresponding to the data of each sampling point to obtain the polynomial coefficients of the uncertainty proxy model. The least square method has high stability and reliability and can improve more accurate and reliable parameter estimation.

[0020] Optionally, predicting the process quality of the target process according to the process parameters by using an uncertainty proxy model includes:

[0021] Obtaining multi-order moments of the process parameters;

[0022] Constructing an orthogonal basis corresponding to the uncertainty proxy model using the multi-order moments;

[0023] The process quality of the target process is obtained according to the orthogonal basis and the polynomial coefficients.

[0024] In the above implementation process, the uncertainty proxy model is constructed using the above method, so that the impact of parameter fluctuations on process quality prediction can be considered to improve the accuracy of the prediction.

[0025] Optionally, the machine learning model is constructed in the following manner:

[0026] Obtain the process parameters for training involved in various process processing;

[0027] Inputting the process parameters for training into the simulation model for simulation to obtain simulation data of the corresponding process;

[0028] The machine learning model is trained using the training process parameters and the simulation data to obtain a trained machine learning model.

[0029] In the above implementation process, the machine learning model is trained by fusing simulation data and process parameters, which can combine the advantages of high accuracy of the simulation model and further improve the training accuracy of the machine learning model.

[0030] Optionally, when the uncertainty proxy model performs online prediction, after predicting the process quality of the target process according to the process parameters by using the uncertainty proxy model, the method further includes:

[0031] Output corresponding prompt information according to the process quality, so that online prompts can be provided in real time.

[0032] Optionally, when the uncertainty proxy model performs offline prediction, after predicting the process quality of the target process according to the process parameters using the uncertainty proxy model, the method further includes:

[0033] The parameter optimization scheme corresponding to the process quality output is then used to perform parameter fluctuation analysis and fluctuation control, providing a basis for the design of process parameters.

[0034] In a second aspect, an embodiment of the present application provides a coefficient determination method, the method comprising:

[0035] Obtaining process parameters involved in target process processing;

[0036] Inputting the process parameters into a machine learning model to predict process quality, thereby obtaining a process quality prediction result;

[0037] The polynomial coefficients of the uncertainty proxy model are determined according to the process quality prediction result, wherein the uncertainty proxy model is a model based on chaotic polynomial expansion, and the uncertainty proxy model is used for process quality prediction.

[0038] In the above implementation process, the polynomial coefficients of the uncertainty proxy model are obtained through the process quality prediction results of the machine learning model. Since the machine learning model can achieve fast prediction, the polynomial coefficients of the uncertainty model can be quickly obtained. Then, when the uncertainty model is used to predict the process quality, the prediction results can be quickly obtained, which is more effective.

[0039] In a third aspect, an embodiment of the present application provides a process quality prediction device, the device comprising:

[0040] A parameter acquisition module is used to acquire process parameters involved in target process processing;

[0041] A prediction module is used to predict the process quality of the target process according to the process parameters through an uncertainty proxy model, wherein the uncertainty proxy model is a model based on chaotic polynomial expansion, and the polynomial coefficients in the uncertainty proxy model are obtained by the process quality prediction results obtained based on the process parameters predicted by the machine learning model.

[0042] In a fourth aspect, an embodiment of the present application provides a coefficient determination device, the device comprising:

[0043] An information acquisition module is used to obtain process parameters involved in target process processing;

[0044] A process quality prediction module, used for inputting the process parameters into a machine learning model to predict the process quality and obtain a process quality prediction result;

[0045] The coefficient determination module is used to determine the polynomial coefficients of the uncertainty proxy model according to the process quality prediction result, wherein the uncertainty proxy model is a model based on chaotic polynomial expansion, and the uncertainty proxy model is used for process quality prediction.

[0046] In a fifth aspect, an embodiment of the present application provides an electronic device, comprising a processor and a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the steps in the method provided in the first aspect are executed.

[0047] In a sixth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps in the method provided in the first aspect are performed.

[0048] Other features and advantages of the present application will be described in the following description, and partly become apparent from the description, or be understood by practicing the embodiments of the present application. The purpose and other advantages of the present application can be realized and obtained by the structures specifically pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0050] Figure 1 A flowchart of a process quality prediction method provided in an embodiment of the present application;

[0051] Figure 2 An overall detailed flow chart of a process quality prediction method provided in an embodiment of the present application;

[0052] Figure 3 A flowchart of a coefficient determination method provided in an embodiment of the present application;

[0053] Figure 4 A structural block diagram of a process quality prediction device provided in an embodiment of the present application;

[0054] Figure 5 A structural block diagram of a coefficient determination device provided in an embodiment of the present application;

[0055] Figure 6 A schematic diagram of the structure of an electronic device for executing a process quality prediction method or a coefficient determination method provided in an embodiment of the present application. DETAILED DESCRIPTION

[0056] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present application.

[0057] It should be noted that the terms "system" and "network" in the embodiments of the present invention can be used interchangeably. "Multiple" means two or more. In view of this, in the embodiments of the present invention, "multiple" can also be understood as "at least two". "And / or" describes the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / ", unless otherwise specified, generally indicates that the previously associated objects are in an "or" relationship.

[0058] In the related art, in order to realize the prediction of process quality, a simulation model is generally used for simulation. However, due to the complexity of simulation model calculation and poor timeliness, it cannot meet the requirements of real-time online prediction.

[0059] Based on this, an embodiment of the present application provides a process quality prediction method, which predicts the process quality of a target process according to process parameters through an uncertainty proxy model. The uncertainty proxy model is a model based on chaotic polynomial expansion. The polynomial coefficients in the uncertainty proxy model are obtained based on the process quality prediction results obtained by the machine learning model based on the process parameters. Since the machine learning model can quickly realize prediction and has higher timeliness, it is more conducive to online deployment for real-time prediction of process quality. Moreover, by realizing prediction through the uncertainty proxy model, the uncertainty in the process parameters can also be fully considered, which can improve the accuracy of the prediction.

[0060] Please refer to Figure 1 , Figure 1 A flowchart of a process quality prediction method provided in an embodiment of the present application, the method comprising the following steps:

[0061] Step S110: obtaining process parameters involved in target process processing.

[0062] In some application scenarios, the target process may be a process involved in the processing of equipment (such as vehicles), such as welding process, stamping process, painting process, planing process, etc. The target process can be understood as a process in which process quality prediction is required, and the process parameters involved refer to the process parameters required in the processing process, such as information about the processed materials and environmental parameters, etc. For example, for welding process, the process parameters may include the plate thickness, size, shape of the welded plate, and the temperature, pressure, welding voltage, current, welding speed, welding angle, processing time and other parameters required in the welding process.

[0063] It is understandable that different processes may involve different process parameters, and changes in these process parameters have a certain impact on the process quality of the process, which can be based on the specific process parameters used in the actual production process.

[0064] The execution subject of the process quality prediction method of this scheme can be a prediction system, and the process parameters can be input into the prediction system by the user himself, or obtained by the prediction system from the processing system of the actual processing target process. For example, before processing the target process, the corresponding process parameters need to be set in the processing system. Later in the processing process, each processing equipment in the processing system can be processed according to the set process parameters, such as welding speed, welding temperature, etc. Of course, before processing, it is necessary to predict the process quality of the target process processed under the currently set process parameters, so as to know in advance whether the process quality processed under the process parameters is qualified. Therefore, the prediction system can first obtain the corresponding process parameters from the processing system, or the user can input the corresponding processing parameters in the prediction system to predict the process quality.

[0065] Step S120: predicting the process quality of the target process according to the process parameters using the uncertainty proxy model.

[0066] Among them, the uncertainty proxy model is a model based on chaotic polynomial expansion, and the polynomial coefficients in the uncertainty proxy model are obtained by the process quality prediction results obtained based on the process parameter prediction according to the machine learning model.

[0067] The uncertainty proxy model can be understood as a method for describing and analyzing the uncertainty of random variables. It can be based on the chaotic polynomial theory and can accurately describe the randomness of random variables of any distribution form by expanding the random variables into a linear combination of a series of polynomial functions.

[0068] In the uncertainty proxy model based on chaotic polynomial expansion, it is necessary to determine the polynomial coefficients in the uncertainty proxy model. In this scheme, the polynomial coefficients are obtained according to the process quality prediction results obtained by the machine learning model based on the process parameter prediction.

[0069] Since the machine learning model is more timely, the process quality prediction results can be quickly predicted based on the process parameters. In this way, the uncertainty proxy model and the machine learning model can be deployed online to achieve real-time prediction of process quality.

[0070] In addition, the polynomial coefficients are obtained by using the process quality prediction results obtained through the machine learning model, so that the probability distribution function of the process parameters can be obtained, thereby describing the uncertainty of the process parameters, and then analyzing the impact of the uncertainty of the process parameters on the output results.

[0071] In some embodiments, the machine learning model may be a linear regression model, a support vector machine model, a random forest model, a decision tree model, a neural network model, etc. In practical applications, a suitable type of machine learning model may be selected according to actual needs.

[0072] In the above implementation process, the uncertainty proxy model is used to predict the process quality of the target process according to the process parameters. The uncertainty proxy model is a model based on chaotic polynomial expansion. The polynomial coefficients in the uncertainty proxy model are obtained by the process quality prediction results obtained by the machine learning model based on the process parameters. Since the machine learning model can quickly realize prediction and has higher timeliness, it is more conducive to online deployment to realize real-time prediction of process quality. Moreover, by realizing prediction through the uncertainty proxy model, the uncertainty in the process parameters can be fully considered, which can improve the accuracy of the prediction.

[0073] Based on the above embodiment, after obtaining the process parameters, the process parameters can be first input into the machine learning model to predict the process quality to obtain the process quality prediction results, and then the polynomial coefficients of the uncertainty proxy model are determined according to the process quality prediction results.

[0074] It can be understood that the machine learning model can be pre-trained, and a large number of process parameters involved in different processes are used for training. Therefore, by inputting the process parameters of the target process into the machine learning model for prediction, a more accurate process quality prediction result can be obtained.

[0075] Since the machine learning model can quickly predict and obtain process quality prediction results, the polynomial coefficients of the uncertainty proxy model can be quickly obtained based on the process quality prediction results, thereby meeting the requirements of real-time online prediction.

[0076] Understandably, the machine learning model is a deterministic model that cannot describe the impact of process parameter fluctuations on process quality and is difficult to solve the occasional quality defects in the current process. Therefore, by using the uncertainty proxy model for prediction, we can obtain the impact of process parameter fluctuations on process quality and reduce the "dimensionality curse" problem caused by too many process parameters.

[0077] In the above implementation process, since the machine learning model is a pre-trained model, it can obtain more accurate process quality prediction results. Therefore, more accurate polynomial coefficients can be obtained based on the process quality prediction results, which is beneficial to improving the prediction accuracy of the uncertainty model. The machine learning model can quickly obtain results and is more efficient in determining the polynomial coefficients, which is beneficial to online real-time prediction.

[0078] On the basis of the above embodiment, since the amount of data of the acquired process parameters is limited, in order to obtain more process quality prediction results corresponding to the process parameters and then construct more accurate polynomial coefficients, the process parameters can be sampled in this embodiment to obtain sampling point data, and then the sampling point data can be input into the machine learning model to predict the process quality and obtain the process quality prediction results corresponding to each sampling point data.

[0079] The process parameters can be sampled randomly or at certain intervals. For example, if the input process parameters include a temperature of 120°C, then during sampling, the data point of 130°C may be sampled. In this way, some data points that were not originally included in the process parameters can be obtained, and the process quality of these data points can then be predicted.

[0080] In the above implementation process, the process parameters are sampled and then input into the machine learning model for prediction, so that the process quality prediction results corresponding to the data at different sampling points can be predicted, which can provide more data for the subsequent determination of the polynomial coefficients.

[0081] On the basis of the above embodiment, the Latin hypercube sampling method can be used to sample the process parameters to obtain sampling point data.

[0082] Among them, the Latin hypercube sampling method is a method of approximate random sampling from multivariate parameters, which belongs to the stratified sampling technology. The specific implementation process includes: dividing each variable (such as each process parameter) into several intervals with the same probability, and then selecting sample points that meet the Latin hypercube conditions to ensure that there is only one sample in each row and column. According to the uniform distribution, a number is randomly generated in each small interval, the order of the random numbers is disrupted, and the randomly distributed values ​​are generated according to the inverse function of the probability distribution function.

[0083] In the above implementation process, Latin hypercube sampling is used to use fewer samples when reaching the same threshold, while reducing the complexity of calculation.

[0084] On the basis of the above embodiment, when determining the polynomial coefficients of the uncertainty proxy model according to the process quality prediction results, the least squares method can be used to fit the process quality prediction results corresponding to each sampling point data to obtain the polynomial coefficients of the uncertainty proxy model.

[0085] Among them, the least squares method is to find the best function matching of the data by minimizing the sum of squares of the errors. Assume that the uncertainty proxy model is expressed as follows:

[0086]

[0087] Among them, k ranges from 0 to n-1, cx represents the polynomial coefficients, p k (x) represents a k-order orthogonal basis, and x represents a process parameter.

[0088] Among them, c x It is expressed as follows:

[0089] in, It represents the process quality prediction results corresponding to the data of each sampling point.

[0090] In this way, the polynomial coefficients in the uncertainty proxy model can be determined, and then when predicting the process quality, the process parameter x is substituted into the above uncertainty proxy model to obtain the process quality.

[0091] In the above implementation process, the least square method has high stability and reliability, and can provide more accurate and reliable parameter estimation.

[0092] On the basis of the above embodiments, when constructing an uncertainty proxy model, it is also necessary to first determine other parameters therein, such as first obtaining the multi-order moments of the process parameters, and then using the multi-order moments to construct the orthogonal basis corresponding to the uncertainty proxy model, and then obtaining the process quality of the target process based on the orthogonal basis and polynomial coefficients.

[0093] For example, statistical process parameters x1, x2, ..., x m The first 2n-1 moments of the polynomial, where n is the highest order of the polynomial, and the kth moment of the process parameter x Where ρ(x) represents the probability distribution function, x i is the i-th sampling point data of process parameter x, and N is the number of sampling points of random variable x.

[0094] A set of orthogonal bases in any given probability space are constructed using the moments of the process parameters. The k-th order polynomial basis of the process parameter x is in YesP (k) The coefficients of (x), according to the properties of the orthogonal basis, any two basis functions satisfy: Among them, P (k) (x) and P (j) (x) are the kth and jth order polynomial bases, δ kj is a non-zero constant, and the sequence number k, j = 0, 1, ...

[0095] Then the leading coefficients of all polynomial bases are normalized and a k-order polynomial base P is constructed. (k) (x) and all polynomial bases with order less than k, that is, to construct an orthogonal basis, whose constraints satisfy:

[0096] According to the definition of moment, the above formula can be reduced to:

[0097] Written in matrix form: That is, the first 2k-1 moments are required to construct the first k-order polynomial basis.

[0098] The Latin hypercube sampling method is used to sample the specific values ​​of the process parameters to obtain the number of sampling point data. Among them, m is the dimension of the random variable x, and n is the highest order of the polynomial. Then these multiple sampling point data are input into the machine learning model to predict the process quality prediction results corresponding to each sampling point data. The above-mentioned polynomial coefficients can be obtained by using the process quality prediction results, and then the uncertainty proxy model can be constructed using the polynomial coefficients and orthogonal basis.

[0099] In the above implementation process, the uncertainty proxy model is constructed using the above method, so that the impact of parameter fluctuations on process quality prediction can be considered to improve the accuracy of the prediction.

[0100] On the basis of the above embodiments, since the process parameters involved in different processes generally cover a narrow range, in order to improve the prediction accuracy of the machine learning model, the process parameters can also be supplemented by simulation data. When training the machine learning model, first obtain the training process parameters involved in various process processing, then input the training process parameters into the simulation model for simulation, obtain the simulation data of the corresponding process, and then use the training process parameters and simulation data to train the machine learning model to obtain a trained machine learning model.

[0101] Since the simulation models in the process are generally more complex, require high computing resources, and have a long test cycle, if the simulation results of the simulation model (i.e., the process quality prediction results) are directly used to determine the polynomial coefficients of the uncertainty model, it will be difficult to meet the requirements of real-time online prediction. Therefore, in order to combine the advantages of high accuracy of the simulation model and the ability to obtain more data, the simulation data is used to train the machine learning model in this scheme, which can improve the training accuracy of the machine learning model.

[0102] Among them, the process parameters used for training refer to the process parameters used during training. These process parameters can be input into the simulation model for simulation to observe the relationship between the process quality and the process parameters. The output of the simulation model is simulation data, which may include process parameters and process quality results. The simulation data can be used as a supplement to the process parameters, that is, the simulation data may include other data different from the process parameters used for training. In this way, the process parameters and simulation data can be integrated to train the machine learning model. During training, the label data used is the process quality prediction result.

[0103] The process parameters used for training can be obtained by collecting the process in the actual production process. After the simulation model obtains the simulation data, the process quality results in the simulation data can be compared with the corresponding set process quality results to verify the accuracy of the simulation model. If it is inaccurate, it can be returned to the simulation model to correct the simulation model. After it is accurate, the simulation data and process parameters can be integrated as training data for training the machine learning model.

[0104] It is understandable that when the machine learning model is a different model, the training process is different and will not be described in detail here.

[0105] In addition, during the training process of the machine learning model, in order to better extract the features in the input data, the simulation data and the process data used for training can be cleaned first, such as cleaning of error values, missing values ​​and outliers, and then feature engineering can be used to extract features from the cleaned data. The extracted features are then input into the machine learning model for training.

[0106] In the above implementation process, the machine learning model is trained by fusing simulation data and process parameters, which can combine the advantages of high accuracy of the simulation model and further improve the training accuracy of the machine learning model.

[0107] Based on the above embodiments, the machine learning model can perform online real-time prediction, so the uncertainty proxy model can also be deployed online. When the uncertainty proxy model performs online prediction, after obtaining the process quality of the target process, the corresponding prompt information can be output according to the process quality.

[0108] After obtaining the process quality output by the uncertainty proxy model, corresponding prompt information can be output according to different process qualities. The process quality output by the uncertainty proxy model can be a specific value, which can be used to reflect the corresponding process quality. For example, data within different numerical ranges represent the excellence of its process quality. The process quality can be divided into multiple levels from excellent to poor. In this way, the corresponding process quality can be determined according to the data range of the data output by the uncertainty proxy model, and different prompt information can be output for different process qualities. For example, if the process quality is excellent, the output includes prompt information about the process quality and the ability to continue processing. If the process quality is poor, the output includes prompt information about the process quality and the need to suspend further processing.

[0109] The output prompt information can be fed back to the processing system, so that the processing system can know the process quality corresponding to the currently set process parameters before processing, and then decide whether to continue processing.

[0110] On the basis of the above-mentioned embodiment, the uncertainty proxy model can also perform offline prediction. When performing offline prediction, after obtaining the process quality, the corresponding parameter optimization solution can be output according to the process quality.

[0111] Since the uncertainty proxy model can analyze the impact of fluctuations in process parameters on process quality, it is possible to analyze the parameter range under which the process quality is excellent. The parameter optimization plan can include information such as how to adjust the current process parameters and the range of adjustment. Therefore, if the process quality is poor under the process parameters of the current target process, the corresponding process parameters can be adjusted according to the obtained parameter optimization plan. The parameter optimization plan can be fed back to the processing system, and the processing system can adjust the corresponding process parameters according to the parameter optimization plan, thereby achieving the purpose of controllable process quality.

[0112] It can be understood that when the uncertainty proxy model is deployed offline, different process parameters of different processes can be input for prediction. In this way, the process quality under different parameters can be observed, and the impact of parameter fluctuations on process quality can be obtained. In this way, the process quality can be improved and a basis for the design of process parameters can be provided.

[0113] The specific implementation flow chart of the embodiment of this application can be referred to Figure 2 As shown, the process parameters are first collected and input into the simulation model. After the obtained simulation data is integrated with the process parameters, the machine learning model is trained to obtain a trained machine learning model. The currently collected process parameters can be input into the uncertainty model for prediction. For example, the multi-order moments are first obtained, and then the orthogonal basis is calculated. Then, the polynomial coefficients of the uncertainty model are constructed through the prediction results of the machine learning model. The process quality can be predicted based on the orthogonal basis and the polynomial coefficients. The uncertainty proxy model can be used for online real-time prediction, process robust design optimization, and parameter fluctuation analysis and control.

[0114] Please refer to Figure 3 , Figure 3 A flow chart of a coefficient determination method provided in an embodiment of the present application includes the following steps:

[0115] Step S210: Acquire process parameters involved in target process processing.

[0116] Step S220: input the process parameters into the machine learning model to predict the process quality and obtain a process quality prediction result.

[0117] Step S230: determining the polynomial coefficients of the uncertainty proxy model according to the process quality prediction result, wherein the uncertainty proxy model is a model based on chaotic polynomial expansion, and the uncertainty proxy model is used for process quality prediction.

[0118] In this implementation method, the polynomial coefficients of the uncertainty proxy model can be obtained by deploying a machine learning model online. Since the machine learning model can achieve rapid prediction, the polynomial coefficients of the uncertainty model can be quickly obtained, and then when the uncertainty model is used to predict the process quality, the prediction results can be quickly obtained, which is more effective. Subsequently, before processing the target process, the obtained process parameters can be input into the machine learning model and the uncertainty proxy model respectively, and finally the corresponding process quality can be obtained through the uncertainty proxy model, which can realize real-time online prediction of process quality with high timeliness. It can be understood that the specific implementation process of obtaining polynomial coefficients and using uncertainty models to predict process quality in this method can refer to the relevant description in the above embodiment. For the sake of brevity, it will not be repeated here.

[0119] Please refer to Figure 4 , Figure 4 The structure block diagram of a process quality prediction device 300 provided in an embodiment of the present application, the device 300 may be a module, program segment or code on an electronic device. It should be understood that the device 300 is similar to the above Figure 1 The method embodiment corresponds to and can be executed Figure 1 The various steps involved in the method embodiment and the specific functions of the device 300 can be found in the above description. To avoid repetition, the detailed description is appropriately omitted here.

[0120] Optionally, the device 300 includes:

[0121] The parameter acquisition module 310 is used to acquire the process parameters involved in the target process processing;

[0122] The prediction module 320 is used to predict the process quality of the target process according to the process parameters through an uncertainty proxy model, wherein the uncertainty proxy model is a model based on chaotic polynomial expansion, and the polynomial coefficients in the uncertainty proxy model are obtained by the process quality prediction results obtained based on the process parameters predicted by the machine learning model.

[0123] Optionally, the prediction module 320 is further used to input the process parameters into the machine learning model to predict the process quality and obtain a process quality prediction result; and determine the polynomial coefficients of the uncertainty proxy model based on the process quality prediction result.

[0124] Optionally, the prediction module 320 is used to sample the process parameters to obtain sampling point data; input the sampling point data into the machine learning model to predict the process quality, and obtain the process quality prediction results corresponding to each sampling point data.

[0125] Optionally, the prediction module 320 is used to sample the process parameters using a Latin hypercube sampling method to obtain sampling point data.

[0126] Optionally, the prediction module 320 is used to fit the process quality prediction result corresponding to each sampling point data using the least square method to obtain the polynomial coefficients of the uncertainty proxy model.

[0127] Optionally, the prediction module 320 is used to obtain multi-order moments of the process parameters; construct an orthogonal basis corresponding to the uncertainty proxy model using the multi-order moments; and obtain the process quality of the target process based on the orthogonal basis and the polynomial coefficients.

[0128] Optionally, the machine learning model is constructed in the following manner:

[0129] Obtain the process parameters for training involved in various process processing;

[0130] Inputting the process parameters for training into the simulation model for simulation to obtain simulation data of the corresponding process;

[0131] The machine learning model is trained using the training process parameters and the simulation data to obtain a trained machine learning model.

[0132] Optionally, when the uncertainty proxy model performs online prediction, the device 300 further includes:

[0133] The information output module is used to output corresponding prompt information according to the process quality.

[0134] Optionally, when the uncertainty proxy model performs offline prediction, the device 300 further includes:

[0135] The information output module is used to output a corresponding parameter optimization solution according to the process quality.

[0136] Please refer to Figure 5 , Figure 5 4 is a structural block diagram of a coefficient determination device 400 provided in an embodiment of the present application. The device 400 may be a module, a program segment or a code on an electronic device. It should be understood that the device 400 is similar to the above Figure 3 The method embodiment corresponds to and can be executed Figure 3 The various steps involved in the method embodiment and the specific functions of the device 400 can be found in the above description. To avoid repetition, the detailed description is appropriately omitted here.

[0137] Optionally, the device 400 includes:

[0138] The information acquisition module 410 is used to acquire the process parameters involved in the target process processing;

[0139] A process quality prediction module 420 is used to input the process parameters into a machine learning model to predict the process quality and obtain a process quality prediction result;

[0140] The coefficient determination module 430 is used to determine the polynomial coefficients of the uncertainty proxy model according to the process quality prediction result, wherein the uncertainty proxy model is a model based on chaotic polynomial expansion, and the uncertainty proxy model is used for process quality prediction.

[0141] It should be noted that those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0142] Please refer to Figure 6 , Figure 6 A schematic diagram of the structure of an electronic device for executing a process quality prediction method or a coefficient determination method provided in an embodiment of the present application, the electronic device may include: at least one processor 510, such as a CPU, at least one communication interface 520, at least one memory 530 and at least one communication bus 540. Among them, the communication bus 540 is used to realize the connection and communication between these components. Among them, the communication interface 520 of the device in the embodiment of the present application is used to communicate signaling or data with other node devices. The memory 530 can be a high-speed RAM memory, or it can be a non-volatile memory (non-volatile memory), such as at least one disk storage. The memory 530 can optionally also be at least one storage device located away from the aforementioned processor. Computer-readable instructions are stored in the memory 530. When the computer-readable instructions are executed by the processor 510, the electronic device executes the above-mentioned Figure 1 or Figure 3 The method process is shown.

[0143] Understandably, Figure 6 The structure shown is for illustration only, and the electronic device may also include Figure 6 More or fewer components as shown, or with Figure 6 Different configurations shown. Figure 6 Each component shown in the figure can be implemented by hardware, software or a combination thereof.

[0144] The present application embodiment provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the following is performed: Figure 1 or Figure 3The method process in the method embodiment shown is executed by the electronic device.

[0145] This embodiment discloses a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions. When the program instructions are executed by a computer, the computer can perform the methods provided by the above method embodiments, for example, including:

[0146] Obtaining process parameters involved in target process processing;

[0147] The process quality of the target process is predicted according to the process parameters through an uncertainty proxy model, wherein the uncertainty proxy model is a model based on chaotic polynomial expansion, and the polynomial coefficients in the uncertainty proxy model are obtained by the process quality prediction results obtained based on the process parameters predicted by a machine learning model.

[0148] In summary, the embodiments of the present application provide a process quality prediction method, a coefficient determination method, a device and an electronic device. The method predicts the process quality of a target process according to process parameters through an uncertainty proxy model. The uncertainty proxy model is a model based on chaotic polynomial expansion. The polynomial coefficients in the uncertainty proxy model are obtained according to the process quality prediction results obtained by the machine learning model based on the process parameter prediction. Since the machine learning model can quickly realize prediction and has higher timeliness, it is more conducive to online deployment to realize real-time prediction of process quality. Moreover, by realizing prediction through the uncertainty proxy model, the uncertainty in the process parameters can also be fully considered, which can improve the accuracy of the prediction.

[0149] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0150] In addition, the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0151] Furthermore, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0152] In this document, relational terms such as first and second, etc. are used merely to distinguish one entity or operation from another entity or operation, but do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0153] The above description is only an embodiment of the present application and is not intended to limit the protection scope of the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A process quality prediction method, characterized in that: The method comprises: Obtaining process parameters involved in target process processing; The process quality of the target process is predicted according to the process parameters through an uncertainty proxy model, wherein the uncertainty proxy model is a model based on chaotic polynomial expansion, and the polynomial coefficients in the uncertainty proxy model are obtained by the process quality prediction results obtained based on the process parameters predicted by a machine learning model.

2. The method according to claim 1, characterized in that Before predicting the process quality of the target process according to the process parameters by using the uncertainty proxy model, the method further includes: Inputting the process parameters into the machine learning model to predict the process quality and obtain a process quality prediction result; The polynomial coefficients of the uncertainty proxy model are determined according to the process quality prediction result.

3. The method according to claim 2, characterized in that The step of inputting the process parameters into the machine learning model to predict the process quality and obtaining the process quality prediction result includes: Sampling the process parameters to obtain sampling point data; The sampling point data is input into the machine learning model to predict the process quality, and the process quality prediction results corresponding to each sampling point data are obtained.

4. The method according to claim 3, characterized in that The step of sampling the process parameters to obtain sampling point data includes: The process parameters are sampled using a Latin hypercube sampling method to obtain sampling point data.

5. The method according to claim 3, characterized in that: The polynomial coefficients of the uncertainty proxy model are constructed according to the process quality prediction result, including: The least square method is used to fit the process quality prediction results corresponding to the data of each sampling point to obtain the polynomial coefficients of the uncertainty proxy model.

6. The method according to claim 1, characterized in that The predicting the process quality of the target process according to the process parameters by using an uncertainty proxy model includes: Obtaining multi-order moments of the process parameters; Constructing an orthogonal basis corresponding to the uncertainty proxy model using the multi-order moments; The process quality of the target process is obtained according to the orthogonal basis and the polynomial coefficients.

7. The method according to claim 1, characterized in that The machine learning model is constructed in the following way: Obtain the process parameters for training involved in various process processing; Inputting the process parameters for training into the simulation model for simulation to obtain simulation data of the corresponding process; The machine learning model is trained using the training process parameters and the simulation data to obtain a trained machine learning model.

8. The method according to any one of claims 1 to 7, characterized in that: When the uncertainty proxy model performs online prediction, after predicting the process quality of the target process according to the process parameters by using the uncertainty proxy model, the method further includes: Output corresponding prompt information according to the process quality.

9. The method according to any one of claims 1 to 7, characterized in that: When the uncertainty proxy model performs offline prediction, after predicting the process quality of the target process according to the process parameters by using the uncertainty proxy model, the method further includes: Output a corresponding parameter optimization solution according to the process quality.

10. A coefficient determination method, characterized in that: The method comprises: Obtaining process parameters involved in target process processing; Inputting the process parameters into a machine learning model to predict process quality, thereby obtaining a process quality prediction result; The polynomial coefficients of the uncertainty proxy model are determined according to the process quality prediction result, wherein the uncertainty proxy model is a model based on chaotic polynomial expansion, and the uncertainty proxy model is used for process quality prediction.

11. A process quality prediction device, characterized in that: The device comprises: A parameter acquisition module is used to acquire process parameters involved in target process processing; A prediction module is used to predict the process quality of the target process according to the process parameters through an uncertainty proxy model, wherein the uncertainty proxy model is a model based on chaotic polynomial expansion, and the polynomial coefficients in the uncertainty proxy model are obtained by the process quality prediction results obtained based on the process parameters predicted by the machine learning model.

12. A coefficient determination device, characterized in that: The device comprises: An information acquisition module is used to obtain process parameters involved in target process processing; A process quality prediction module, used for inputting the process parameters into a machine learning model to predict the process quality and obtain a process quality prediction result; A coefficient determination module is used to determine the polynomial coefficients of the uncertainty proxy model according to the process quality prediction result, wherein the uncertainty proxy model is a model based on chaotic polynomial expansion, and the uncertainty proxy model is used for process quality prediction.

13. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, the method according to any one of claims 1 to 10 is executed.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 10 is executed.

Images