Copper foil preparation process optimization method, system and equipment based on machine learning

Abstract

The present application relates to the field of data processing technology, and in particular to a method, system, and device for optimizing a copper foil preparation process based on machine learning, comprising collecting historical preparation data, historical quality data, and probability values ​​of all copper foil preparation equipment as first sample data, and historical real-time preparation data and historical real-time quality data of a target object as second sample data, respectively training a first model and a second model, periodically collecting the real-time preparation data of the target object, inputting the first model to obtain a first prediction result and probability value, inputting the second model to obtain a second prediction result, and then combining the third model to obtain a target prediction result. When the standard deviation between the target prediction result and the copper foil preparation is greater than a preset value, adjusting the preparation data to obtain the optimal preparation parameters. The present invention solves the problem of low precision of process optimization parameters due to insufficient sample data when the copper foil preparation equipment is not put into use for a long enough time, thereby achieving the effect of improving the precision of process optimization parameters during the production process.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method, system, and equipment for optimizing a copper foil preparation process based on machine learning. Background Art

[0002] As the market continues to demand higher performance from copper foil, including stricter control over thickness uniformity, surface roughness, and mechanical properties, traditional process optimization methods are increasingly unable to keep up. When faced with new quality standards and product demands, finding the right process parameters often requires significant time and effort through repeated trials and exploration, resulting in low efficiency and high costs. A similar prior art includes a Chinese application with publication number CN118446084A, which proposes a method, system and equipment for optimizing the process parameters of electrolytic copper foil, including the following steps: collecting the process parameter data of the electrolytic copper foil and determining the boundaries of each process parameter; constructing a material mechanical property prediction model for the electrolytic copper foil; optimizing the process parameters through an intelligent evolutionary algorithm to determine the optimal electrolytic copper foil process parameters; verifying the obtained optimal electrolytic copper foil process parameters through mechanical property experiments, optimizing the process parameters by constructing an intelligent evolutionary algorithm based on the material mechanical property prediction model, and obtaining the optimal process parameters with the goal of improving the mechanical properties of the electrolytic copper foil, and conducting material mechanical property verification experiments under the optimal process parameters, thereby solving the problem of optimizing the key process parameters in the preparation of electrolytic copper foil. In addition, similar prior art includes a Chinese application with publication number CN118446084A, which proposes a copper smelting optimization method, device, electronic device and storage medium, relating to the field of copper smelting technology. The method includes: constructing an initial data set, which is obtained by combining sample decision parameters with multiple copper smelting indicators respectively; inputting the initial data set into a corresponding copper smelting agent model, and outputting a target prediction value used to characterize the copper smelting indicator prediction result, wherein the copper smelting agent model is determined for each copper smelting indicator based on a deep Gaussian process and Gaussian noise; multiple merged prediction groups are constructed based on the target prediction values ​​corresponding to each copper smelting indicator, and a target solution set is determined. When a new merged prediction group is added to the target solution set, the optimal solution of the merged prediction group for copper smelting optimization and its corresponding optimal solution of decision parameters are determined based on the expected super volume improvement. This can achieve copper smelting optimization under multiple competing objectives and obtain the optimal process parameter combination in real time. Although the above two patent documents solve the problem of parameter optimization in the production process, and both use prediction models to predict the quality data in the production process to derive the optimal parameters, for production equipment with a short production time, there is not enough sample data, which makes the accuracy of the above quality data low, and therefore it is impossible to obtain more accurate process optimization parameters. Summary of the Invention

[0003] This application provides a method, system, and device for optimizing the copper foil production process based on machine learning, which is used to improve the accuracy of process optimization parameters in the production process when the copper foil production equipment has not been invested in for a long time. In a first aspect, this application provides a method for optimizing the copper foil production process based on machine learning, the method comprising:

[0004] Collecting first sample data of all copper foil preparation equipment during the copper foil preparation process, the first sample data including historical preparation data and corresponding historical quality data and probability distribution, preprocessing the sample data, and training a first model based on the preprocessed first sample data;

[0005] Also collecting second sample data of the target object, and training a second model based on the preprocessed second sample data, wherein the second sample data includes historical real-time preparation data and corresponding historical real-time quality data of the target object;

[0006] Periodically collecting real-time preparation data of the target object, inputting the real-time preparation data into the first model to obtain a first prediction result and a corresponding probability value, further inputting the real-time preparation data into the second model to obtain a second prediction result, and obtaining a target prediction result corresponding to a preset time period of the target object based on the first prediction result and the corresponding probability value, the second prediction result, and the third model;

[0007] When the difference between the target prediction result and the copper foil preparation standard is greater than a preset difference, the preparation data is adjusted and optimal preparation parameters are obtained based on the first model and the second model.

[0008] As a preferred technical solution of the present invention, preprocessing the first sample data and training the first model based on the preprocessed first sample data includes:

[0009] Dividing the historical production data of all the copper foil preparation equipment in the copper foil preparation process into groups according to the historical preparation data, wherein the historical preparation data in each group is the same, and the historical production data includes a plurality of the historical preparation data and the corresponding historical quality data;

[0010] Calculating the probability distribution of each of the historical quality data corresponding to each of the groups, and using the historical quality data with the highest probability value and the probability value in the probability distribution as a label of the historical preparation data corresponding to the group;

[0011] The historical preparation data and labels corresponding to each of the groups are used as the preprocessed first sample data, and the first model is trained using the first sample data.

[0012] As a preferred technical solution of the present invention, obtaining a target prediction result corresponding to a preset time period of the target object based on the first prediction result and the probability value of the first prediction result, the second prediction result, and the third model includes:

[0013] periodically collecting the real-time preparation data of the target object, inputting the real-time preparation data into the first model, and obtaining the first prediction result and the probability value of the first prediction result, and further inputting the real-time preparation data into the second model to obtain a second prediction result;

[0014] When the probability value of the first prediction result is greater than or equal to a set threshold, weighting the first prediction result and the second prediction result according to the probability value to obtain the target prediction result, wherein the larger the probability value, the greater the weight of the first prediction result, and the smaller the weight of the second prediction result;

[0015] When the probability value of the first prediction result is less than the set threshold, the real-time preparation data, the first prediction result and the second prediction result are input into the third model to obtain a target prediction result.

[0016] As a preferred technical solution of the present invention, the training of the second model further includes:

[0017] adding every N pieces of the real-time preparation data and the corresponding real-time quality data to the second sample data, and periodically training the second model using the second sample data;

[0018] The acquisition of the target prediction result further includes:

[0019] Calculate the sample data ratio of the first quantity of the second sample data to the second quantity of the first sample data, and adjust the weights of the first prediction result and the second prediction result according to the sample data ratio, wherein the larger the sample data ratio, the greater the weight of the second prediction result, and the smaller the weight of the first prediction result.

[0020] As a preferred technical solution of the present invention, the process of obtaining the third model includes:

[0021] The historical real-time preparation data of the target object is input into the first model to obtain a first historical prediction result and a corresponding probability value, and the historical real-time preparation data is also input into the second model to obtain a second historical prediction result, and each historical real-time preparation data of the target object and the corresponding first historical prediction result, second historical prediction result and historical real-time quality data are used as third sample data, wherein the probability value corresponding to the first historical prediction result is less than a set threshold, and the third model is trained using the third sample data.

[0022] As a preferred technical solution of the present invention, the acquisition of the optimal preparation parameters includes:

[0023] When the difference between the target prediction result and the copper foil preparation standard is greater than the preset difference, a plurality of simulated preparation data are obtained by a Monte Carlo algorithm within a preset range of the real-time preparation data corresponding to the target prediction result, and the first model, the second model and the third model are input according to each of the simulated preparation data to obtain the target simulation result, and the best simulation result is selected from the multiple target simulation results, and the simulated preparation data corresponding to the best simulation result is used as the target optimized preparation data, and the best preparation parameters are obtained according to the correlation between the target optimized preparation data and the parameters of the target object.

[0024] As a preferred technical solution of the present invention, the real-time preparation data, the historical preparation data and the historical real-time preparation data all include electrolyte concentration, current density, electrolyte temperature, drum rotation speed and environmental cleanliness, temperature and humidity;

[0025] The real-time quality data, the historical quality data and the historical real-time quality data all include at least the thickness and uniformity of the copper foil, the surface roughness and the mechanical properties.

[0026] The present invention also provides a copper foil preparation process optimization system based on machine learning, which is used to perform the above method, and the system includes:

[0027] A collecting unit, configured to collect first sample data of all copper foil preparation devices during the copper foil preparation process and second sample data of the target object, and further configured to periodically collect real-time preparation data of the target object;

[0028] a model training unit, configured to preprocess the second sample data and train the first model based on the preprocessed first sample data, wherein the first sample data includes historical preparation data and corresponding historical quality data and probability values; and further configured to train the second model based on the preprocessed second sample data, wherein the second sample data includes historical real-time preparation data and corresponding historical real-time quality data of the target object;

[0029] a prediction unit, configured to input the real-time preparation data into the first model to obtain a first prediction result and a corresponding probability value, further input the preparation data into the second model to obtain a second prediction result, and obtain a target prediction result corresponding to a preset time period based on the first prediction result and the corresponding probability value, the second prediction result, and the third model, wherein when the probability value is less than a set threshold, the target prediction result is obtained through the third model;

[0030] A simulation unit is used to obtain optimal preparation parameters by simulating the real-time preparation data and based on the first model, the second model and the third model when the difference between the target prediction result and the copper foil preparation standard is greater than a preset difference.

[0031] The present invention also provides a copper foil preparation process optimization device based on machine learning, the device comprising: a memory and at least one processor, wherein the memory stores instructions;

[0032] The at least one processor calls the instructions in the memory to enable the copper foil preparation process optimization device based on machine learning to perform the above method.

[0033] The present invention also provides a computer-readable storage medium having instructions stored thereon, and the above-mentioned method is implemented when the instructions are executed by a processor.

[0034] Effect

[0035] The present invention collects historical preparation data, historical quality data and probability values ​​of all copper foil preparation equipment as the first sample data, and historical real-time preparation data and historical real-time quality data of the target object as the second sample data, and respectively trains the first model and the second model, thereby establishing an accurate prediction model. In the actual production process, the real-time preparation data of the target object is periodically collected, input into the first model to obtain the first prediction result and probability value, input into the second model to obtain the second prediction result, and then combine with the third model to obtain the target prediction result. When the target prediction result and the standard deviation of the copper foil preparation are greater than the preset value, the preparation data is simulated and the optimal preparation parameters are obtained based on the three models, thereby achieving the goal of copper foil. The precise optimization of the preparation process, through the mutual coordination of the above-mentioned technical solutions, effectively solves the problem of low accuracy of process optimization parameters due to insufficient sample data when the copper foil preparation equipment is not invested long enough. It can make full use of limited sample data and improve the accuracy of process optimization parameters in the production process through model training and prediction, thereby improving the preparation quality of copper foil and meeting the market's increasingly stringent requirements for copper foil performance. At the same time, this technology also has strong flexibility and adaptability, and can dynamically adjust preparation parameters according to different production conditions and quality standards, providing copper foil manufacturers with an efficient and intelligent process optimization solution, which helps to reduce production costs, improve production efficiency and product quality stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0037] Figure 1 This is a flow chart of a method for optimizing a copper foil preparation process based on machine learning in an embodiment of the present application;

[0038] Figure 2 This is a flowchart of the first sample data preprocessing and first model training in an embodiment of the present application;

[0039] Figure 3 This is a flow chart of the process of obtaining the target prediction result corresponding to the preset time period of the target object in the embodiment of the present application;

[0040] Figure 4 This is a structural diagram of the copper foil preparation process optimization system based on machine learning in an embodiment of the present application. DETAILED DESCRIPTION

[0041] The embodiments of the present application provide a method, system, and apparatus for optimizing a copper foil preparation process based on machine learning. The terms "first," "second," "third," "fourth," and the like (if any) in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or precedence. It should be understood that the data used in this manner are interchangeable where appropriate so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to these processes, methods, products, or apparatus.

[0042] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 As shown, an embodiment of the copper foil preparation process optimization method based on machine learning in the embodiment of the present application includes:

[0043] Step S1: collecting first sample data of all copper foil preparation equipment during the copper foil preparation process, preprocessing the first sample data, and training a first model based on the preprocessed first sample data, wherein the first sample data includes historical preparation data and corresponding historical quality data and probability values ​​of the historical quality data;

[0044] Specifically, when the target object has not been in use for a long enough time and the sample data corresponding to the target object is not sufficient, the copper foil quality data within the preset time period cannot be accurately obtained through the second model of the target object. Therefore, the first sample data is obtained by collecting the historical production data of all other copper foil preparation equipment in the copper foil preparation process, and the first model is trained with the above-mentioned first sample data, and then the first model is used to predict and assist in obtaining the above-mentioned target prediction result. Since different copper foil preparation equipment has individual differences, even under the conditions of the same historical preparation data, the corresponding historical quality data will be different. Therefore, the historical production data corresponding to the same historical preparation data are grouped, and the probability distribution of the historical quality data in the same group is obtained, and the above-mentioned historical quality data with the highest probability value and the above-mentioned probability value are used as the labels of the above-mentioned historical preparation data, and the above-mentioned historical preparation data and the corresponding labels are used as the above-mentioned first sample data, and the above-mentioned first model is trained. Through the above-mentioned technical solution, the quality data of the target object corresponding to the preset time period that can be universally predicted, that is, the first prediction result, can be obtained.

[0045] Step S2: collecting second sample data of the target object and training a second model based on the preprocessed second sample data, wherein the second sample data includes historical real-time preparation data and corresponding historical real-time quality data of the target object;

[0046] Specifically, by performing preprocessing such as outlier elimination and feature value extraction on the historical real-time preparation data and the corresponding historical real-time quality data of the above-mentioned target object, and using the preprocessed historical real-time preparation data and the corresponding historical real-time quality data as the above-mentioned second sample data to train the above-mentioned second model, through the above-mentioned technical solution, a prediction model suitable for the above-mentioned target object can be obtained, laying the foundation for cooperating with the above-mentioned first model to predict the quality data of the above-mentioned target object within a preset time period.

[0047] Step S3: Periodically collect real-time preparation data of the target object, input the real-time preparation data into the first model to obtain a first prediction result and a corresponding probability value, further input the real-time preparation data into the second model to obtain a second prediction result, and obtain a target prediction result for the target object within a preset time period based on the first prediction result and the corresponding probability value, the second prediction result, and the third model;

[0048] Specifically, by periodically collecting the above-mentioned real-time preparation data of the target object, and inputting the above-mentioned real-time preparation data into the above-mentioned first model, obtaining the above-mentioned first prediction result and the corresponding above-mentioned probability value, and also inputting the above-mentioned real-time preparation data into the above-mentioned second model, obtaining the above-mentioned second prediction result, the larger the above-mentioned probability value is, the better the convergence of the quality data of the above-mentioned copper foil preparation equipment under the above-mentioned real-time preparation data conditions, and vice versa. In order to obtain a more accurate target prediction value of the above-mentioned target object, when the above-mentioned probability value is greater than or equal to the above-mentioned set threshold, that is, when the copper foil preparation equipment has better convergence under the above-mentioned real-time preparation data conditions, the above-mentioned first prediction result and the above-mentioned second prediction result are weighted, and in the above-mentioned first prediction When the probability value of the test result is less than the above-mentioned set threshold, the above-mentioned real-time preparation data, the above-mentioned first prediction result and the above-mentioned second prediction result are input into the above-mentioned third model, and the target prediction result is obtained, wherein the above-mentioned third model is based on the historical real-time preparation data of the above-mentioned target object and the historical real-time quality data corresponding to the above-mentioned historical real-time preparation data, the historical first prediction result and the historical second prediction result corresponding to the above-mentioned historical real-time preparation data, and the probability value corresponding to the above-mentioned historical first prediction result is less than the set threshold. Through the above-mentioned technical solution, even when the second sample data of the above-mentioned target object is insufficient, an accurate target prediction value can be obtained to lay the foundation for process optimization of the target object in the copper foil preparation process.

[0049] Step S4: when the difference between the target prediction result and the copper foil preparation standard is greater than a preset difference, obtaining optimal preparation parameters by simulating preparation data and based on the first model, the second model and the third model.

[0050] Specifically, when the difference between the above-mentioned target prediction result and the above-mentioned copper foil preparation standard is greater than the above-mentioned preset difference, it means that the quality of the copper foil produced by the above-mentioned target object within the above-mentioned preset time period may be unqualified or substandard, and the preparation parameters of the above-mentioned target object need to be optimized and adjusted. Within the preset range of the above-mentioned real-time preparation data, multiple simulated preparation data are obtained through the above-mentioned Monte Carlo algorithm, and each of the above-mentioned simulated preparation data is input into the above-mentioned first model and the above-mentioned second model, and the first simulation result and the corresponding probability value and the second simulation result are obtained. According to the size of the above-mentioned probability value, each of the above-mentioned simulated preparation data is input into the above-mentioned third model, and the target simulation result is obtained. The above-mentioned best simulation result is selected from the multiple above-mentioned target simulation results according to the above-mentioned copper foil preparation standard, and the above-mentioned best preparation parameters are obtained based on the correlation between the above-mentioned simulated preparation data corresponding to the above-mentioned best simulation result and the above-mentioned target object parameters. Through the above-mentioned technical solution, even when the above-mentioned target object is put into production for a short time and the sample data is insufficient, accurate optimal preparation parameters can be obtained, thereby improving the preparation quality of copper foil.

[0051] Furthermore, the first sample data is preprocessed, and the first model is trained based on the preprocessed first sample data, such as Figure 2 Shown, including:

[0052] Step S11: dividing the historical production data of all the copper foil preparation equipment in the copper foil preparation process into groups according to the historical preparation data, wherein the historical preparation data in each group is the same, and the historical production data includes a plurality of the historical preparation data and the corresponding historical quality data;

[0053] Step S12: Calculate the probability distribution of each of the historical quality data corresponding to each group, and use the historical quality data with the highest probability value and the probability value in the probability distribution as the label of the historical preparation data corresponding to the group;

[0054] Step S13: using the historical prepared data and labels corresponding to each of the groups as the preprocessed first sample data, and training the first model using the first sample data.

[0055] Specifically, the first sample data is obtained by collecting the historical production data of all other copper foil preparation equipment in the copper foil preparation process, and the first model is trained with the first sample data, and then the first model is used to predict and assist in obtaining the above-mentioned target prediction result, wherein the above-mentioned historical production data is the historical preparation data and corresponding historical quality data of other copper foil preparation equipment different from the above-mentioned target object in the copper foil preparation process. Since different copper foil preparation equipment have individual differences, even under the conditions of the same historical preparation data, the corresponding historical quality data will be different. Therefore, the historical production data corresponding to the same historical preparation data are grouped and the same historical production data are obtained. The probability distribution of historical quality data in a group, and the historical quality data with the highest probability value and the probability value are used as labels of the historical preparation data, and the historical preparation data and the corresponding labels are used as the first sample data, and the first model is trained, wherein the historical preparation data at least include the concentration of the electrolyte, the current density, the electrolyte temperature, the rotation speed of the drum and the cleanliness, temperature and humidity of the environment, and the historical quality data at least include the thickness and uniformity of the copper foil, the surface roughness and the mechanical properties. Through the above technical solution, the quality data of the target object corresponding to the preset time period, i.e., the first prediction result, can be obtained which can be universally predicted.

[0056] Furthermore, based on the first prediction result and the probability value of the first prediction result, the second prediction result and the third model, the target prediction result corresponding to the preset time period of the target object is obtained, such as Figure 3 Shown, including:

[0057] periodically collecting the real-time preparation data of the target object, inputting the real-time preparation data into the first model, and obtaining the first prediction result and the probability value of the first prediction result, and further inputting the real-time preparation data into the second model to obtain a second prediction result;

[0058] When the probability value of the first prediction result is greater than or equal to a set threshold, weighting the first prediction result and the second prediction result according to the probability value to obtain the target prediction result, wherein the larger the probability value, the greater the weight of the first prediction result, and the smaller the weight of the second prediction result;

[0059] When the probability value of the first prediction result is less than the set threshold, the real-time preparation data, the first prediction result and the second prediction result are input into the third model to obtain a target prediction result.

[0060] Specifically, by periodically collecting the above-mentioned real-time preparation data of the target object, wherein the above-mentioned target object is the target copper foil preparation equipment, and inputting the above-mentioned real-time preparation data into the above-mentioned first model, obtaining the above-mentioned first prediction result and the corresponding above-mentioned probability value, and also inputting the above-mentioned real-time preparation data into the above-mentioned second model, obtaining the above-mentioned second prediction result, since the above-mentioned first prediction result reflects the quality data of other copper foil preparation equipment under the above-mentioned real-time preparation data conditions, the above-mentioned probability value reflects the probability distribution value of the above-mentioned quality data under the above-mentioned real-time preparation data conditions, the larger the above-mentioned probability value is, the better the convergence of the quality data of the above-mentioned copper foil preparation equipment under the above-mentioned real-time preparation data conditions, and vice versa, the convergence is poor, since the above-mentioned second model is trained by the second sample data of the above-mentioned target object, it can more accurately predict the target prediction value of the above-mentioned target object within the preset time period, but because the number of the second sample data is not large enough, the accuracy is not high enough. In order to obtain a more accurate target prediction value of the above-mentioned target object, when the above-mentioned probability value is greater than or equal to the above-mentioned set threshold, that is, when the copper foil preparation equipment has a better convergence under the above-mentioned real-time preparation data conditions, for example: the above-mentioned set threshold is 0.9, the above-mentioned first prediction result and the above-mentioned second prediction result are weighted, wherein, The above-mentioned probability value is used as the weight of the above-mentioned first prediction result, and the sum of the weights of the first prediction result and the second prediction result is 1. When the probability value of the above-mentioned first prediction result is less than the above-mentioned set threshold, that is, the convergence of the quality data corresponding to the above-mentioned real-time preparation data is poor, therefore, the above-mentioned target prediction value of the above-mentioned target object cannot be accurately obtained only through the above-mentioned first prediction result and the above-mentioned second prediction result. Although the above-mentioned first prediction result and the above-mentioned second prediction result have a certain correlation with the real-time quality data corresponding to the above-mentioned real-time preparation data, the above-mentioned real-time preparation data, the above-mentioned first prediction result and the above-mentioned second prediction result are input into the above-mentioned third model, and the target prediction result is obtained, wherein the above-mentioned third model is based on the historical real-time preparation data of the above-mentioned target object and the historical real-time quality data corresponding to the above-mentioned historical real-time preparation data, the first historical prediction result and the second historical prediction result corresponding to the above-mentioned historical real-time preparation data, and the probability value corresponding to the above-mentioned first historical prediction result is less than the above-mentioned set threshold. Through the above-mentioned technical solution, even when the second sample data of the above-mentioned target object is insufficient, an accurate target prediction value can be obtained, laying a foundation for process optimization of the target object in the copper foil preparation process.

[0061] Furthermore, the training of the second model further includes:

[0062] adding every N pieces of the real-time preparation data and the corresponding real-time quality data to the second sample data, and periodically training the second model using the second sample data;

[0063] The acquisition of the target prediction result further includes:

[0064] Calculate the sample data ratio of the first quantity of the second sample data to the second quantity of the first sample data, and adjust the weights of the first prediction result and the second prediction result according to the sample data ratio, wherein the larger the sample data ratio, the greater the weight of the second prediction result, and the smaller the weight of the first prediction result.

[0065] Specifically, as the target object is put into use for a longer time, more real-time preparation data is collected. Therefore, the N real-time preparation data and corresponding real-time quality data collected are added to the second sample data at each set interval, and the second model is trained to improve the prediction accuracy of the second model. The ratio of the second sample data to the first sample data is calculated, and the ratio is used as the sample data ratio. The larger the data volume of the second sample data, that is, the larger the sample data ratio, the higher the accuracy of the second model. Therefore, when obtaining the target prediction result through the first prediction result and the second prediction result, the larger the sample data ratio, the greater the weight of the second prediction result is, and the weight is reduced based on the weight of the first prediction result. Through the above technical solution, the accuracy of the target prediction result can be further improved.

[0066] Furthermore, the process of obtaining the third model includes:

[0067] The historical real-time preparation data of the target object is input into the first model to obtain a first historical prediction result and a corresponding probability value, and the historical real-time preparation data is also input into the second model to obtain a second historical prediction result, and each historical real-time preparation data of the target object and the corresponding first historical prediction result, second historical prediction result and historical real-time quality data are used as third sample data, wherein the probability value corresponding to the first historical prediction result is less than a set threshold, and the third model is trained using the third sample data.

[0068] Specifically, the historical real-time preparation data of the target object is input into the first model to obtain the first historical prediction result and the corresponding probability value. When the probability value is less than the set threshold, that is, when the quality data corresponding to the historical real-time preparation data has poor convergence, the historical real-time preparation data is also input into the second model to obtain the second historical prediction result. The historical real-time preparation data, the first historical prediction result, the second historical prediction result and the historical real-time quality data corresponding to the historical real-time preparation data are used as the third sample data, and the third model is trained by the third sample data. Through the technical solution, the third model is trained for the third sample data with poor convergence of the corresponding quality data, which not only improves the accuracy of the target prediction result, but also reduces the amount of training data and improves the training quality.

[0069] Furthermore, the acquisition of the optimal preparation parameters includes:

[0070] When the difference between the target prediction result and the copper foil preparation standard is greater than the preset difference, a plurality of simulated preparation data are obtained by a Monte Carlo algorithm within a preset range of the real-time preparation data corresponding to the target prediction result, and the first model, the second model and the third model are input according to each of the simulated preparation data to obtain the target simulation result, and the best simulation result is selected from the multiple target simulation results, and the simulated preparation data corresponding to the best simulation result is used as the target optimized preparation data, and the best preparation parameters are obtained according to the correlation between the target optimized preparation data and the target object parameters.

[0071] Specifically, when the difference between the target prediction result and the copper foil preparation standard is greater than the preset difference, it indicates that the quality of the copper foil produced by the target object within the preset time period may be unqualified or substandard, and the preparation parameters of the target object need to be optimized and adjusted. Since the real-time preparation data corresponding to the target prediction result is the preparation data closest to the real-time quality data that meets the copper foil preparation standard, a plurality of simulated preparation data are obtained through the Monte Carlo algorithm within the preset range of the real-time preparation data, and each of the simulated preparation data is input into the first model and the second model, and the first simulation result and the corresponding probability value and the second simulation result are obtained, and according to the size of the probability value, Each of the above-mentioned simulated preparation data is input into the above-mentioned third model, and the target simulation result is obtained, wherein the method for obtaining the above-mentioned target simulation result is the same as the method for obtaining the above-mentioned target prediction result, and the above-mentioned best simulation result is selected from the above-mentioned target simulation results corresponding to the multiple above-mentioned simulated preparation data according to the above-mentioned copper foil preparation standard, and the above-mentioned best preparation parameters are obtained based on the correlation between the above-mentioned simulated preparation data corresponding to the above-mentioned best simulation result, that is, the above-mentioned target optimized preparation data and the above-mentioned target object parameters, and the above-mentioned target object is configured based on the above-mentioned best preparation parameters. Through the above-mentioned technical solution, even when the above-mentioned target object is invested in a short time and the sample data is insufficient, accurate best preparation parameters can be obtained, thereby improving the quality of copper foil preparation.

[0072] Furthermore, the real-time preparation data, the historical preparation data and the historical real-time preparation data all include electrolyte concentration, current density, electrolyte temperature, drum rotation speed and environmental cleanliness, temperature and humidity;

[0073] The real-time quality data, the historical quality data and the historical real-time quality data all include at least the thickness and uniformity of the copper foil, the surface roughness and the mechanical properties.

[0074] The present invention also provides a copper foil preparation process optimization system based on machine learning, which is used to implement the above method, such as Figure 4 As shown, the system includes:

[0075] A collecting unit, configured to collect first sample data of all copper foil preparation devices during the copper foil preparation process and second sample data of the target object, and further configured to periodically collect real-time preparation data of the target object;

[0076] a model training unit, configured to preprocess the second sample data and train the first model based on the preprocessed first sample data, wherein the first sample data includes historical preparation data and corresponding historical quality data and probability values; and further configured to train the second model based on the preprocessed second sample data, wherein the second sample data includes historical real-time preparation data and corresponding historical real-time quality data of the target object;

[0077] a prediction unit, configured to input the real-time prepared data into the first model to obtain a first prediction result and a corresponding probability value, further input the real-time prepared data into the second model to obtain a second prediction result, and obtain a target prediction result corresponding to a preset time period of the target object based on the first prediction result and the corresponding probability value, the second prediction result, and the third model;

[0078] A simulation unit is used to obtain optimal preparation parameters by simulating the real-time preparation data and based on the first model, the second model and the third model when the difference between the target prediction result and the copper foil preparation standard is greater than a preset difference.

[0079] The present invention also provides a copper foil preparation process optimization device based on machine learning, the device comprising: a memory and at least one processor, wherein the memory stores instructions;

[0080] The at least one processor calls the instructions in the memory to enable the copper foil preparation process optimization device based on machine learning to perform the above method.

[0081] The present invention also provides a computer-readable storage medium having instructions stored thereon, and the above-mentioned method is implemented when the instructions are executed by a processor.

[0082] In summary, the present invention collects the historical preparation data, historical quality data and probability values ​​of all copper foil preparation equipment as the first sample data, and the historical real-time preparation data and historical real-time quality data of the target object as the second sample data, and trains the first model and the second model respectively, thereby establishing an accurate prediction model. In the actual production process, the real-time preparation data of the target object is periodically collected, and the first prediction result and probability value are obtained by inputting the first model, and the second prediction result is obtained by inputting the second model, and then the target prediction result is obtained by combining the third model. When the standard deviation between the target prediction result and the copper foil preparation is greater than the preset value, the optimal preparation parameters are obtained by simulating the preparation data and based on the three models, thereby achieving The precise optimization of the copper foil preparation process, through the mutual coordination of the above-mentioned technical solutions, effectively solves the problem of low accuracy of process optimization parameters due to insufficient sample data when the copper foil preparation equipment is not invested long enough. It can make full use of limited sample data, and through model training and prediction, improve the accuracy of process optimization parameters in the production process, thereby improving the preparation quality of copper foil and meeting the market's increasingly stringent requirements for copper foil performance. At the same time, this technology also has strong flexibility and adaptability, and can dynamically adjust preparation parameters according to different production conditions and quality standards, providing copper foil production companies with an efficient and intelligent process optimization solution, which helps to reduce production costs, improve production efficiency and product quality stability.

[0083] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0084] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program code.

[0085] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A copper foil preparation process optimization method based on machine learning, characterized in that: The method comprises: Collecting first sample data of all copper foil preparation equipment during the copper foil preparation process, preprocessing the first sample data, and training a first model based on the preprocessed first sample data, wherein the first sample data includes historical preparation data and corresponding historical quality data and probability values ​​of the historical quality data; Also collecting second sample data of the target object, and training a second model based on the preprocessed second sample data, the second sample data including historical real-time preparation data and corresponding historical real-time quality data of the target object; Periodically collecting real-time preparation data of the target object, inputting the real-time preparation data into the first model to obtain a first prediction result and a corresponding probability value, further inputting the real-time preparation data into the second model to obtain a second prediction result, and obtaining a target prediction result for the target object within a preset time period based on the first prediction result and the corresponding probability value, the second prediction result, and the third model; When the difference between the target prediction result and the copper foil preparation standard is greater than a preset difference, obtaining optimal preparation parameters by simulating preparation data and based on the first model, the second model and the third model; Obtaining target prediction results, including: periodically collecting the real-time preparation data of the target object, inputting the real-time preparation data into the first model, and obtaining the first prediction result and the probability value of the first prediction result, and further inputting the real-time preparation data into the second model to obtain a second prediction result; When the probability value of the first prediction result is greater than or equal to a set threshold, weighting the first prediction result and the second prediction result according to the probability value to obtain the target prediction result, wherein the larger the probability value, the greater the weight of the first prediction result, and the smaller the weight of the second prediction result; When the probability value of the first prediction result is less than the set threshold, the real-time preparation data, the first prediction result and the second prediction result are input into the third model to obtain a target prediction result.

2. The method according to claim 1, characterized in that Preprocessing the first sample data and training the first model based on the preprocessed first sample data includes: Dividing the historical production data of all the copper foil preparation equipment in the copper foil preparation process into groups according to the historical preparation data, wherein the historical preparation data in each group is the same, and the historical production data includes a plurality of the historical preparation data and the corresponding historical quality data; Calculating the probability distribution of each of the historical quality data corresponding to each of the groups, and using the historical quality data with the highest probability value and the probability value in the probability distribution as a label of the historical preparation data corresponding to the group; The historical preparation data and labels corresponding to each of the groups are used as the preprocessed first sample data, and the first model is trained using the first sample data.

3. The method according to claim 1, characterized in that The training of the second model further includes: adding every N pieces of the real-time preparation data and the corresponding real-time quality data to the second sample data, and periodically training the second model using the second sample data; The acquisition of the target prediction result further includes: Calculate the sample data ratio of the first quantity of the second sample data to the second quantity of the first sample data, and adjust the weights of the first prediction result and the second prediction result according to the sample data ratio, wherein the larger the sample data ratio, the greater the weight of the second prediction result, and the smaller the weight of the first prediction result.

4. The method according to claim 1, wherein The process of obtaining the third model includes: The historical real-time preparation data of the target object is input into the first model to obtain a first historical prediction result and a corresponding probability value, and the historical real-time preparation data is also input into the second model to obtain a second historical prediction result, and each historical real-time preparation data of the target object and the corresponding first historical prediction result, second historical prediction result and historical real-time quality data are used as third sample data, wherein the probability value corresponding to the first historical prediction result is less than a set threshold, and the third model is trained using the third sample data.

5. The method according to claim 1, wherein The acquisition of the optimal preparation parameters includes: When the difference between the target prediction result and the copper foil preparation standard is greater than the preset difference, a plurality of simulated preparation data are obtained by a Monte Carlo algorithm within a preset range of the real-time preparation data corresponding to the target prediction result, and the first model, the second model and the third model are input according to each of the simulated preparation data to obtain the target simulation result, and the best simulation result is selected from the multiple target simulation results, and the simulated preparation data corresponding to the best simulation result is used as the target optimized preparation data, and the best preparation parameters are obtained according to the correlation between the target optimized preparation data and the parameters of the target object.

6. The method according to claim 1, wherein The real-time preparation data, the historical preparation data and the historical real-time preparation data all include electrolyte concentration, current density, electrolyte temperature, drum rotation speed and environmental cleanliness, temperature and humidity; The real-time quality data, the historical quality data and the historical real-time quality data all include at least the thickness and uniformity of the copper foil, the surface roughness and the mechanical properties.

7. A copper foil preparation process optimization system based on machine learning, wherein the system is used to execute the method according to any one of claims 1 to 6, characterized in that: The system comprises: A collecting unit, configured to collect first sample data of all copper foil preparation devices during the copper foil preparation process and second sample data of the target object, and further configured to periodically collect real-time preparation data of the target object; a model training unit, configured to preprocess the second sample data and train the first model based on the preprocessed first sample data, wherein the first sample data includes historical preparation data and corresponding historical quality data and probability values; and further configured to train the second model based on the preprocessed second sample data, wherein the second sample data includes historical real-time preparation data and corresponding historical real-time quality data of the target object; a prediction unit, configured to input the real-time prepared data into the first model to obtain a first prediction result and a corresponding probability value, further input the real-time prepared data into the second model to obtain a second prediction result, and obtain a target prediction result corresponding to a preset time period of the target object based on the first prediction result and the corresponding probability value, the second prediction result, and the third model; A simulation unit is used to obtain optimal preparation parameters by simulating preparation data and based on the first model, the second model and the third model when the difference between the target prediction result and the copper foil preparation standard is greater than a preset difference.

8. A copper foil preparation process optimization device based on machine learning, characterized in that: The device includes: a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instructions in the memory to enable the copper foil preparation process optimization device based on machine learning to execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images