Process parameter time sequence registration method based on time sequence incidence matrix and improved particle swarm optimization

Abstract

The invention discloses a process parameter time sequence registration method based on a time sequence incidence matrix and an improved particle swarm algorithm, and the method comprises the steps: carrying out the preprocessing of each batch of flow process time sequence data, and obtaining the preprocessed flow process time sequence data of each batch; constructing a regular feature screening method based on a time sequence incidence matrix, and screening out a process parameter set obviously influenced by time delay; obtaining a screened process parameter time sequence incidence matrix according to the screened process parameter set; optimizing the screened process parameter time sequence incidence matrix according to an improved particle swarm algorithm, and searching an optimal lag time solution set of the process parameters; and according to the optimal lag time solution set of the process parameters, performing backward translation on the screened process parameter time sequences obviously influenced by time delay for corresponding optimal lag time to realize accurate alignment with the quality indexes on the time sequences, and obtaining registered data. The method provides important support for guaranteeing the process quality and improving the processing efficiency for a process manufacturing enterprise.

Description

Technical Field

[0001] The invention relates to a process parameter time series registration method based on a time series association matrix and an improved particle swarm algorithm, and belongs to the technical field of data processing. Background Art

[0002] Process industry is an indispensable part of modern manufacturing industry and an important pillar industry for the current national economic and social development. In the actual process of process industry production, process data is usually collected by sensors in real time and recorded synchronously at fixed time intervals to form continuous time series data. However, since the production process involves complex physical and chemical reactions, changes in process parameters often cannot be immediately reflected in quality indicators, resulting in a time delay effect between the two. This delay effect causes the input of the production system to be unable to be reflected in the output in a timely manner, resulting in decisions based on distorted data, which increases potential risks. Therefore, effectively eliminating the time delay effect is crucial to improving the accuracy and reliability of production decisions. Summary of the invention

[0003] The present invention provides a process parameter timing registration method based on a timing association matrix and an improved particle swarm algorithm, which is used to solve the time delay effect existing in a process manufacturing process.

[0004] The technical solution of the present invention is:

[0005] According to a first aspect of the present invention, a method for timing alignment of process parameters based on a timing association matrix and an improved particle swarm algorithm is provided, wherein the method comprises the following steps:

[0006] S1. Obtain process technology time series data of different batches of process production lines; preprocess each batch of process technology time series data to obtain the preprocessed process technology time series data of each batch; wherein the preprocessed process technology time series data is constructed by M process parameters and one quality indicator;

[0007] S2. Constructing a regularity feature screening method based on a time series correlation matrix to obtain a time series correlation matrix of process parameters, and identifying and screening a set of process parameters significantly affected by time delay according to the time series correlation matrix of process parameters;

[0008] S3. Obtaining a screened process parameter time series correlation matrix based on the screened process parameter set;

[0009] S4, optimizing the screened process parameter time series correlation matrix based on the improved particle swarm algorithm to find the optimal lag time solution set of the process parameters;

[0010] S5. Based on the optimal lag time solution set of process parameters, the time series of each process parameter significantly affected by time delay is shifted backward to the corresponding optimal lag time to achieve precise alignment with the quality index in the time series, and obtain the registered data.

[0011] Furthermore, the S2 includes:

[0012] S2.1. Constructing a time series correlation matrix of process parameters: Calculate the correlation coefficients of the time series relative quality indicators of each process parameter in the process time series data after preprocessing of each batch at different lag times, so as to construct a time series correlation matrix of process parameters;

[0013] S2.2. Use the regular feature screening method to analyze the time series correlation matrix of the quantitative process parameters, so as to screen out the set of process parameters that are significantly affected by the time delay effect.

[0014] Furthermore, the regularity feature screening method includes: obtaining the delay correlation coefficient of each process parameter at different lag times based on the timing correlation matrix of the process parameters; obtaining the average delay correlation coefficient of each process parameter based on the delay correlation coefficient of each process parameter at different lag times; making a judgment based on the average delay correlation coefficient of each process parameter and a preset threshold: retaining the current process parameter when the average delay correlation coefficient of the current process parameter is less than the preset threshold; and obtaining a process parameter set based on the retained process parameters.

[0015] Furthermore, the delay correlation coefficients of the process parameters at different lag times and the average delay correlation coefficients of the process parameters are expressed as follows:

[0016]

[0017] Among them, G i (t l ) is the lag time t l The time delay correlation coefficient of the ith process parameter at sec; N represents the total batch; f i,q (t l ) is the lag time t l The correlation coefficient corresponding to the i-th process parameter of the q-th batch at second, The lag time is t l The average value of the mutual correlation coefficient of the i-th process parameter at second; G i is the average delay correlation coefficient of the i-th process parameter.

[0018] Furthermore, the mutual correlation coefficient of the screened process parameters and quality indicator data that are significantly affected by time delay is taken as the fitness, and the maximum mutual correlation coefficient is taken as the goal. The improved particle swarm optimization algorithm is used to perform global optimization and search for the global optimal lag time, thereby obtaining the optimal lag time solution set of the process parameter time series.

[0019] According to a second aspect of the present invention, a process parameter timing alignment system based on a timing correlation matrix and an improved particle swarm algorithm is provided, comprising a module of any one of the above methods.

[0020] According to a third aspect of the present invention, a processor is provided, the processor being used to run a program, wherein the program, when running, executes any one of the above-mentioned process parameter timing registration methods based on a timing correlation matrix and an improved particle swarm algorithm.

[0021] According to a fourth aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium comprising a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned process parameter timing alignment methods based on the timing correlation matrix and the improved particle swarm algorithm.

[0022] The beneficial effects of the present invention are as follows: the present invention identifies and screens out process parameters affected by the time delay effect through a regularity feature screening method based on the time series association matrix, thereby ensuring the effective extraction of key parameters; further adopts an improved particle swarm algorithm to eliminate the influence of process errors and reduce the complexity of the time series association matrix, thereby quickly finding the optimal lag time solution set and achieving accurate alignment of process parameters and quality indicators in the time series. In addition, through the verification of a process production line data, the feasibility and effectiveness of the method of the present invention in time series alignment are proved, which provides important support for process manufacturing enterprises to ensure process quality and improve processing efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a schematic diagram of the process structure of the present invention;

[0024] Figure 2 It is a structural schematic diagram of the regularity feature screening method based on the time series association matrix of the present invention;

[0025] Figure 3 It is a structural schematic diagram of the present invention based on the improved particle swarm algorithm. DETAILED DESCRIPTION

[0026] In order to make the purpose, technical scheme and advantages of the embodiments of the present invention clearer, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. It should be noted that the embodiments in this application and the features in the embodiments can be combined with each other arbitrarily without conflict.

[0027] Example 1: Figure 1-3 As shown, according to a first aspect of an embodiment of the present invention, a method for timing registration of process parameters based on a timing association matrix and an improved particle swarm algorithm is provided, comprising the following steps:

[0028] S1. Obtain process technology time series data of different batches of process production lines; preprocess each batch of process technology time series data to obtain the preprocessed process technology time series data of each batch; wherein the preprocessed process technology time series data is constructed by M process parameters and one quality indicator; it should be noted that the preprocessing is specifically to remove the material head and material tail operation, that is, retain the data between the critical positions of the material head and the material tail;

[0029] S2. Constructing a regularity feature screening method based on a time series correlation matrix to obtain a time series correlation matrix of process parameters, and identifying and screening a set of process parameters significantly affected by time delay according to the time series correlation matrix of process parameters;

[0030] S3. Obtaining a screened process parameter time series correlation matrix based on the screened process parameter set;

[0031] S4. Optimize the screened process parameter timing association matrix based on the improved particle swarm algorithm to find the optimal lag time solution set of the process parameters; that is, use the improved particle swarm algorithm to obtain the optimal lag time of each process parameter respectively, and establish the optimal lag time solution set of the process parameters based on the optimal lag time of each process parameter, which is the optimal lag time set of all process parameters.

[0032] S5. Based on the optimal lag time solution set of process parameters, the time series of each process parameter significantly affected by time delay is shifted backward to the corresponding optimal lag time to achieve precise alignment with the quality index in the time series, and obtain the registered data.

[0033] Furthermore, after the registered data are registered, the missing sequences / redundant sequences are removed based on the quality index to obtain the registered sequence; or the registered data are completed using the LSTM preprocessing model.

[0034] Furthermore, the S2 includes:

[0035] S2.1. Constructing a time series correlation matrix of process parameters: Calculate the correlation coefficients of the time series relative quality indicators of each process parameter in the process time series data after preprocessing of each batch at different lag times, so as to construct a time series correlation matrix of process parameters; Move the time series relative quality indicators of the process parameters backward by h seconds (in the embodiment of the present invention, h = 1, 2, ..., 360, which means a lag of h seconds), and the expression of the established correlation coefficient R (h) is as follows:

[0036]

[0037] in, It means that the mth process parameter of the nth batch collected at time t is shifted backward by h seconds. represents the quality index of the nth batch collected at time t; express joint expectations; express expectations; express expectations; express The standard deviation of express It should be noted that when h is 0s, it indicates that there is no translation, that is, the corresponding positions of the time series of process parameters and quality indicators are the same;

[0038] like Figure 2 As shown in the figure, the original time series represents the original process parameter time series of a batch, X1, X2,…, X M represents the 1st, 2nd, and Mth process parameter time series, Y represents the quality index time series, and the correlation coefficients of the 1st process parameter time series relative to the quality index at different lag times are used for explanation, that is, the 1st process parameter time series represented by the 1st column is moved downward by the required time (i.e., the lag time), and then the correlation coefficients of the 1st process parameter time series relative to the quality index after the movement are calculated, and similarly, the correlation coefficients of the 2nd to Mth process parameter time series relative to the quality index after the movement are calculated to construct the time series correlation matrix of the process parameters; for example, the expression is as follows:

[0039]

[0040] Among them, f 1,n (t T ) is the lag time t T The correlation coefficient corresponding to the first process parameter of the nth batch at second; f M,n (t T ) is the lag time t TThe mutual correlation coefficient corresponding to the Mth process parameter of the nth batch at second; in the above matrix, from the 1st column to the Tth column, that is, the time series correlation matrix of the 1st process parameter, followed by the time series correlation matrix of the 2nd to Mth process parameters;

[0041] S2.2, using the regularity feature screening method to analyze the time series correlation matrix of the quantitative process parameters, thereby screening out the process parameter set significantly affected by the time delay effect. The present invention uses the regularity feature screening method to screen out the process parameter set significantly affected by the time delay effect, which helps to remove irrelevant factors in the process data that are not related to the processing time delay effect.

[0042] Furthermore, the regularity feature screening method comprises:

[0043] According to the time series correlation matrix of process parameters, the time delay correlation coefficient of each process parameter under different lag times is obtained;

[0044] According to the time delay correlation coefficients of each process parameter at different lag times, the average time delay correlation coefficient of each process parameter is obtained;

[0045] A judgment is made based on the average delay correlation coefficient of each process parameter and a preset threshold: if the average delay correlation coefficient of the current process parameter is less than the preset threshold, the current process parameter is retained; based on the retained process parameter, a process parameter set is obtained. In an embodiment of the present invention, a preset threshold is set to 0.041887, that is, the process parameters corresponding to the average delay correlation coefficient of the process parameter greater than or equal to 0.041887 are eliminated; the preset threshold in the present invention uses a relatively large value, and based on this design, the process parameters that are greatly affected by the time delay effect can be retained to the greatest extent, thereby avoiding as much as possible the elimination of the process parameters that are greatly affected by the time delay effect, which will have a bad impact on the subsequent steps.

[0046] Furthermore, the delay correlation coefficients of the process parameters at different lag times and the average delay correlation coefficients of the process parameters are expressed as follows:

[0047]

[0048] Among them, G i (t l ) is the lag time t l The time delay correlation coefficient of the ith process parameter at sec; N represents the total batch; f i,q (t l ) is the lag time t l The correlation coefficient corresponding to the i-th process parameter of the q-th batch at second, The lag time is t lThe average value of the correlation coefficient of the ith process parameter at t seconds; l = 1, 2, ..., T, t T Indicates the maximum lag time; G i is the average delay correlation coefficient of the i-th process parameter; in the embodiment of the present invention, t1 is set to 1 second, t T =360 seconds, that is, 360 seconds is the maximum lag value of the process parameter time series; taking the thin plate wire drying process as an example, the process time from raw material input to output is generally 5-6 minutes. Therefore, the present invention sets 360 seconds as the maximum lag value, which conforms to the lag law in the actual production process.

[0049] Furthermore, between step S1 and step S2, a normalization process is also included, that is, the pre-processed process time series data of each batch is normalized to increase the calculation efficiency.

[0050] The improvement of the improved particle swarm optimization algorithm lies in: adopting the strategy of linear decrease of inertia weight, adaptive change of acceleration factor and restriction of particle position; taking the cross-correlation coefficient of the screened process parameters and quality index data significantly affected by time delay as the fitness, taking the maximum cross-correlation coefficient as the goal, using the improved particle swarm optimization algorithm to perform global optimization, search for the global optimal lag time, and then obtain the optimal lag time solution set of the process parameter time series.

[0051] like Figure 3 As shown, the specific steps are as follows:

[0052] S4.1. Randomly initialize the particle swarm: The size of the particle swarm is set to h particles, the dimension of each particle is 1, and each particle in the particle swarm represents a lag time solution; the initial position of the particle is in the search space [0,t T ] is randomly generated within , and the speed vector is initialized to 0; in the embodiment of the present invention, h is 12, and the lag time range of the process parameters is set to [0,360], which is used to capture the correlation between the process parameters and the quality indicators at different lag times;

[0053] S4.2, calculating the fitness of each particle; the fitness value is determined by the correlation coefficient of the screened process parameters and quality index data that are significantly affected by the time delay;

[0054] S4.3, update the individual optimal position and global optimal position of the particle;

[0055] S4.4, update the inertia weight according to the linear decrease strategy of the inertia weight, and update the acceleration factor according to the adaptive change strategy of the acceleration factor;

[0056] S4.5, update the velocity and position of the particle;

[0057] S4.6, strategy for limiting particle positions: used to check whether there are particles whose positions are beyond the search range, and put the particles beyond the search range back into the search range;

[0058] S4.7, determine whether the set termination condition is met: if the termination condition is met, output the optimal lag time; if not, repeat steps S4.2 to S4.7 until the termination condition is met;

[0059] S4.8. Determine whether the upper limit of the process parameters has been reached. If the maximum number of iterations has been reached, output the optimal lag time. If the maximum number of iterations has not been reached, repeat steps S4.2 to S4.8.

[0060] The inertia weight linear reduction strategy expression is:

[0061]

[0062] Where: (t) is the inertia weight at the tth iteration, ω max is the maximum inertia weight, ω min is the minimum inertia weight, and Tmax is the maximum number of iterations.

[0063] The adaptive change strategy expression of the acceleration factor is:

[0064]

[0065] Where: is the individual and global acceleration factor for the tth iteration, c 1max 、c 2max is the individual and global maximum acceleration factor, c 1min 、c 2min is the individual and global minimum acceleration factor, and Tmax is the maximum number of iterations.

[0066] The search strategy expression in the strategy of limiting the particle position is:

[0067] 0.1h min ≤h i ≤0.1h max

[0068] Where: h i represents the particle position, h min is the particle search lower limit, h max The upper limit for particle search.

[0069] Furthermore, in the above, the inertia weight is reduced from 0.9 to 0.2, gradually reducing the particle's dependence on the previous velocity and enhancing the convergence ability, that is, ω max Take 0.9, ω minTake 0.2; Individual learning coefficient: Reduce from 2 to 1, so that each particle gradually reduces its dependence on its own historical optimal position during the search process, that is, c 1max Take 2, c 1min Take 1; Global learning coefficient: Increase from 1 to 2, gradually enhance the ability of particles to follow the global optimal position, and promote the group to explore the global optimal solution, that is, c 2max Take 2, c 2min Take 1; the algorithm sets Tmax to 100.

[0070] Illustratively, an optional implementation manner of the present invention is described in detail below, but the content of the present invention is not limited to the scope.

[0071] In view of the problems of high coupling correlation, strong sample time series, and time delay effect between process parameters and quality indicators in process data, this embodiment proposes a process parameter time series alignment method based on time series correlation matrix and improved particle swarm algorithm. First, the time series correlation matrix is ​​constructed using the regularity feature screening method based on the time series correlation matrix, and the set of process parameters significantly affected by time delay is screened out; then, the improved particle swarm algorithm is used to optimize the screened time series correlation matrix, so as to find the optimal lag time solution set of process parameters and realize the precise alignment of process parameters and quality indicators in time series.

[0072] The sample data set comes from a thin-plate drying process production line of a process manufacturing enterprise. The process data of a process in October 2022 is derived from the process production line. The obtained process data is normalized, the time series correlation matrix is ​​constructed, the regularity feature is screened, the particle swarm optimization is improved, and the data is registered through the above-mentioned registration method to obtain the registered data. Then, the registered data are filled according to the LSTM model, and the training set and the test set are divided into 7:3 ratio. At the same time, a variety of machine learning regression models are used for training, and the feasibility and effectiveness of the present invention are verified based on the test set. The results are shown in Table 1.

[0073] Table 1 Prediction results of each model

[0074]

[0075] In this embodiment, the model evaluation indicators used include mean absolute error (MAE), mean square error (MSE) and goodness of fit (R 2 ).

[0076]

[0077] Where: s is the number of test set samples, y 1,j is the predicted quality index value, y 2,jis the actual quality index value, is the average value of the actual quality indicator data.

[0078] As shown in Table 1, the data aligned in time series by the method proposed in the invention have significantly better goodness of fit and error in different prediction models than the data that have not been aligned in time series. This shows that processing time delay effects can effectively improve the accuracy of the prediction model. In addition, these results further verify that the process data aligned using the method proposed in the invention can more accurately reflect the causal relationship between the model input and output, further improving the quality of the modeling data.

[0079] According to a second aspect of an embodiment of the present invention, a process parameter time series alignment system based on a time series association matrix and an improved particle swarm algorithm is provided, comprising a first module for executing S1, obtaining process process time series data of different batches of a process production line; preprocessing each batch of process process time series data to obtain the preprocessed process process time series data of each batch; wherein the preprocessed process process time series data is constructed from M process parameters and a quality indicator; a second module for executing S2, constructing a regularity feature screening method based on a time series association matrix to obtain a process parameter time series association matrix, and according to the time of the process parameters, The first module is used to identify and screen out the process parameter set significantly affected by the time delay based on the sequence association matrix; the third module is used to execute S3, and obtain the screened process parameter sequence association matrix based on the screened process parameter set; the fourth module is used to execute S4, and optimize the screened process parameter sequence association matrix based on the improved particle swarm algorithm to find the optimal lag time solution set of the process parameters; the fifth module is used to execute S5, and according to the optimal lag time solution set of the process parameters, the time series of each process parameter significantly affected by the time delay is shifted backward to the corresponding optimal lag time to achieve accurate alignment with the quality index in the time series, and obtain the registered data. The term "module" as used above can be a combination of software and / or hardware that implements a predetermined function. Although the system described in the above embodiment is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceived. For the parts not described in detail in each module, please refer to the relevant description of this embodiment.

[0080] According to a third aspect of an embodiment of the present invention, a processor is provided, wherein the processor is used to run a program, wherein when the program is running, any one of the above-mentioned process parameter timing alignment methods based on a timing correlation matrix and an improved particle swarm algorithm is executed.

[0081] According to a fourth aspect of an embodiment of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned process parameter timing alignment methods based on the timing correlation matrix and the improved particle swarm algorithm.

[0082] In an exemplary embodiment, the computer-readable storage medium may include, but is not limited to, various media that can store computer programs, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk.

[0083] The specific implementation modes of the present invention are described in detail above in conjunction with the accompanying drawings, but the present invention is not limited to the above implementation modes, and various changes can be made within the knowledge scope of ordinary technicians in this field without departing from the purpose of the present invention.

Claims

1. A process parameter timing registration method based on timing correlation matrix and improved particle swarm algorithm, characterized in that: The following steps are involved: S1. Obtain the process technology time series data of different batches of the process production line; Preprocessing each batch of process technology time series data to obtain the preprocessed process technology time series data of each batch; wherein the preprocessed process technology time series data is constructed by M process parameters and one quality indicator; S2. Constructing a regularity feature screening method based on a time series correlation matrix to obtain a time series correlation matrix of process parameters, and identifying and screening a set of process parameters significantly affected by time delay according to the time series correlation matrix of process parameters; S3. Obtaining a screened process parameter time series correlation matrix based on the screened process parameter set; S4, optimizing the screened process parameter time series correlation matrix based on the improved particle swarm algorithm to find the optimal lag time solution set of the process parameters; S5. Based on the optimal lag time solution set of process parameters, the time series of each process parameter significantly affected by time delay is shifted backward to the corresponding optimal lag time to achieve precise alignment with the quality index in the time series, and obtain the registered data.

2. The process parameter timing registration method based on timing correlation matrix and improved particle swarm algorithm according to claim 1 is characterized in that: The S2 comprises: S2.

1. Constructing a time series correlation matrix of process parameters: Calculate the correlation coefficients of the time series relative quality indicators of each process parameter in the process time series data after preprocessing of each batch at different lag times, so as to construct a time series correlation matrix of process parameters; S2.

2. Use the regular feature screening method to analyze the time series correlation matrix of the quantitative process parameters, so as to screen out the set of process parameters that are significantly affected by the time delay effect.

3. The process parameter timing registration method based on timing correlation matrix and improved particle swarm algorithm according to claim 2 is characterized in that: The regularity feature screening method comprises: According to the time series correlation matrix of process parameters, the time delay correlation coefficient of each process parameter under different lag times is obtained; According to the time delay correlation coefficients of each process parameter at different lag times, the average time delay correlation coefficient of each process parameter is obtained; A judgment is made based on the average delay correlation coefficient of each process parameter and a preset threshold: when the average delay correlation coefficient of the current process parameter is less than the preset threshold, the current process parameter is retained; and a process parameter set is obtained based on the retained process parameter.

4. The method for timing registration of process parameters based on timing correlation matrix and improved particle swarm algorithm according to claim 3 is characterized in that: The time delay correlation coefficients of the process parameters at different lag times and the average time delay correlation coefficients of the process parameters are expressed as follows: Among them, G i (t l ) is the lag time t l The time delay correlation coefficient of the ith process parameter at sec; N represents the total batch; f i，q (t l ) is the lag time t l The correlation coefficient corresponding to the i-th process parameter of the q-th batch at second, The lag time is t l The average value of the mutual correlation coefficient of the i-th process parameter at second; G i is the average delay correlation coefficient of the i-th process parameter.

5. The process parameter timing registration method based on timing correlation matrix and improved particle swarm algorithm according to claim 1 is characterized in that: Taking the mutual correlation coefficient of the screened process parameters and quality index data significantly affected by time delay as the fitness and the maximum mutual correlation coefficient as the goal, the improved particle swarm optimization algorithm is used to perform global optimization and search for the global optimal lag time, thereby obtaining the optimal lag time solution set of the process parameter time series.

6. A process parameter timing registration system based on timing correlation matrix and improved particle swarm algorithm, characterized in that: A module comprising the method according to any one of claims 1 to 5.

7. A processor, characterized in that: The processor is used to run a program, wherein the program, when running, executes the process parameter timing registration method based on the timing association matrix and the improved particle swarm algorithm as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute the process parameter timing alignment method based on the timing correlation matrix and the improved particle swarm algorithm as described in any one of claims 1 to 5.

Images