A method for processing process data of insulated cables

Abstract

The present invention provides a method for processing process data of an insulated cable, and relates to the technical field of process parameter optimization. The method comprises: obtaining multiple groups of process operation parameters of a target cable and product performance parameters corresponding to each group of process operation parameters, extracting a process data set of each product performance parameter, and determining multiple target process parameters corresponding to each product performance parameter; performing data enhancement on each process data set to obtain an extended data set, screening multiple first data enhancement sets from multiple extended data sets and performing cross-validation to determine multiple second data enhancement sets; performing data enhancement on multiple groups of process operation parameters and product performance parameters corresponding to each group of process operation parameters to obtain a process performance data set, processing the process performance data set through a process parameter optimization model, and obtaining optimized process parameters of the target cable. The present invention achieves the improvement of the process parameter optimization efficiency of the insulated cable.

Description

Technical Field

[0001] The present invention relates to the technical field of process parameter optimization, and in particular to a method for processing process data of an insulating cable. Background Art

[0002] With the continuous development of power transmission technology, insulated cables are increasingly used in modern industrial and civil fields. The manufacturing process of insulated cables requires strict process control to ensure their reliability and long life in complex environments. However, due to the diverse and interrelated process operating parameters, how to optimize these parameters to improve product performance has become a challenging task.

[0003] Data processing technology has been used to optimize process parameters. However, in the R&D process of insulated cable products, due to the limitations of experimental conditions in the initial stage, the collected process operation parameters and corresponding product performance data are relatively small. Faced with the situation of small data volume and high dimensionality, some methods are insufficient in dealing with data correlation, building models, and expanding data sets. How to effectively extract and utilize process parameters that are closely related to specific product performance parameters in limited sample data is one of the key issues that need to be solved to improve the quality of process parameter optimization. The simple method of expanding the data set is also prone to the problem of reduced prediction accuracy of the model due to the lack of accurate correlation analysis, which cannot meet actual R&D needs. Summary of the invention

[0004] The present invention proposes a method for processing process data of an insulated cable, aiming to solve at least one technical problem in the above-mentioned background technology.

[0005] To achieve the above-mentioned purpose, the present invention provides a method for processing process data of an insulated cable, comprising: obtaining multiple groups of process operation parameters of a target cable and product performance parameters corresponding to each group of process operation parameters, extracting a process data set for each product performance parameter, each process data set including multiple characteristic values ​​of one of the product performance parameters and a group of process operation parameters corresponding to each characteristic value; configuring a process performance association analysis model for each product performance parameter, processing the process data set corresponding to the product performance parameter through the process performance association analysis model, generating a first process operation parameter association list for each product performance parameter, and determining multiple target process parameters corresponding to each product performance parameter according to the first process operation parameter association list; performing data enhancement on each process data set based on the target process parameter to obtain to an expanded data set for each process data set, input the expanded data set into a corresponding process performance association analysis model, generate a second process operation parameter association list for each product performance parameter, the process operation parameter association list includes an association score for each process operation parameter; screen out multiple first data enhancement sets from multiple expanded data sets according to the first process operation parameter association list and the second process operation parameter association list, cross-validate the multiple first data enhancement sets to determine multiple second data enhancement sets; based on the multiple second data enhancement sets, perform data expansion on multiple groups of process operation parameters and product performance parameters corresponding to each group of process operation parameters to obtain a process performance data set, process the process performance data set through a process parameter optimization model to obtain optimized process parameters for the target cable.

[0006] Preferably, multiple first data enhancement sets are screened out from multiple extended data sets according to the first process operation parameter association list and the second process operation parameter association list, including: extracting a first process parameter score vector for each product performance parameter according to the association score of each process operation parameter in the first process operation parameter association list; extracting a second process parameter score vector for each product performance parameter according to the association score of each process operation parameter in the second process operation parameter association list, calculating the data correlation coefficient between the first process parameter score vector and the second process parameter score vector for each product performance parameter, and screening out extended data sets having a data correlation coefficient greater than a first correlation threshold from multiple extended data sets and merging them as the first data enhancement set.

[0007] Preferably, multiple first data enhancement sets are cross-validated to determine multiple second data enhancement sets, including: calculating the performance correlation coefficient between any two product performance parameters based on the first process parameter score vector, and determining multiple performance parameter combinations from multiple product performance parameters based on the second correlation threshold; for any first data enhancement set, if the product performance parameter corresponding to the first data enhancement set does not belong to any performance parameter combination, the first data enhancement set is recorded as the second data enhancement set; otherwise, determining a target model for verifying the first data enhancement set based on the product performance parameters and performance parameter combinations to which the first data enhancement set belongs, inputting the first data enhancement set into the target model, and generating a cross-correlation list corresponding to the first data enhancement set; extracting a first process parameter relationship vector from the second process operation parameter association list of the product performance parameters to which the first data enhancement set belongs, and extracting a second process parameter relationship vector from the cross-correlation list corresponding to the first data enhancement set; calculating the cross-correlation coefficient between the first process parameter relationship vector and the second process parameter relationship vector, and recording the first data enhancement set whose cross-correlation coefficient is greater than the first correlation threshold as the second data enhancement set.

[0008] Preferably, data enhancement is performed on each process data set based on the target process parameters to obtain an extended data set for each process data set, including: respectively extracting the normally distributed data of the product performance parameters and multiple process operation parameters in each process data set, respectively adding normally distributed noise to multiple characteristic values ​​of the product performance parameters in each process data set based on a preset proportional factor, and adding normally distributed noise to multiple target process parameters in a group of process operation parameters corresponding to each characteristic value in the process data set based on a preset proportional factor according to multiple target process parameters corresponding to the product performance parameters to which the process data set belongs, to obtain an extended data set for each process data set.

[0009] Preferably, data expansion is performed on multiple groups of process operation parameters and product performance parameters corresponding to each group of process operation parameters based on multiple second data enhancement sets to obtain a process performance data set, including: associating each group of process operation parameters with the corresponding product performance parameters to form a group of training data, extracting multiple groups of noise combination data from the second data enhancement set, including noise data of one of the product performance parameters and noise data of multiple target process parameters in a group of process operation parameters corresponding to the product performance parameters; using the process operation parameters as indexes, determining the training data corresponding to each group of noise combination data, adding multiple noise data in the noise combination to the corresponding training data to form a new group of training data, and constructing a process performance data set containing multiple groups of training data.

[0010] Preferably, multiple target process parameters corresponding to each product performance parameter are determined according to the first process operation parameter association list, including: extracting multiple target process parameters corresponding to each product performance parameter based on a preset association cumulative threshold, including sorting multiple product performance parameters based on the association score of each process operation parameter in the first process operation parameter association list, selecting product performance parameters based on the association score from high to low, and accumulating the association scores corresponding to the multiple selected product performance parameters, if the corresponding association scores of the current n product performance parameters are accumulated and are less than the preset association cumulative threshold, and the corresponding association scores of the first n+1 product performance parameters are accumulated and are not less than the preset association cumulative threshold, then the first n+1 product performance parameters among the multiple product performance parameters sorted from high to low based on the association scores are recorded as target process parameters.

[0011] Preferably, it also includes: for the first process parameter scoring vector and the second process parameter scoring vector, respectively normalizing the association scores of multiple process operation parameters in the first process operation parameter association list and the second process operation parameter association list to obtain the normalized association score corresponding to each process operation parameter in the first process operation parameter association list and the second process operation parameter association list, and constructing the first process parameter scoring vector and the second process parameter scoring vector, each element in the process parameter scoring vector represents the normalized association score of one of the process operation parameters; for the first process parameter relationship vector and the second process parameter relationship vector, respectively sorting the multiple process operation parameters in the second process operation parameter association list and the cross-correlation list based on the association score, determining the relationship score of each process operation parameter based on the sequence number of the process operation parameter, respectively normalizing the relationship scores of multiple process operation parameters in the first process parameter relationship vector and the second process parameter relationship vector to obtain the normalized relationship score of each process operation parameter in the first process parameter relationship vector and the second process parameter relationship vector, and constructing the first process parameter relationship vector and the second process parameter relationship vector, each element in the process parameter relationship vector represents the normalized relationship score of one of the process operation parameters.

[0012] Preferably, the process performance data set is processed by a process parameter optimization model to obtain optimized process parameters of the target cable, including: determining the target performance optimization parameters of the target cable, constructing the objective function of the process parameter optimization model based on minimizing the product performance parameters, constructing multiple initial particles based on multiple groups of training data in the process performance data set, including determining the position of each particle based on the process operation parameters contained in the multiple groups of training data, and initializing the speed of the particle based on the position of each particle to obtain multiple initial particles, iteratively updating the positions and speeds of the multiple initial particles, determining the target position after reaching a preset number of iterations, and determining the optimized process parameters of the target cable according to the target position.

[0013] Preferably, the data correlation coefficient between the first process parameter scoring vector and the second process parameter scoring vector is determined based on the absolute value of the Pearson correlation coefficient between the first process parameter scoring vector and the second process parameter scoring vector, and the cross correlation coefficient between the first process parameter relationship vector and the second process parameter relationship vector is determined based on the cosine similarity between the first process parameter relationship vector and the second process parameter relationship vector.

[0014] Preferably, the process performance correlation analysis model is configured as a random forest model.

[0015] The present invention has the following beneficial effects: the present invention analyzes the complex relationship between various process operation parameters and product performance parameters through a process performance correlation analysis model, determines multiple target process parameters for each process operation parameter according to the complex relationship between various process operation parameters and product performance parameters, and performs data enhancement on the original data based on the multiple target process parameters to enhance the diversity and adaptability of the data set, so that even when the data samples are small, the accuracy of the optimization model can be effectively improved to avoid overfitting of the model or falling into a local optimum; by analyzing the correlation between different product performance parameters, the enhanced data is cross-validated to screen out a combination of process parameters that exhibits consistency and stability under different process conditions, thereby constructing a process parameter optimization model. The process performance data set effectively improves the quality of the data set, so that the process parameter optimization model can more accurately capture the impact of process conditions on product performance, thereby obtaining more reliable optimized process parameters; the particle swarm algorithm is used for multi-objective optimization to ensure that the final optimization results can reach or approach the expected goals in multiple performance indicators. The generated optimized process parameters have strong generalization ability. Compared with randomly generated particles, the process performance data set is used for particle initialization, which can reduce the amount of calculation to a certain extent, effectively improve the quality of the initial solution and improve the stability of the algorithm, which is helpful to better locate the global optimal solution. Through precise process parameter optimization, the process parameter optimization efficiency of insulated cables is improved, providing a solid technical guarantee for the production of high-quality cable products. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic flow chart of a method for processing process data of an insulated cable provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to make those skilled in the art better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0018] Figure 1 A schematic diagram of a process flow of an insulated cable process data processing method provided by an embodiment of the present invention is shown, such as Figure 1 As shown, the method includes the following steps: S1, obtaining multiple groups of process operation parameters of the target cable and product performance parameters corresponding to each group of process operation parameters, and extracting a process data set of each product performance parameter.

[0019] In this embodiment, the operating process parameters of the target cable refer to the variables controlled and adjusted during the production process. These parameters directly affect the performance of the final product. For the manufacture of insulated cables, the operating process parameters include but are not limited to material ratio, extrusion temperature / pressure, curing temperature / pressure, drawing speed, cooling speed, etc. The product performance parameters corresponding to each set of process operating parameters are specifically various performance parameters exhibited by the insulated cable under these operating process parameters, such as insulation strength, pressure resistance, conductivity, etc. These process operating parameters and their corresponding product performance parameters are sorted and classified to extract the process data set of each product performance parameter (such as insulation strength). Each process data set contains multiple characteristic values ​​of the product performance parameter, which reflect the changes of the product performance parameter under different process conditions, and each characteristic value corresponds to a specific set of process operating parameters.

[0020] S2. Configure a process performance association analysis model for each product performance parameter, process the process data set corresponding to the product performance parameter through the process performance association analysis model, generate a first process operation parameter association list for each product performance parameter, and determine multiple target process parameters corresponding to each product performance parameter based on the first process operation parameter association list.

[0021] In this embodiment, a process performance correlation analysis model is configured for each product performance parameter to analyze the complex relationship between the process operation parameters and the product performance parameters, thereby obtaining a first process operation parameter correlation list for each product performance parameter. The first process operation parameter correlation list records correlation scores corresponding to multiple process operation parameters. The correlation scores are used to characterize the degree of correlation between the process operation parameters and the product performance parameters. The multiple target process parameters are multiple process operation parameters that have a greater impact on product performance.

[0022] Optionally, for each product performance parameter, multiple target process parameters corresponding to each product performance parameter are extracted based on a preset correlation accumulation threshold. Specifically, based on the correlation score of each process operation parameter in the first process operation parameter association list, multiple product performance parameters are sorted from high to low based on the correlation score, and product performance parameters are selected based on the correlation score from high to low. After each process operation parameter is selected, the corresponding correlation scores of the selected process operation parameters are accumulated, and it is determined whether the corresponding correlation scores of the selected process operation parameters are less than the preset correlation accumulation threshold after accumulation. If so, continue to select the next process operation parameter, otherwise stop selecting. The corresponding correlation scores of the current n product performance parameters are less than the preset correlation accumulation threshold after accumulation, and the corresponding correlation scores of the first n+1 product performance parameters are not less than the preset correlation accumulation threshold after accumulation, the process operation parameter selection operation is completed, and the first n+1 product performance parameters of the multiple product performance parameters sorted from high to low based on the correlation score are recorded as target process parameters. In this way, multiple representative process operation parameters of each product performance parameter can be selected.

[0023] Among them, the process performance association analysis model is preferably a random forest model. It is worth noting that in the process of developing insulated cables, due to limited test data, less sample data is collected. Directly analyzing the data based on the machine learning model to optimize the process operation parameters is not conducive to capturing the specific impact of different process operation parameters on each product performance parameter. Therefore, in the form of single variable analysis, multiple random forest models are used to process the data in the process data set corresponding to one of the product performance parameters. Specifically, multiple groups of process operation parameters in the process data set are used as input variables of the model, and the characteristic values ​​corresponding to each group of process operation parameters are used as the target variables of the model. The model can automatically learn the relationship between each process operation parameter and the product performance parameter by constructing multiple decision trees to identify which process operation parameters are more critical to the prediction of product performance parameters. In the end, the model will evaluate the importance of each process operation parameter, that is, the feature importance score, which is specifically achieved by analyzing the contribution of each process operation parameter at the split node in all decision trees. The main purpose of analyzing different process operation parameters through random forests is to identify which process operation parameters have a greater impact on product performance parameters. Since random forests do not directly reveal the specific relationship between process operation parameters and product performance parameters (such as linearity, nonlinearity, interaction, etc.), when there are fewer sample data, it can better analyze the relative importance of different process operation parameters to product performance parameters. S3. Perform data enhancement on each process data set based on the target process parameters to obtain an expanded data set for each process data set, input the expanded data set into the corresponding process performance association analysis model, and generate a second process operation parameter association list for each product performance parameter.

[0024] In this embodiment, data enhancement is performed on each process data set based on the target process parameters determined in step S2. Data enhancement is a method of expanding an existing data set, usually by adding noise or simulating additional experimental conditions to generate new data points. In this scheme, data enhancement is performed based on the target process parameters. The specific method is to reasonably change or expand the process data related to these target process parameters to generate a richer combination of process conditions, and obtain an expanded data set after data enhancement. These expanded data sets are input into the corresponding process performance association analysis model, that is, the process performance association analysis model configured for the product performance parameters corresponding to each process data set. The model will re-analyze based on the newly generated data to generate a second process operation parameter association list for each product performance parameter. Similarly, the second process operation parameter association list includes the association score of each process operation parameter after verification by the expanded data, reflecting the influence of each process operation parameter on the performance parameter under the conditions of the expanded data.

[0025] S4. Filter out a plurality of first data enhancement sets from a plurality of expanded data sets according to the first process operation parameter association list and the second process operation parameter association list, and cross-validate the plurality of first data enhancement sets to determine a plurality of second data enhancement sets.

[0026] In this embodiment, the first process operation parameter association list and the second process operation parameter association list corresponding to any product performance parameter respectively record the association scores between each process operation parameter and the product performance parameter in the original data and the enhanced data, reflecting the influence of each process operation parameter on the product performance parameter. The association between the original data and the enhanced data can be analyzed through the first process operation parameter association list and the second process operation parameter association list to determine the quality of data enhancement.

[0027] In an optional implementation, multiple first data enhancement sets are screened out from multiple extended data sets based on the first process operation parameter association list and the second process operation parameter association list, specifically including: extracting a first process parameter score vector for each product performance parameter based on the association score of each process operation parameter in the first process operation parameter association list, and extracting a second process parameter score vector for each product performance parameter based on the association score of each process operation parameter in the second process operation parameter association list; specifically, the association scores in the process operation parameter association list can be normalized to determine the normalized association score of each process operation parameter, and then a corresponding process parameter score vector is constructed, each element in the process parameter score vector corresponds to the normalized association score of one of the process operation parameters, and the first process parameter score vector and the second process parameter score vector are constructed based on the same template to quantify the relative importance of each process operation parameter in the original data and the enhanced data.

[0028] After determining the first process parameter scoring vector and the second process parameter scoring vector, the data correlation coefficient between the first process parameter scoring vector and the second process parameter scoring vector for each product performance parameter is calculated, and the expanded data sets having data correlation coefficients greater than a first correlation threshold are screened out from multiple expanded data sets and merged as a first data enhancement set.

[0029] Specifically, the data correlation coefficient between the first process parameter scoring vector and the second process parameter scoring vector of each product performance parameter calculated is a statistical indicator used to measure the similarity between the two vectors, for example, by calculating the Pearson correlation coefficient or cosine similarity between the two vectors. In this embodiment, the Pearson correlation coefficient is taken as an example. Then, the first correlation threshold is used to screen out those extended data sets whose data correlation coefficients are greater than the threshold from multiple extended data sets. The purpose of setting the first correlation threshold is to ensure that the selected data enhancement set shows consistency and stability in the association analysis at different stages. Only when the correlation between the first process parameter scoring vector and the second process parameter scoring vector is strong (that is, the data correlation coefficient is greater than the set threshold), the extended data set is considered to be reliable and valid. These screened extended data sets are ultimately recorded as the first data enhancement set, providing a basis for subsequent data analysis and process parameter optimization. It is worth noting that when calculating the data correlation coefficient based on the Pearson correlation coefficient, since the Pearson correlation coefficient ranges from -1 to 1, when the correlation coefficient tends to -1 or 1, it indicates a strong correlation. In this case, the data correlation coefficient can be taken as the absolute value of the Pearson correlation coefficient, so as to screen out the expanded data set with a strong correlation with the original data based on a pre-set threshold and record it as the first data enhancement set.

[0030] In order to further improve the reliability and rationality of the data, cross-validation is performed on these first data enhancement sets to determine multiple second data enhancement sets. Cross-validation further analyzes the relationship between different data sets, so that the data after data enhancement conforms to the diversity and correlation between process operation parameters, ensuring that the selected data enhancement set has good generalization ability and consistency.

[0031] S5. Based on multiple second data enhancement sets, data expansion is performed on multiple groups of process operation parameters and product performance parameters corresponding to each group of process operation parameters to obtain a process performance data set, and the process performance data set is processed by a process parameter optimization model to obtain optimized process parameters of the target cable.

[0032] In this embodiment, based on multiple second data enhancement sets, comprehensive data expansion is performed on multiple groups of process operation parameters and product performance parameters corresponding to each group of process operation parameters. Through this expansion process, a process performance data set covering a wider range of process conditions and performance performance is generated.

[0033] In an optional implementation, data expansion is performed on multiple groups of process operation parameters and product performance parameters corresponding to each group of process operation parameters based on multiple second data enhancement sets, specifically including: associating each group of process operation parameters with the corresponding product performance parameters to form a group of training data, extracting multiple groups of noise combination data from the second data enhancement set, determining the training data corresponding to each group of noise combination data with the process operation parameters as indexes, adding multiple noise data in the noise combination to the corresponding training data to form a new group of training data, and constructing a process performance data set containing multiple groups of training data.

[0034] In this embodiment, in order to further enrich the process performance data set, each set of process operation parameters is first associated with its corresponding product performance parameters to form a complete set of training data. These training data are constructed based on the original data. As an important basis for model learning and optimization, they help to build a more accurate process parameter optimization model by reflecting the actual impact of process operation parameters on product performance. And extract multiple sets of noise combination data from the second data enhancement set that has been screened, these noise combination data include noise data of a specific product performance parameter (such as insulation strength, voltage resistance, etc.), and noise data of multiple target process parameters in a set of process operation parameters associated with the product performance parameters. The noise data here is based on the noise added in the previous data enhancement process, aiming to simulate the possible changes in product performance under different process conditions. The design of this noise combination data can effectively increase the diversity of training data, so that the generated process performance data set has a wider applicability.

[0035] After obtaining these noise combination data, the training data corresponding to each set of noise combination data is determined with the process operation parameters as the index. Specifically, it is necessary to first identify the process operation parameters and performance parameters involved in each set of noise combination data, and then find data pairs matching these parameters in the existing training data set. This matching process ensures that the addition of noise is based on the existing actual data, thereby maintaining the rationality and physical meaning of the data. Next, multiple noise data in the noise combination are added to the corresponding training data to form a new set of training data. This process can be regarded as expanding and enhancing the data based on the original data in combination with simulated noise to generate more possible process scenarios. These newly generated training data retain the basic structure of the original data, while expanding the distribution range and possibility of the data through the introduction of noise, so that the model can learn and optimize under more diverse conditions. The process performance data set constructed through the above steps contains multiple sets of training data. This expanded data set not only covers the correlation between the original process operation parameters and product performance parameters, but also introduces the changes that may occur under simulated different process conditions, further improving the breadth and adaptability of the data set. Through these rich training data, the process parameter optimization model can better capture the changing patterns of product performance under complex process conditions, thereby deriving a more optimized and reliable process parameter combination and improving the quality and performance of the final product.

[0036] Finally, the process performance data set is input into the process parameter optimization model. Based on the process parameter optimization model and the expected product performance parameter analysis, the optimized process parameters of the target cable are obtained. The optimized process parameters can be used to provide a reference for the high-quality production of the target cable, so as to improve the product quality of the insulated cable.

[0037] In an optional implementation, in step S3, data enhancement is performed on each process data set based on the target process parameters to obtain an expanded data set for each process data set, specifically including: extracting the normally distributed data of the product performance parameters and multiple process operation parameters in each process data set, and adding normally distributed noise to multiple characteristic values ​​of the product performance parameters in each process data set based on a preset proportional factor.

[0038] Specifically, the normally distributed data extracted from the product performance parameters and multiple process operation parameters in the process data set can accurately reflect the distribution characteristics of each parameter in the actual process. The preset proportional factor is used to control the amplitude of the noise, so that the added noise will not deviate too much from the actual data range while simulating the fluctuations that may occur under real process conditions. By adding normally distributed noise to the eigenvalues, more diverse performance data can be generated, thereby effectively expanding the coverage of the original data and enabling the model to be trained and optimized under a wider range of conditions.

[0039] According to the multiple target process parameters corresponding to the product performance parameters to which the process data set belongs, normal distribution noise is added to the multiple target process parameters in a set of process operation parameters corresponding to each characteristic value in the process data set based on a preset proportional factor to obtain an expanded data set for each process data set.

[0040] Specifically, the target process parameters are key parameters that have a greater impact on product performance determined through preliminary analysis. Similar to the noise addition of product performance parameters, the noise addition of these process operation parameters is also based on a preset proportional factor. The process operation parameter values ​​after adding noise can simulate the small fluctuations that may occur in the actual production process. In this way, the added noise data can be made to conform to the strong correlation between product performance parameters and process operation parameters, avoiding the introduction of more interference noise. Adding moderate noise introduces more possible process conditions and performance performance.

[0041] In an optional implementation, in step S4, multiple first data enhancement sets are cross-validated to determine multiple second data enhancement sets, specifically including: calculating the performance correlation coefficient between any two product performance parameters based on the first process parameter scoring vector, determining multiple performance parameter combinations from multiple product performance parameters based on the second correlation threshold, and for any first data enhancement set, if the product performance parameter corresponding to the first data enhancement set does not belong to any performance parameter combination, the first data enhancement set is recorded as the second data enhancement set.

[0042] In this embodiment, the performance correlation coefficient between any two product performance parameters is calculated based on the first process parameter scoring vector. The first process parameter scoring vector is generated by analyzing the influence of the process operation parameters on each product performance parameter, reflecting the importance of each process operation parameter in different product performance indicators. The performance correlation coefficient is used to measure the degree of correlation between different product performance parameters. By comparing the similarity between these scoring vectors, it is possible to identify which product performance parameters have a strong correlation. Through the second correlation threshold, combinations of product performance parameters with a strong correlation between each other are determined from multiple product performance parameters, thereby obtaining multiple performance parameter combinations.

[0043] After determining multiple performance parameter combinations, for each first data enhancement set, first check whether the corresponding product performance parameter belongs to any of the above performance parameter combinations. If the product performance parameter of the first data enhancement set does not belong to any performance parameter combination, then the first data enhancement set is directly recorded as the second data enhancement set, because if a product performance parameter has a low correlation with other product performance parameters, there is no need to confirm its reliability through complex cross-validation.

[0044] Otherwise, determine the target model for verifying the first data enhancement set based on the product performance parameters and performance parameter combinations to which the first data enhancement set belongs, input the first data enhancement set into the target model, and generate a cross-correlation list corresponding to the first data enhancement set.

[0045] In this embodiment, if the product performance parameters of the first data enhancement set belong to a certain performance parameter combination, further cross-validation is required. Specifically, a target model for validating the set is determined, and the process performance correlation analysis model configured by another set specifically associated with the target model can be used to verify the degree of correlation between the data in the set and another product performance parameter. The cross-correlation list generated by the target model reflects the correlation between the process operation parameters of the data enhancement set in the target model and the product performance parameters to which the model belongs.

[0046] A first process parameter relationship vector is extracted from a second process operation parameter association list of product performance parameters to which the first data enhancement set belongs, and a second process parameter relationship vector is extracted from a cross-association list corresponding to the first data enhancement set; a cross-correlation coefficient between the first process parameter relationship vector and the second process parameter relationship vector is calculated, and the first data enhancement set whose cross-correlation coefficient is greater than a first correlation threshold is recorded as the second data enhancement set.

[0047] In this embodiment, the process parameter relationship vector is specifically obtained based on the importance ranking and relative position information extraction of each process operation parameter. Exemplarily, the second process operation parameter association list and the cross-association list respectively record the relative position relationship of multiple process operation parameters, and the relative position relationship can be specifically obtained by sorting based on the association score corresponding to each process operation parameter, so as to determine the relationship score of each process operation parameter based on the sequence number of the process operation parameter. Exemplarily, the sorting result is converted into a relationship score of 0-10, and each relationship score is equally spaced. For example, for 6 process operation parameters, according to the order of association scores from high to low, the relationship scores are 10, 8, 6, 4, 2, and 0 respectively. Then, the relationship scores of multiple process operation parameters in the first process parameter relationship vector and the second process parameter relationship vector are normalized respectively to obtain the normalized relationship score of each process operation parameter in the first process parameter relationship vector and the second process parameter relationship vector, and the process parameter relationship vector is constructed based on the normalized relationship score. In the first process parameter relationship vector and the second process parameter relationship vector, each element in the same position represents the normalized relationship score of the same process operation parameter.

[0048] After determining the first process parameter relationship vector and the second process parameter relationship vector, the cross-correlation coefficient between the first process parameter relationship vector and the second process parameter relationship vector is calculated. The cross-correlation coefficient is used to measure whether the performance of the first data enhancement set in the target model is consistent with its performance in the original model, for example, calculated based on cosine similarity. If the cross-correlation coefficient is greater than the preset first correlation threshold, the first data enhancement set is considered to be stable and relatively consistent in the target model corresponding to the first data enhancement set associated with it, and is considered to have passed cross-validation, and can be recorded as the second data enhancement set. These cross-validated second data enhancement sets will be used for further analysis and optimization to ensure the reliability of the data and the accuracy of the model. Through the above cross-validation process, a more stable and consistent second data enhancement set can be screened out, thereby providing a more solid data foundation for the optimization of process parameters and the improvement of final product performance.

[0049] It is worth mentioning that the cross-correlation coefficient is used to analyze whether the relative importance of the same process operation parameters in different models is consistent or relatively similar, that is, whether the importance rankings of the two product performance parameters with strong correlation are consistent or relatively similar. Therefore, after calculating the cross-correlation coefficient based on cosine similarity, there is no need to take the absolute value.

[0050] In an optional implementation, for step S5, the process performance data set is processed by a process parameter optimization model to obtain optimized process parameters of the target cable, specifically including: determining the target performance optimization parameters of the target cable, and constructing an objective function of the process parameter optimization model based on minimizing the product performance parameters.

[0051] In this embodiment, the target performance optimization parameter is a specific performance indicator that needs to be achieved during the optimization process, which can be determined based on actual needs. The target performance optimization parameter can be a specific electrical and mechanical performance indicator such as insulation strength, voltage resistance, and conductivity. After obtaining the target performance optimization parameter, the objective function of the process parameter optimization model is constructed. In this embodiment, the objective function is constructed with the goal of minimizing the product performance parameters, where: objective function = , where For the The predicted value of a product performance parameter under given process operating parameters, For the The target value of a product performance parameter is the value corresponding to the product performance parameter in the target performance optimization parameter. For the The weight of each product performance parameter, The number of product performance parameters.

[0052] In the process parameter optimization model, a particle swarm algorithm is used to search for the optimal process operation parameter combination. In this process, multiple initial particles are first constructed based on multiple sets of training data in the process performance data set, including determining the position of each particle based on the process operation parameters contained in the multiple sets of training data, and initializing the particle speed based on the position of each particle to obtain multiple initial particles. Specifically, the process parameter combination in each set of training data can be used as an initial position of the particle, and the initial value of the particle speed is set based on the variation range of the process operation parameters. For example, the difference between the maximum and minimum values ​​of one of the process operation parameters in the process performance data set is determined, and the product of the preset proportional coefficient and the difference is used as the initial value of the speed of one of the examples on this process operation parameter. After initializing the position and speed of the particle, the position and speed of the particle are iteratively updated to continuously approach the global optimal solution. After multiple iterations, when the preset number of iterations is reached, the algorithm will determine the current global optimal position, that is, the target position, and extract the process parameter combination corresponding to the target position. These parameters are the optimized process parameters of the target cable. This optimized process parameter can enable the target cable to reach or approach the expected target in various performance indicators.

[0053] It is worth noting that particle initialization based on the process performance data set generated by analyzing the original data can ensure that the initial solution has a certain degree of rationality and effectiveness. These initial solutions usually come from historical experience or actual experimental data, and may be close to the global optimal solution, so that the algorithm searches in the good area of ​​the solution space from the beginning, reducing the scope and time of the search, thereby accelerating the convergence speed, avoiding the inferior solutions that may appear during random initialization, and thus improving the efficiency of the entire optimization process. For some complex process parameter optimization problems, initialization with raw data can make it easier for the algorithm to explore local optimal solutions and gradually approach the global optimal solution through iteration. In contrast, completely random initialization may cause particles to skip some important local optimal solution areas in the early stages of the search.

[0054] An insulated cable process data processing method provided by an embodiment of the present invention analyzes the complex relationship between various process operation parameters and product performance parameters through a process performance correlation analysis model, determines multiple target process parameters for each process operation parameter according to the complex relationship between various process operation parameters and product performance parameters, and performs data enhancement on the original data based on the multiple target process parameters to enhance the diversity and adaptability of the data set, so that even when the data samples are small, the accuracy of the optimization model can be effectively improved to avoid model overfitting or falling into a local optimum; by analyzing the correlation between different product performance parameters, the enhanced data is cross-validated to screen out a combination of process parameters that exhibits consistency and stability under different process conditions, thereby A process performance data set was constructed, which effectively improved the quality of the data set, so that the process parameter optimization model can more accurately capture the impact of process conditions on product performance, thereby obtaining more reliable optimized process parameters; the particle swarm algorithm was used for multi-objective optimization to ensure that the final optimization results can reach or approach the expected goals in multiple performance indicators. The generated optimized process parameters have strong generalization ability. Compared with randomly generated particles, the use of process performance data sets for particle initialization can reduce the amount of calculation to a certain extent, effectively improve the quality of the initial solution and improve the stability of the algorithm, which is helpful to better locate the global optimal solution. Through precise process parameter optimization, the process parameter optimization efficiency of insulated cables is improved, providing a solid technical guarantee for the production of high-quality cable products.

[0055] The above is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. Parts not described in detail in this specification belong to the prior art known to those skilled in the art.

Claims

1. A method for processing process data of an insulated cable, characterized in that: include: Acquire multiple groups of process operation parameters of the target cable and product performance parameters corresponding to each group of process operation parameters, and extract a process data set of each product performance parameter, wherein each process data set includes multiple characteristic values ​​of one of the product performance parameters and a group of process operation parameters corresponding to each characteristic value; configuring a process performance association analysis model for each product performance parameter, processing a process data set corresponding to the product performance parameter through the process performance association analysis model, generating a first process operation parameter association list for each product performance parameter, and determining a plurality of target process parameters corresponding to each product performance parameter according to the first process operation parameter association list; Performing data enhancement on each process data set based on the target process parameter to obtain an expanded data set of each process data set, inputting the expanded data set into the corresponding process performance association analysis model to generate a second process operation parameter association list for each product performance parameter, wherein the process operation parameter association list includes an association score for each process operation parameter; Screening out a plurality of first data enhancement sets from a plurality of augmented data sets according to the first process operation parameter association list and the second process operation parameter association list, and cross-validating the plurality of first data enhancement sets to determine a plurality of second data enhancement sets; Based on multiple second data enhancement sets, data expansion is performed on multiple groups of process operation parameters and product performance parameters corresponding to each group of process operation parameters to obtain a process performance data set, and the process performance data set is processed by a process parameter optimization model to obtain optimized process parameters for the target cable.

2. The method for processing process data of an insulated cable according to claim 1, characterized in that: Screening out a plurality of first data enhancement sets from a plurality of extended data sets according to the first process operation parameter association list and the second process operation parameter association list includes: Extracting a first process parameter score vector for each product performance parameter according to the association score of each process operation parameter in the first process operation parameter association list; A second process parameter scoring vector for each product performance parameter is extracted according to the association score of each process operation parameter in the second process operation parameter association list, a data correlation coefficient between the first process parameter scoring vector and the second process parameter scoring vector for each product performance parameter is calculated, and expanded data sets having data correlation coefficients greater than a first correlation threshold are screened out from multiple expanded data sets and merged and recorded as a first data enhancement set.

3. The method for processing process data of an insulated cable according to claim 2, characterized in that: Cross-validating the multiple first data augmentation sets to determine the multiple second data augmentation sets includes: Calculating a performance correlation coefficient between any two product performance parameters based on the first process parameter score vector, and determining a plurality of performance parameter combinations from the plurality of product performance parameters based on a second correlation threshold; For any first data enhancement set, if the product performance parameter corresponding to the first data enhancement set does not belong to any performance parameter combination, the first data enhancement set is recorded as the second data enhancement set; Otherwise, determining a target model for verifying the first data enhancement set according to the product performance parameters and the performance parameter combinations to which the first data enhancement set belongs, inputting the first data enhancement set into the target model, and generating a cross-correlation list corresponding to the first data enhancement set; Extracting a first process parameter relationship vector from a second process operation parameter association list of product performance parameters to which the first data enhancement set belongs, and extracting a second process parameter relationship vector from a cross-association list corresponding to the first data enhancement set; A cross-correlation coefficient between the first process parameter relationship vector and the second process parameter relationship vector is calculated, and a first data enhancement set whose cross-correlation coefficient is greater than a first correlation threshold is recorded as a second data enhancement set.

4. The method for processing process data of an insulated cable according to claim 1, characterized in that: Data enhancement is performed on each process data set based on the target process parameters to obtain an expanded data set for each process data set, including: Normal distribution data of product performance parameters and multiple process operation parameters in each process data set are extracted respectively, and normal distribution noise is added to multiple characteristic values ​​of product performance parameters in each process data set based on a preset proportional factor. According to multiple target process parameters corresponding to the product performance parameters belonging to the process data set, normal distribution noise is added to multiple target process parameters in a group of process operation parameters corresponding to each characteristic value in the process data set based on a preset proportional factor to obtain an extended data set for each process data set.

5. The method for processing process data of an insulated cable according to claim 1, characterized in that: Based on the multiple second data enhancement sets, data expansion is performed on the multiple groups of process operation parameters and the product performance parameters corresponding to each group of process operation parameters to obtain a process performance data set, including: Associating each set of process operation parameters with the corresponding product performance parameters to form a set of training data, and extracting multiple sets of noise combination data from the second data enhancement set, including noise data of one of the product performance parameters and noise data of multiple target process parameters in a set of process operation parameters corresponding to the product performance parameter; Taking the process operation parameters as indexes, the training data corresponding to each group of noise combination data is determined, and multiple noise data in the noise combination are added to the corresponding training data to form a new group of training data, so as to construct a process performance data set containing multiple groups of training data.

6. The method for processing process data of an insulated cable according to claim 2, characterized in that: Determine a plurality of target process parameters corresponding to each product performance parameter according to the first process operation parameter association list, including: Based on a preset association cumulative threshold, multiple target process parameters corresponding to each product performance parameter are extracted, including sorting multiple product performance parameters based on the association score of each process operation parameter in the first process operation parameter association list, selecting product performance parameters based on the association score from high to low, and accumulating the association scores corresponding to the multiple selected product performance parameters. If the corresponding association scores of the current n product performance parameters are accumulated and are less than the preset association cumulative threshold, and the corresponding association scores of the first n+1 product performance parameters are accumulated and are not less than the preset association cumulative threshold, then the first n+1 product performance parameters among the multiple product performance parameters sorted from high to low based on the association scores are recorded as target process parameters.

7. The method for processing process data of an insulated cable according to claim 3, characterized in that: Also includes: For the first process parameter scoring vector and the second process parameter scoring vector, respectively normalize the association scores of multiple process operation parameters in the first process operation parameter association list and the second process operation parameter association list to obtain the normalized association score corresponding to each process operation parameter in the first process operation parameter association list and the second process operation parameter association list, and construct the first process parameter scoring vector and the second process parameter scoring vector, wherein each element in the process parameter scoring vector represents the normalized association score of one of the process operation parameters; For the first process parameter relationship vector and the second process parameter relationship vector, multiple process operation parameters in the second process operation parameter association list and the cross-association list are sorted based on the association score, the relationship score of each process operation parameter is determined based on the sequence number of the process operation parameter, and the relationship scores of the multiple process operation parameters in the first process parameter relationship vector and the second process parameter relationship vector are normalized to obtain the normalized relationship score of each process operation parameter in the first process parameter relationship vector and the second process parameter relationship vector, and the first process parameter relationship vector and the second process parameter relationship vector are constructed, and each element in the process parameter relationship vector represents the normalized relationship score of one of the process operation parameters.

8. The method for processing process data of an insulated cable according to claim 5, characterized in that: The process performance data set is processed through the process parameter optimization model to obtain the optimized process parameters of the target cable, including: Determine the target performance optimization parameters of the target cable, construct the objective function of the process parameter optimization model based on minimizing the product performance parameters, construct multiple initial particles based on multiple sets of training data in the process performance data set, including determining the position of each particle based on the process operation parameters contained in the multiple sets of training data, and initializing the speed of the particle based on the position of each particle to obtain multiple initial particles, iteratively update the positions and speeds of the multiple initial particles, determine the target position after reaching a preset number of iterations, and determine the optimized process parameters of the target cable according to the target position.

9. The method for processing process data of an insulated cable according to claim 3, characterized in that: The data correlation coefficient between the first process parameter scoring vector and the second process parameter scoring vector is determined based on the absolute value of the Pearson correlation coefficient between the first process parameter scoring vector and the second process parameter scoring vector, and the cross correlation coefficient between the first process parameter relationship vector and the second process parameter relationship vector is determined based on the cosine similarity between the first process parameter relationship vector and the second process parameter relationship vector.

10. The method for processing process data of an insulated cable according to claim 1, characterized in that: The process performance association analysis model is configured as a random forest model.

Images