Lithium battery diaphragm process parameter self-correction method and system based on big data analysis

Abstract

The application provides a lithium battery diaphragm process parameter self-correction method and system based on big data analysis. The lithium battery diaphragm process parameter self-correction method based on big data analysis comprises the following steps: in the preparation process of the lithium battery diaphragm, the thickness values measured at multiple continuous time nodes after stretching of raw materials are obtained; a thickness data set corresponding to each time node is constructed, wherein the thickness data set comprises thickness data for representing the thickness conditions of each region of the raw material, and the thickness data is generated according to the thickness values; each group of thickness data sets is compared with other thickness data sets of adjacent time nodes, and a target thickness data set is determined in the thickness data set according to the data difference; and the process parameters of the lithium battery diaphragm are corrected according to the target thickness data set. The application has the beneficial effect that the process parameter correction in the lithium battery diaphragm manufacturing process is more accurate.

Description

Technical Field

[0001] This application relates to the field of lithium battery manufacturing technology, specifically to a self-calibration method and system for lithium battery separator process parameters based on big data analysis. Background Technology

[0002] Against the backdrop of the rapid development of the lithium battery industry, the quality of the lithium battery separator, as a core component of the battery, directly affects the battery's safety, cycle life, and energy density. In the separator manufacturing process, the raw material stretching step is a crucial stage that determines key properties such as separator thickness uniformity and porosity. This stage involves the coordinated control of multiple parameters, including stretching temperature, stretching rate, and clamping force. However, current lithium battery separator manufacturing processes commonly suffer from inaccurate process parameter calibration.

[0003] Traditional separator production relies on manual experience. Technicians primarily determine whether the thickness meets standards by periodically sampling separator samples and combining this with their own experience, thereby adjusting process parameters. This approach has significant drawbacks: for example, heating and stretching devices need to operate continuously for a period of time to reach stability during the stretching process. However, manual inspection often only collects thickness data at a single point in time to adjust the equipment, neglecting the fluctuations and changes in the operating status of production equipment during lithium battery production. This results in inaccurate calibration and adjustment of equipment process parameters. Summary of the Invention

[0004] The embodiments of this application provide a self-calibration method and system for lithium battery separator process parameters based on big data analysis, which can improve the technical problem of inaccurate calibration of process parameters in the lithium battery separator manufacturing process.

[0005] In a first aspect, embodiments of this application provide a self-calibration method for lithium battery separator process parameters based on big data analysis, comprising the following steps:

[0006] In the process of preparing lithium battery separators, the thickness values ​​of the raw materials are obtained at multiple consecutive time points after stretching.

[0007] Construct a thickness dataset corresponding to each time node, wherein the thickness dataset includes thickness data for characterizing the thickness of each region of the raw material, and the thickness data is generated based on the thickness value;

[0008] Compare each set of thickness datasets with other thickness datasets at adjacent time points, and determine the target thickness dataset from the thickness datasets based on the data differences;

[0009] The process parameters of the lithium battery separator are corrected based on the target thickness dataset.

[0010] In one embodiment, the process parameters include the clamping force of the clamping device and the rotation speed of the stretching device, and the correction of the process parameters of the lithium battery separator based on the target thickness dataset includes:

[0011] Based on the target thickness dataset, determine the first thickness variation of the raw material from its center to its edge in the transverse direction and the second thickness variation of the raw material in the longitudinal direction;

[0012] The clamping force of the clamping device and the rotation speed of the stretching device are corrected by combining the first thickness change and the second thickness change.

[0013] In one embodiment, the first thickness variation includes the lateral thickness data deviation value of two adjacent regions of the raw material in the lateral direction, and the second thickness variation includes the longitudinal thickness data deviation value of two adjacent regions of the raw material in the longitudinal direction. The step of correcting the clamping force of the clamping device and the rotational speed of the stretching device by combining the first thickness variation and the second thickness variation includes:

[0014] If the lateral thickness data deviation value is greater than a preset first deviation threshold, and the longitudinal thickness data deviation value is less than or equal to a preset second deviation threshold, then the clamping force of the clamping device is adjusted according to the lateral thickness data deviation value.

[0015] If the longitudinal thickness data deviation value is greater than the second deviation threshold, and the transverse thickness data deviation value is less than or equal to the first deviation threshold, then the rotational speed of the stretching device is corrected according to the longitudinal thickness data deviation value.

[0016] If the lateral thickness data deviation value is greater than the first deviation threshold and the longitudinal thickness data deviation value is greater than the second deviation threshold, then the clamping force of the clamping device and the rotation speed of the stretching device are adjusted in combination with the lateral thickness data deviation value and the longitudinal thickness data deviation value.

[0017] In one embodiment, correcting the clamping force of the clamping device based on the lateral thickness data deviation value includes:

[0018] Based on the lateral thickness data deviation value, all areas of abnormal thickness in the raw material are located;

[0019] Calculate the distance between abnormal regions between all the aforementioned thickness difference abnormal regions;

[0020] The initial value of the clamping force is determined based on the deviation value of the lateral thickness data and using regression analysis.

[0021] The initial value of the clamping force is corrected based on the distance to the abnormal area to obtain the target value of the clamping force;

[0022] The clamping force of the clamping device is corrected according to the target clamping force value.

[0023] In one embodiment, the step of combining the lateral thickness data deviation value and the longitudinal thickness data deviation value to collaboratively correct the clamping force of the clamping device and the rotational speed of the stretching device includes:

[0024] The lateral thickness data deviation value and the longitudinal thickness data deviation value are input into the pre-constructed deviation coupling correction model. The deviation coupling correction model outputs the clamping force correction amplitude and the rotation speed correction amplitude. The deviation coupling correction model is constructed based on a gradient boosting regression tree.

[0025] The clamping force of the clamping device and the rotational speed of the stretching device are corrected according to the clamping force correction range and the rotational speed correction range, respectively.

[0026] In one embodiment, comparing each set of thickness datasets with other thickness datasets at adjacent time points, and determining the target thickness dataset from the thickness datasets based on data differences, includes:

[0027] Determine the distance value in the data space between the thickness data in each set of thickness datasets and the thickness data in other thickness datasets at adjacent time points, wherein the distance value includes Euclidean distance or Manhattan distance;

[0028] The target thickness dataset is determined based on the distance value.

[0029] In one embodiment, the thickness data includes the average value and dispersion value of the thickness values ​​in each region, and the distance value includes a first distance value calculated from the average value corresponding to each region in each group of thickness datasets and other thickness datasets at adjacent time nodes, and a second distance value calculated from the dispersion value.

[0030] Determining the target thickness dataset based on the distance value includes:

[0031] A comprehensive distance index is determined based on the first distance value and the second distance value, wherein the formula for determining the comprehensive distance index is: D=w 1 D 1 +w 2 D 2 D1 is the first distance value, D2 is the second distance value, w1 is the first preset weight, w2 is the second preset weight, and w2 is greater than w1;

[0032] The target thickness dataset is determined based on the distance comprehensive index.

[0033] In one embodiment, the distance composite index includes a first distance composite index of the thickness dataset at the current time node relative to the thickness dataset at the previous time node, and a second distance composite index of the thickness dataset at the current time node relative to the thickness dataset at the next time node.

[0034] When both the first comprehensive distance index and the second comprehensive distance index are less than or equal to a preset index threshold, the thickness dataset at the current time node is determined to be the target thickness dataset.

[0035] In one embodiment, correcting the process parameters of the lithium battery separator based on the target thickness dataset includes:

[0036] Based on the average value corresponding to each region in the target thickness dataset, the thickness variation of the raw material along the transverse and longitudinal directions is determined;

[0037] When the thickness variation indicates that the thickness of the raw material varies uniformly in the same direction, compare whether the variation of the dispersion value of each region in the same direction matches the thickness variation.

[0038] If so, the equipment requiring process parameter correction is located based on the thickness variation and the dispersion value, and the process parameters are corrected accordingly.

[0039] If not, then repeat the step of determining the target thickness dataset.

[0040] Secondly, embodiments of this application provide a self-calibration system for lithium battery separator process parameters based on big data analysis, the self-calibration system for lithium battery separator process parameters based on big data analysis comprising:

[0041] One or more processors;

[0042] Memory;

[0043] and one or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the processor to implement the methods described above.

[0044] The beneficial effects of the embodiments of this application are as follows:

[0045] In the embodiments of this application, during the preparation of lithium battery separators, a large number of thickness values ​​reflecting the thickness of the stretched raw material are obtained at multiple consecutive time points after the raw material is stretched. Then, the thickness value data at each time point is processed to obtain a thickness dataset. The dataset includes thickness data reflecting the thickness of each region. The thickness data is generated from the thickness values ​​to reduce the amount of data. By comparing the differences in thickness data at adjacent time points, it is determined whether the equipment operation has reached a relatively stable state, thereby determining the target thickness dataset under stable conditions for data analysis. This enables the correction of the process parameters of the lithium battery separator, making parameter correction more accurate and faster, saving manpower and material costs, and preventing misjudgments. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a flowchart illustrating the self-calibration method for lithium battery separator process parameters based on big data analysis provided in this application embodiment. Figure 1 ;

[0048] Figure 2 This is a flowchart illustrating the self-calibration method for lithium battery separator process parameters based on big data analysis provided in this application embodiment. Figure 2 ;

[0049] Figure 3 This is a flowchart illustrating the self-calibration method for lithium battery separator process parameters based on big data analysis provided in this application embodiment. Figure 3 ;

[0050] Figure 4 This is a structural block diagram of a self-calibration system for lithium battery separator process parameters based on big data analysis, provided in the application embodiment. Detailed Implementation

[0051] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0052] In the description of this application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined as "first" or "second" may explicitly or implicitly include one or more features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0053] In this application, the term "exemplary" is used to mean "used as an example, illustration, or description." Any embodiment described as "exemplary" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use this application. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that this application can be made without using these specific details. In other instances, well-known structures and processes are not described in detail to avoid obscuring the description of this application with unnecessary detail. Therefore, this application is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0054] To address the issue of inaccurate process parameter calibration during lithium-ion battery separator manufacturing, this application provides a self-calibration method for lithium-ion battery separator process parameters based on big data analysis. Please refer to... Figure 1 , Figure 1 This is a flowchart illustrating a self-calibration method for lithium battery separator process parameters based on big data analysis provided in this application embodiment. The method includes the following steps:

[0055] Step 100: During the preparation of the lithium battery separator, obtain the thickness values ​​of the raw material at multiple consecutive time points after stretching.

[0056] In the manufacturing process of lithium-ion battery separators, stretching equipment continuously stretches the raw materials. For example, in the wet production process of lithium-ion battery separators, stretching includes longitudinal stretching and transverse stretching of the raw material film. During longitudinal stretching, after heating, the film is stretched by the pressure of the stretching rollers and the speed ratio of the front and rear rollers, so that the molecular chains are longitudinally oriented to achieve the specified thickness. Transverse stretching is a similar operation, in which the longitudinally stretched raw material film is heated and then stretched laterally by the pressure of the stretching rollers.

[0057] In this embodiment of the application, the thickness values ​​of the stretched raw material are obtained continuously multiple times. High-precision equipment such as laser thickness gauges and capacitance thickness gauges can be used to continuously scan and collect real-time thickness data of the diaphragm. A large number of thickness values ​​are collected at each time point to reflect the thickness of the stretched raw material in the current collection area.

[0058] Step 200: Construct the thickness dataset corresponding to each time point. The thickness dataset includes thickness data used to characterize the thickness of each region of the raw material. The thickness data is generated based on the thickness values.

[0059] It is understood that for a piece of stretched raw material, the thickness value collected includes data from multiple points. In this embodiment of the application, the large amount of thickness value data is processed, and a thickness dataset is generated from the thickness values ​​of multiple points to represent the overall thickness of the area covered by these points. This reduces the amount of data while still being able to accurately represent the thickness of the raw material.

[0060] In the stretching process of raw materials, there may be thickness deviations in the central area, the edge area, and the transition area between the center and the edge. Therefore, the raw material area can be divided according to the center, transition and edge, and the thickness data can characterize the overall thickness of these areas.

[0061] For the generation of thickness data, it can be derived from the fusion of the original measurement values, i.e., the thickness values, for example by calculating the average thickness value, so as to express the overall thickness of the region and reduce the amount of data for subsequent comparisons.

[0062] Step 300: Compare each set of thickness datasets with other thickness datasets at adjacent time points, and determine the target thickness dataset from the thickness datasets based on the data differences.

[0063] Step 400: Correct the process parameters of the lithium battery separator based on the target thickness dataset.

[0064] It is understandable that for the continuous stretching preparation of lithium battery separators, after a certain stretching time, the equipment is in a relatively stable working stage. For example, the heating of the heating equipment reaches a relatively stable stage, and the operation of the stretching rollers reaches a relatively stable stage. At this time, the thickness parameters of the stretched raw material are more likely to match the actual process parameters. Therefore, in this embodiment, for multiple sets of thickness datasets measured sequentially at time nodes, the thickness dataset of each time node is compared with the thickness dataset of adjacent time nodes. By judging the differences between the datasets, it is indirectly determined whether the operation of each device has reached stability. Thus, the thickness dataset of the raw material obtained when the equipment reaches stable operation is determined as the target thickness dataset. The target thickness dataset is then matched with the process parameters of the current equipment. When the target thickness dataset is abnormal, it can more accurately indicate that some process parameters are problematic. Therefore, by analyzing the target thickness dataset, the process parameters of the lithium battery separator can be more accurately corrected, which facilitates the production of more qualified lithium battery separators.

[0065] Therefore, the self-calibration method for lithium battery separator process parameters based on big data analysis in this application embodiment acquires a large number of thickness values ​​reflecting the thickness of the stretched raw material at multiple consecutive time points during the preparation of the lithium battery separator. Then, the thickness value data at each time point is processed to obtain a thickness dataset. The dataset includes thickness data reflecting the thickness of each region. The thickness data is generated from the thickness values ​​to reduce the amount of data. By comparing the differences in thickness data at adjacent time points, it is determined whether the equipment operation has reached a relatively stable state, thereby determining the target thickness dataset under stable conditions for data analysis. This enables the calibration of the process parameters of the lithium battery separator, making parameter calibration more accurate and faster, saving manpower and material costs, and preventing misjudgments.

[0066] In addition, in this embodiment of the application, the target thickness dataset that meets the requirements is determined in real time by acquiring thickness data from multiple consecutive time nodes. After the process parameters are corrected, the required material can be produced directly. This judgment process has higher real-time performance and accuracy. Compared with human judgment based on experience, it can save the raw material costs consumed in the early stage as much as possible, thereby reducing the overall material cost in the production process.

[0067] Reference Figure 2 As shown, in some embodiments of this application, the process parameters include the clamping force of the clamping device and the rotation speed of the stretching device. Correcting the process parameters of the lithium battery separator based on the target thickness dataset includes:

[0068] Step 411: Determine the first thickness variation of the raw material from the center to the edge in the transverse direction and the second thickness variation of the raw material in the longitudinal direction based on the target thickness dataset;

[0069] Step 412: Adjust the clamping force of the clamping device and the rotation speed of the stretching device based on the first and second thickness changes.

[0070] In this embodiment, for the stretching of the raw material, the transverse direction refers to the direction parallel to the horizontal plane and perpendicular to the material preparation transport direction, while the longitudinal direction refers to the material preparation transport direction. During the stretching process, the raw material is clamped by clamping devices on both sides and stretched and pressed by stretching rollers after heating. For the transverse direction of the raw material, that is, the width direction of the raw material, there may be a first thickness variation between the edge and the center area. When the thickness variation meets the requirements, it is within the allowable error range of the preparation. However, if the thickness variation is too large and exceeds the expectation, it indicates that the current preparation does not meet the requirements.

[0071] Typically, significant thickness variations between the edge and center are related to the clamping force of the clamping device. For example, a large difference in clamping force can cause a stretching deviation between the edge and center, resulting in a substantial thickness change. Therefore, in this embodiment, the clamping device requiring correction and its corresponding clamping force are determined based on the first thickness change, such as the magnitude and location of the thickness difference, thereby achieving precise adjustment.

[0072] In the embodiments of this application, the thickness data in the target thickness dataset is obtained by processing the actual thickness value data. Therefore, for the first thickness change of the raw material from the center to the edge in the transverse direction determined by the target thickness dataset, the difference in the data reflects more the magnitude of the thickness difference than the actual thickness difference. Therefore, when setting the error value for comparison and judgment with the first thickness change, it can be adjusted adaptively according to the actual situation and experience, which will not be elaborated here.

[0073] In addition, during the preparation of lithium battery separators, the thickness of the raw materials may fluctuate periodically or randomly along the length direction. If the thickness fluctuation exceeds the error in the longitudinal direction, it will not meet the production requirements.

[0074] For example, periodic thickness fluctuations are usually related to the periodic and regular deviations in the rotation speed of the stretching equipment. If the encoder signal of the drive motor is interfered with, causing the pulse signal to be lost periodically, the encoder pulse signal can be analyzed by an algorithm to eliminate abnormal pulses and correct the rotation speed calculation results to ensure accurate rotation speed feedback. This enables automatic correction of the rotation speed of the stretching equipment, thereby adjusting the thickness of the raw material obtained by subsequent stretching so that the second thickness change conforms to the error.

[0075] In some embodiments of this application, the self-calibration method for lithium battery separator process parameters based on big data analysis is characterized in that the first thickness change includes the lateral thickness data deviation value of two adjacent regions of the raw material in the lateral direction, and the second thickness change includes the longitudinal thickness data deviation value of two adjacent regions of the raw material in the longitudinal direction. The method for correcting the clamping force of the clamping device and the rotation speed of the stretching device by combining the first and second thickness changes includes:

[0076] If the lateral thickness data deviation is greater than the preset first deviation threshold and the longitudinal thickness data deviation is less than or equal to the preset second deviation threshold, the clamping force of the clamping device is adjusted according to the lateral thickness data deviation.

[0077] If the longitudinal thickness data deviation is greater than the second deviation threshold and the transverse thickness data deviation is less than or equal to the first deviation threshold, then the speed of the stretching equipment is adjusted according to the longitudinal thickness data deviation.

[0078] If the lateral thickness data deviation is greater than the first deviation threshold and the longitudinal thickness data deviation is greater than the second deviation threshold, then the clamping force of the clamping device and the rotation speed of the stretching device are adjusted in conjunction with the lateral thickness data deviation and the longitudinal thickness data deviation.

[0079] In this embodiment, since the lateral stretching and longitudinal stretching steps of the lithium battery separator are not completely independent, the initial longitudinal stretching alters the material state of the raw material, such as molecular orientation, local density, and ductility. These material states indirectly affect the subsequent lateral stretching. Therefore, when it is necessary to simultaneously correct the clamping force of the clamping device and the rotation speed of the stretching device, the synergistic effect of both on the lateral and longitudinal thickness data deviations, especially the lateral thickness data deviation, needs to be considered to prevent excessive correction of the clamping force during the lateral stretching process, which could lead to an excessively large lateral thickness data deviation. Therefore, if it is necessary to simultaneously correct the clamping force of the clamping device and the rotation speed of the stretching device, their synergistic effect must be considered.

[0080] If the lateral thickness deviation is greater than a preset first deviation threshold, and the longitudinal thickness deviation is less than or equal to a preset second deviation threshold, then it means that only the clamping force of the clamping device needs to be corrected to restore the lateral thickness deviation to normal. Therefore, it is not necessary to consider the synergistic effect of the clamping force and the rotation speed of the stretching device; only the rotation speed of the stretching device needs to be corrected based on the longitudinal thickness deviation. The specific correction steps include: identifying two adjacent areas where the lateral thickness deviation is greater than the preset first deviation threshold as thickness difference abnormal areas. After identifying the thickness difference abnormal areas, the edge of the raw material can be used as a reference to record the vertical distance between the center point of each thickness difference abnormal area and the raw material edge. Based on this vertical distance, the center point distance of each thickness difference abnormal area, i.e., the abnormal area distance, is further calculated. If two abnormal areas are close together, for example, if both sides of a thickness abnormal area are also thickness abnormal areas, then because the clamping force has a certain diffusion range, when it is adjacent to two abnormal areas, the force transmission will overlap or interfere with each other. Therefore, the adjustment range of the clamping force of the clamping device at the corresponding positions of the three thickness abnormal areas needs to be appropriately reduced to reduce interference. Historical lateral thickness deviation values ​​and corresponding historical clamping forces of the clamping equipment are obtained as sample data. The historical lateral thickness deviation values ​​are used as the dependent variable, and the historical clamping forces are used as the independent variable. Nonlinear regression analysis is performed to obtain the regression equation between the lateral thickness deviation values ​​and the clamping forces. The lateral thickness deviation values ​​between the current thickness anomaly area and its adjacent thickness anomaly areas are input into the regression equation to obtain the initial clamping force value of the clamping equipment. If there are adjacent thickness anomaly areas within the current thickness anomaly area, the initial clamping force value needs to be multiplied by a preset correction coefficient to obtain the target clamping force value. The correction coefficient can be determined through multiple clamping force adjustment experiments. Finally, the clamping force of the clamping equipment corresponding to the thickness anomaly area is corrected to the target clamping force value.

[0081] In the manufacturing process of lithium battery separators, the stretching of raw materials involves stretching equipment typically consisting of multiple spaced stretching rollers. Different rollers ultimately affect the thickness of different regions of the raw material differently. When the longitudinal thickness data deviation exceeds a preset deviation value, it not only indicates a large thickness difference between adjacent regions, making the fabrication unsuitable for actual production needs, but also suggests a problem with the rotational speed of the corresponding stretching rollers. For example, if the thickness data for an adjacent region shows one region is too thin and the other too thick, it may be due to the corresponding stretching roller rotating too fast or too slow, causing over- or under-stretching of the raw material in that region. In this case, by mapping the region to the stretching rollers, the target stretching equipment causing the deviation can be quickly located, precisely identifying the roller with the abnormal rotational speed and enabling rapid adjustment. This effectively improves the stability of the production process and the yield rate of the separator product, avoiding the decrease in production efficiency and waste of resources caused by excessively long manual inspection times. The adjustment range for correcting the stretching equipment rotational speed based on the longitudinal thickness data deviation can be determined through regression analysis or by establishing and training a neural network model. Taking regression analysis as an example, a regression model is established using historical longitudinal thickness data deviation values ​​and stretching equipment speed. By inputting the current longitudinal thickness data deviation value into the regression model, the target value of the stretching equipment speed can be directly obtained, and the stretching equipment speed can be corrected based on the target value of the stretching equipment speed.

[0082] If the transverse thickness deviation exceeds the first deviation threshold and the longitudinal thickness deviation exceeds the second deviation threshold, it indicates an anomaly in both the longitudinal and transverse stretching processes during the material stretching procedure. Generally, the material stretching process involves longitudinal stretching followed by transverse stretching. If the clamping force is adjusted solely based on the transverse thickness deviation, and the stretching equipment speed is adjusted based on the longitudinal thickness deviation, without considering the coupling effect between the clamping force and speed, the stretching equipment speed might be corrected to the target value. However, using this corrected speed for longitudinal stretching alters the stress state and microstructure of the material, requiring a different clamping force for subsequent transverse stretching. Since the clamping force adjustment is still based on the material before speed adjustment, it may further increase the transverse thickness deviation. Therefore, it is necessary to combine the transverse and longitudinal thickness deviations to collaboratively adjust both the clamping force and the stretching equipment speed. The clamping force correction range and rotation speed correction range can be predicted based on the gradient boosting regression tree, and the clamping force of the clamping device and the rotation speed of the stretching device can be corrected according to the clamping force correction range and rotation speed correction range, respectively.

[0083] Gradient boosting regression trees are a regression model based on ensemble learning. Their core principle involves progressively building multiple decision trees (base learners), with each new tree focusing on fitting the residuals (errors) between the predictions of all preceding trees and the actual values. Gradient descent is used to continuously optimize these residuals, and the final output is the sum of all the predictions. When handling regression problems, gradient boosting regression trees retain the ability of decision trees to capture nonlinear relationships while reducing the risk of overfitting from individual trees through ensemble strategies, thus achieving high prediction accuracy. In the step-by-step stretching process of lithium-ion battery separators, gradient boosting regression trees efficiently handle nonlinear relationships. For parameters such as clamping force of the clamping equipment, rotation speed of the stretching equipment, and deviations in lateral and longitudinal thickness data, gradient boosting regression trees do not require manually pre-defined function forms; they automatically capture these nonlinear patterns through piecewise fitting of decision trees.

[0084] In some embodiments of this application, correcting the clamping force of the clamping device based on the lateral thickness data deviation includes:

[0085] Based on the lateral thickness data deviation, locate all areas of abnormal thickness in the raw material;

[0086] Calculate the distance between abnormal regions among all regions with varying thicknesses;

[0087] The initial value of the clamping force was determined based on the deviation value of the transverse thickness data and by using regression analysis.

[0088] The initial value of the clamping force is corrected based on the distance to the abnormal area to obtain the target value of the clamping force;

[0089] The clamping force of the clamping device is adjusted according to the target clamping force value.

[0090] In this embodiment, the lateral thickness data deviation value refers to the difference in lateral thickness data between two adjacent regions of the raw material. When the lateral thickness data deviation value is greater than a preset first deviation threshold, it indicates that one of the two adjacent regions of the raw material is too thick and the other is too thin, and both regions are considered to have abnormal thickness. Based on the lateral thickness data deviation value, the excessively thick or thin regions of the raw material are marked as regions with abnormal thickness. For regions with abnormal thickness, the clamping force of the corresponding clamping device may be too large or too small, therefore, it is necessary to correct the clamping force of the clamping device in these regions.

[0091] The lateral thickness deviation values ​​of regions with abnormal thickness can be directly input into a preset neural network model to obtain the target clamping force value. This preset neural network model is trained using a training dataset and an initial neural network model. The training dataset includes historical lateral thickness deviation values ​​and corresponding historical clamping forces of the clamping device. The clamping force of the clamping device is corrected based on the target clamping force value. The clamping force is predicted using the preset neural network model, which is trained using an initial neural network model. The training dataset includes historical lateral thickness deviation values ​​and corresponding historical clamping forces of the clamping device. The initial neural network model can be a multilayer perceptron or a convolutional neural network, and the network parameters are continuously adjusted using a backpropagation algorithm. The historical lateral thickness deviation values ​​from the training dataset are input into the model to calculate the error between the predicted clamping force and the actual historical clamping force. The error is minimized using a gradient descent optimization algorithm, and training is iterated repeatedly until the model can accurately learn the complex mapping relationship between thickness deviation and clamping force. Training is complete when the model achieves a low prediction error on the validation set. In practical applications, the real-time deviation value of the transverse thickness data is input into the trained model, and the model can quickly output the corresponding clamping force target value, providing an accurate basis for adjusting the clamping force of the diaphragm production equipment, thereby ensuring the uniformity of the transverse thickness of the diaphragm and the stability of the production process.

[0092] While neural network models can accurately output the target clamping force value, training such models is complex, requires significant computational power, and necessitates a large amount of historical lateral thickness deviation data and corresponding historical clamping forces, making it highly dependent on the quantity and quality of the data. Therefore, regression analysis can be used to calculate the target clamping force value. However, considering that the accuracy of regression analysis is lower than that of neural network models, anomaly distance is introduced to correct the initial clamping force value obtained through regression analysis, resulting in a more accurate target clamping force value. This method balances computational power while ensuring the accuracy of the final result.

[0093] Specifically, two adjacent areas with a lateral thickness deviation exceeding a preset first deviation threshold are designated as thickness aberration areas. After identifying these areas, the raw material edge is used as a reference, and the vertical distance between the center point of each thickness aberration area and the raw material edge is recorded. Based on this vertical distance, the center point distance of each thickness aberration area, i.e., the aberration area distance, is further calculated. If two aberration areas are close together, for example, if a thickness aberration area is flanked by two other thickness aberration areas, the force transmission will overlap or interfere with each other when the clamping force is adjacent to the two aberration areas due to the certain diffusion range of the clamping force. Therefore, the adjustment range of the clamping force of the clamping device at the corresponding positions of the three thickness aberration areas needs to be appropriately reduced to minimize interference. Historical lateral thickness deviation values ​​and corresponding historical clamping forces of the clamping device are obtained as sample data. The historical lateral thickness deviation values ​​are used as the dependent variable, and the historical clamping forces are used as the independent variable. Nonlinear regression analysis is performed because the relationship between the lateral thickness deviation values ​​and the clamping forces is not linear. When the clamping force is too low, the lateral thickness deviation values ​​decrease rapidly as the clamping force increases. Once the clamping force reaches a reasonable range, the change in lateral thickness deviation values ​​slows down. A regression equation between the lateral thickness deviation values ​​and the clamping forces is obtained through quadratic regression fitting. The lateral thickness deviation values ​​between the current abnormal thickness region and adjacent abnormal thickness regions are input into the regression equation to obtain the initial value of the clamping force of the clamping device. If the current area of ​​abnormal thickness also has adjacent areas of abnormal thickness, for example, if the thickness distribution of multiple consecutive regions in a raw material is as follows: region A is thinner, region B is thicker, region C is thicker, and region D is thinner, then region C is adjacent to two areas of abnormal thickness, namely regions B and D. Therefore, the initial clamping force value of region C needs to be multiplied by a preset correction coefficient to obtain the target clamping force value, in order to compensate for the interference of regions B and D on region C. The correction coefficient can be determined through multiple clamping force adjustment tests of the clamping equipment. Finally, the clamping force of the clamping equipment corresponding to the area of ​​abnormal thickness is corrected to the target clamping force value.

[0094] Through the above steps, precise control of the raw material thickness can be achieved during the transverse stretching process. At the same time, the raw material is divided into regions, and deviation analysis is performed on adjacent regions. This allows for the rapid location of the clamping equipment that needs correction, making the correction process more targeted and achieving precise thickness control, thus providing quality assurance for the lithium battery separators produced subsequently.

[0095] In some embodiments of this application, the coordinated correction of the clamping force of the clamping device and the rotational speed of the stretching device by combining the lateral thickness data deviation value and the longitudinal thickness data deviation value includes:

[0096] The lateral thickness data deviation value and the longitudinal thickness data deviation value are input into the pre-built deviation coupling correction model. The deviation coupling correction model outputs the clamping force correction amplitude and the rotation speed correction amplitude. The deviation coupling correction model is constructed based on the gradient boosting regression tree.

[0097] The clamping force of the clamping device and the rotation speed of the stretching device are corrected according to the clamping force correction range and the rotation speed correction range, respectively.

[0098] In this embodiment, gradient boosting regression trees can handle nonlinear relationships, capture feature interactions (such as the synergistic effect of lateral and longitudinal thickness data deviations), and are highly robust to noise in process data. They are suitable for learning complex mapping relationships from historical data and outputting continuous correction amplitudes, namely, clamping force correction amplitudes and rotational speed correction amplitudes. First, sufficient historical lateral and longitudinal thickness data deviations, along with corresponding historical clamping forces and rotational speeds of clamping and stretching equipment, are collected and integrated into a sample set. After data cleaning and normalization, the sample set is divided into a training set (for model learning) and a test set (for verifying generalization ability) in a certain ratio, such as 7:3, ensuring that the deviation distribution and process condition distribution of the two sets of samples are consistent (e.g., both include high / medium / low deviation scenarios). Then, a deviation coupling correction model is constructed based on the gradient boosting regression tree. The deviation coupling correction model is a multi-output gradient boosting regression tree model. The training set is input into the bias coupling correction model for iterative training. Multiple decision trees are generated iteratively through gradient boosting, with each tree fitting the prediction residuals of the preceding model to gradually reduce the error. During training, model hyperparameters, such as the number of trees, tree depth, and learning rate, are continuously optimized. The model error is validated using a test set. Training is complete when the model error is less than a preset error threshold or the preset maximum number of iterations is reached, resulting in the trained bias coupling correction model. The current lateral and longitudinal thickness data deviations are input into the trained bias coupling correction model, which outputs the clamping force correction amplitude and rotational speed correction amplitude. The clamping force of the clamping device and the rotational speed of the stretching device are then adjusted based on these correction amplitudes.

[0099] Gradient boosting regression trees are a regression model based on ensemble learning. Their core principle involves progressively building multiple decision trees (base learners), with each new tree focusing on fitting the residuals (errors) between the predictions of all preceding trees and the actual values. Gradient descent is used to continuously optimize these residuals, and the final output is the sum of all the predictions. When handling regression problems, gradient boosting regression trees retain the ability of decision trees to capture nonlinear relationships while reducing the risk of overfitting from individual trees through ensemble strategies, thus achieving high prediction accuracy. In the step-by-step stretching process of lithium-ion battery separators, gradient boosting regression trees efficiently handle nonlinear relationships. For parameters such as clamping force of the clamping equipment, rotation speed of the stretching equipment, and deviations in lateral and longitudinal thickness data, gradient boosting regression trees do not require manually pre-defined function forms; they automatically capture these nonlinear patterns through piecewise fitting of decision trees.

[0100] Reference Figure 3 As shown, in some embodiments of this application, comparing each set of thickness datasets with other thickness datasets at adjacent time points, and determining the target thickness dataset from the thickness datasets based on data differences includes:

[0101] Step 301: Determine the distance value in the data space between the thickness data in each thickness dataset and the thickness data in other thickness datasets at adjacent time points, where the distance value includes Euclidean distance or Manhattan distance;

[0102] Step 302: Determine the target thickness dataset based on the distance values.

[0103] In this embodiment, by comparing the corresponding data in the thickness dataset and determining the distance between the two datasets in the data space, the data difference between the two datasets can be determined.

[0104] For example, for thickness datasets at two adjacent time points, the two sets of data are A =[ a 1, a 2,..., a n ]and B =[ b 1, b 2,..., b n The formula for calculating the Euclidean distance in the data space is:

[0105] ;

[0106] It is the square root of the sum of the squares of the differences between corresponding elements; the smaller the distance value, the smaller the data difference.

[0107] For both datasets, the distance value can also be determined by the Manhattan distance. The formula for calculating the Manhattan distance in the data space is:

[0108] ;

[0109] It is the sum of the absolute values ​​of the differences between corresponding elements; the smaller the distance value, the smaller the data difference.

[0110] In some embodiments, determining the target thickness dataset based on the distance value can be achieved by determining the data difference between the thickness dataset at each time point and the thickness dataset at the previous time point. When the determined distance value is less than a preset threshold, the thickness dataset at the current time point is determined as the target thickness dataset.

[0111] It is understood that, for the stretched raw material in the embodiments of this application, which is a plane, it can be a data array of thickness data in each row and column of the corresponding raw material plane when the dataset is divided.

[0112] In some embodiments of this application, the thickness data includes the average thickness value and the dispersion value of each region, and the distance value includes a first distance value calculated from the average value of each region in each thickness dataset and other thickness datasets at adjacent time points, and a second distance value calculated from the dispersion value.

[0113] The dataset for determining target thickness based on distance values ​​includes:

[0114] The comprehensive distance index is determined based on the first and second distance values. The formula for determining the comprehensive distance index is as follows: D = w 1 D 1+ w 2 D 2 D1 is the first distance value, and D2 is the second distance value. w 1 represents the first preset weight. w 2 represents the second preset weight. w 2 greater than w 1;

[0115] The target thickness dataset is determined based on the comprehensive distance index.

[0116] In this embodiment, the thickness data for each region in the thickness dataset determined at each time point has two types. One type represents the average thickness of the region. During the raw material stretching process, if the equipment as a whole does not experience major malfunctions, the thickness at adjacent time points usually does not fluctuate significantly. If significant fluctuations occur, the average value can still relatively accurately indicate the overall thickness of the region. Therefore, after collecting a large number of thickness values, this embodiment uses the mean value calculation. Subsequently, only the dispersion value in each set of data needs to be calculated, without having to calculate the distance between each thickness value in adjacent sets of data one by one, thereby reducing the amount of data. This is more conducive to quickly comparing the differences between data at adjacent time points. In addition, this application further compares the dispersion value of the thickness values ​​in the region. The dispersion value can be the standard deviation or variance, etc. The dispersion value can reflect the uniformity of the thickness in the region at the same time. The smaller the value, the closer the thickness at each point is to the average value, and the better the thickness uniformity. Conversely, the larger the value, the more drastic the thickness fluctuation and the worse the uniformity.

[0117] At two adjacent time points, both the average value and the dispersion value can reflect the differences between the two sets of data. For example, if the average values ​​are similar but the dispersion values ​​differ significantly, it indicates that there are areas with large thickness fluctuations during a single stretching process, and this fluctuation is continuous. While using only the average value saves computation, it may lead to misjudgments. Therefore, combining the dispersion value makes the judgment more accurate. Conversely, if the average values ​​differ but the dispersion values ​​are not significantly different, it indicates that the equipment's operating state may be a stable process during continuous stretching. The thickness of each region of the raw material stretched at adjacent time points varies to some extent, causing changes in the average value. However, the thickness fluctuation within the same region at the same time point is relatively uniform, resulting in similar dispersion values ​​at adjacent time points. In this case, the conclusion regarding the difference between the two sets of data is mainly based on the change in the average value. It can be understood that when both the average value and the dispersion value are similar, it indicates that the difference between the two sets of data is relatively small.

[0118] In this embodiment, the spatial distance between the average values ​​of the same region in two adjacent sets of thickness datasets is the first distance value, and the spatial distance between the dispersion values ​​is the second distance value. The first distance value and the second distance value are combined to construct a distance comprehensive index formula for judgment. Specifically, when comparing and judging, the second distance value corresponding to the dispersion value has a greater judgment weight to comprehensively reflect the changes in the large amount of data collected. Finally, the target thickness dataset is determined by the distance comprehensive index. For example, when the above-determined distance comprehensive index is less than the set preset index threshold, the thickness dataset at the current time node is determined as the target thickness dataset.

[0119] In some embodiments of this application, the distance composite index includes a first distance composite index of the thickness dataset at the current time node relative to the thickness dataset at the previous time node, and a second distance composite index of the thickness dataset at the current time node relative to the thickness dataset at the next time node.

[0120] When both the first and second comprehensive distance indices are less than or equal to the preset index thresholds, the thickness dataset at the current time node is determined as the target thickness dataset.

[0121] In this embodiment, the difference between the thickness dataset at the current time node and the thickness datasets at the two time nodes before and after is combined to more accurately determine whether the device operation has reached a stable state. That is, the first comprehensive distance index and the second comprehensive distance index between the current time node and the two time nodes before and after are calculated by the above-mentioned Euclidean distance or Manhattan distance. Thus, only when the first comprehensive distance index and the second comprehensive distance index are both less than or equal to the preset index threshold, it indicates that the difference between the data at the current time node and the data before and after is not large, that is, the data fluctuation is not large, thus more accurately indicating that the current device operation has reached a stable state.

[0122] The preset threshold values ​​can be set according to the actual situation, and will not be elaborated on here.

[0123] In some embodiments of this application, the thickness datasets of multiple time nodes and the first and second distance comprehensive indices of adjacent time nodes are determined within a preset time period, and the average value of the first and second distance comprehensive indices is determined.

[0124] When the first and second distance comprehensive indicators of multiple time nodes are both less than or equal to the preset indicator threshold, the thickness dataset of the time node with the smallest average indicator value is determined as the target thickness dataset. The target thickness dataset determined in this way has the smallest data difference and the smallest data fluctuation relative to the data before and after, thus more accurately indicating that the current equipment operation has reached a stable state.

[0125] In some embodiments of this application, the thickness value is obtained by scanning the raw material with a thickness measuring device; the construction of the thickness dataset corresponding to each time point includes the following steps:

[0126] The area scanned by the thickness measuring device is divided into multiple regions;

[0127] Determine the average value of the thickness values ​​in each of the said regions;

[0128] The thickness data is generated based on the average value.

[0129] In this embodiment, multiple thickness values ​​are obtained by scanning the stretched raw material using a thickness measuring device. The area of ​​the raw material scanned by the thickness measuring device is divided into multiple regions. It can be understood that there are multiple thickness value data in each region. For a large number of thickness value data, the average thickness value of each region is calculated. This average value can characterize the overall thickness of the region to a certain extent. Subsequently, thickness data in the thickness dataset can be constructed based on this average value. For example, in one embodiment, the thickness data is the average thickness value of the region; in another embodiment, the thickness data is the average thickness value and the dispersion value of the region.

[0130] In some embodiments of this application, correcting the process parameters of the lithium battery separator based on the target thickness dataset includes:

[0131] Based on the average values ​​of each region in the target thickness dataset, determine the thickness variation of the raw material along the transverse and longitudinal directions;

[0132] When the thickness variation characterizes the uniform thickness variation of the raw material in the same direction, compare whether the variation of the dispersion value of each region in the same direction matches the thickness variation.

[0133] If so, the equipment requiring process parameter correction is located based on the thickness variation and dispersion value, and the process parameters are corrected accordingly.

[0134] If not, repeat the steps to determine the target thickness dataset.

[0135] In this embodiment, the lithium battery separator process parameters are corrected by combining the target thickness dataset, which includes average and dispersion values.

[0136] The thickness variation includes the first thickness variation and the second thickness variation mentioned above.

[0137] In lithium-ion battery separator fabrication, the raw material is typically divided into a central region, a transition region, and an edge region arranged sequentially along both the longitudinal and transverse directions. The average value of each of these regions is calculated using a target thickness dataset. This allows for the determination of the thickness variation of the raw material in these regions along both the longitudinal and transverse directions. When the thickness variation indicates a uniform change in the thickness of the raw material in the same direction—for example, during stretching—it may gradually thin or thicken along the stretching direction, or the thickness may be uniform at the center but gradually thin or thicken towards the edges. In this case, the corresponding dispersion value should have relatively small variations in adjacent regions or along the same direction, thus matching the thickness variation. If a mismatch exists, it indicates a potential problem with the currently determined target thickness dataset or a equipment malfunction causing inconsistent thickness fluctuations in each region. Therefore, the target thickness dataset determination steps can be repeated, for example, including steps 100-300, to improve the accuracy of the correction.

[0138] If the changes match, the equipment that needs to be corrected for process parameters can be located based on the thickness changes and the dispersion value. For example, if the thickness changes reflect thickness fluctuations in adjacent areas or in the same horizontal or vertical direction, the stretching equipment that handles that direction can be directly located to correct the process parameters. If the dispersion value is too high, there is a problem in the area corresponding to the dispersion value, and the stretching equipment that handles that area can be located to correct the process parameters.

[0139] For example, when determining the second thickness variation of the raw material in the longitudinal direction based on the target thickness dataset, and correcting the rotation speed of the stretching equipment based on the second thickness variation, on the one hand, the second thickness variation of the overall raw material can be indicated by the average value of two adjacent regions or in the same direction in the longitudinal direction; on the other hand, the thickness variation in each region can be indicated by the dispersion value. In some specific application scenarios, the relatively large range of thickness fluctuations corresponding to multiple adjacent regions in the longitudinal direction may meet the production requirements, but due to equipment process parameter issues, such as rotation speed fluctuations occurring only at certain times, thickness variations may occur. Therefore, combining the dispersion value further reflects the small range of changes in a single region, thereby making the detection of the raw material after stretching more accurate, and thus facilitating better adjustment or correction of process parameters.

[0140] In some embodiments of this application, dividing the area scanned by the thickness measuring device of the raw material into multiple regions includes:

[0141] Obtain the equipment size parameters of the stretching equipment used to stretch the raw material, the equipment size parameters including the size of the stretching rollers and the spacing between adjacent stretching rollers;

[0142] The region size parameters are determined based on the equipment size parameters, so as to divide the area of ​​the raw material scanned by the thickness measuring equipment into multiple regions according to the region size parameters.

[0143] In this embodiment, the above-mentioned regions are divided using the equipment size parameters of the stretching equipment. The stretching of the raw material is usually achieved by multiple stretching rollers arranged sequentially along the longitudinal direction. The stretching is completed after the raw material passes through all the stretching rollers. Therefore, the thickness of each region in the stretched raw material has a certain correspondence with the arrangement and size of the stretching rollers. Thus, the equipment size includes the size of the stretching rollers and the spacing between adjacent stretching rollers in the longitudinal direction. The size of the stretching rollers can specifically include the width of the stretching rollers along the longitudinal direction. Based on these equipment parameters, the region size parameters are determined. Specifically, the region size parameters can be the length of the region boundary along the longitudinal direction. Finally, the divided regions can correspond to the position of the stretching equipment, which can more accurately determine the position that needs adjustment and correction when analyzing the thickness dataset in the subsequent process.

[0144] In one specific embodiment, the above-described scheme in the embodiments of this application is performed after the raw material is longitudinally stretched to facilitate subsequent transverse stretching preparation. In other embodiments, the above-described scheme of this application can also be implemented after transverse stretching.

[0145] This application also provides a self-calibration system for lithium battery separator process parameters based on big data analysis. The self-calibration system for lithium battery separator process parameters includes:

[0146] One or more processors;

[0147] Memory;

[0148] and one or more applications, wherein the one or more applications are stored in memory and configured to be operated by the processor in any of the methods in any of the above method embodiments.

[0149] like Figure 4 As shown, it illustrates a structural schematic diagram of the electronic device involved in the embodiments of this application, specifically:

[0150] The electronic device may include components such as a processor 501 with one or more processing cores, a storage unit 502 with one or more computer-readable storage media, a power supply 503, and an input unit 504. Those skilled in the art will understand that... Figure 4 The electronic device structure shown does not constitute a limitation on the electronic device and may include more or fewer components than shown, or combine certain components, or have different component arrangements. Wherein:

[0151] The processor 501 is the control center of the electronic device. It connects various parts of the electronic device via various interfaces and lines. By running or executing software programs and / or modules stored in the storage unit 502, and by calling data stored in the storage unit 502, it performs various functions and processes data, thereby providing overall monitoring of the electronic device. Optionally, the processor 501 may include one or more processing cores; preferably, the processor 501 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the modem processor may also not be integrated into the processor 501.

[0152] Storage unit 502 can be used to store software programs and modules. Processor 501 executes various functional applications and data processing by running the software programs and modules stored in storage unit 502. Storage unit 502 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the electronic device, etc. In addition, storage unit 502 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, storage unit 502 may also include a memory controller to provide processor 501 with access to storage unit 502.

[0153] The electronic device also includes a power supply 503 that supplies power to various components. Preferably, the power supply 503 can be logically connected to the processor 501 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system. The power supply 503 may also include one or more DC or AC power supplies, recharging systems, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components.

[0154] The electronic device may also include an input unit 504, which can be used to receive input digital or character information and generate keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function control.

[0155] Although not shown, the electronic device may also include a display unit, etc., which will not be described in detail here. Specifically, in the embodiments of this application, the processor 501 in the electronic device loads the executable files corresponding to the processes of one or more applications into the storage unit 502 according to the following instructions, and the processor 501 runs the applications stored in the storage unit 502 to realize various functions, as follows:

[0156] In the process of preparing lithium battery separators, the thickness values ​​of the raw materials are obtained at multiple consecutive time points after stretching.

[0157] Construct a thickness dataset corresponding to each time node, wherein the thickness dataset includes thickness data for characterizing the thickness of each region of the raw material, and the thickness data is generated based on the thickness value;

[0158] Compare each set of thickness datasets with other thickness datasets at adjacent time points, and determine the target thickness dataset from the thickness datasets based on the data differences;

[0159] The process parameters of the lithium battery separator are corrected based on the target thickness dataset.

[0160] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be performed by a program or instructions, or by controlling related hardware through a program or instructions. The program or instructions can be stored in a computer-readable storage medium and loaded and executed by a processor.

[0161] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the detailed descriptions of other embodiments above, which will not be repeated here.

[0162] In practice, each of the above units or structures can be implemented as an independent entity or can be arbitrarily combined to be implemented as the same or several entities. For the specific implementation of each of the above units or structures, please refer to the previous method embodiments, which will not be repeated here.

[0163] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0164] The embodiments of this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A self-calibration method for lithium battery separator process parameters based on big data analysis, characterized in that, Includes the following steps: In the process of preparing lithium battery separators, the thickness values ​​of the raw materials are obtained at multiple consecutive time points after stretching. Construct a thickness dataset corresponding to each time point. The thickness dataset includes thickness data to characterize the thickness of each region of the raw material. The thickness data is generated based on the thickness values. Determine the distance between the thickness data in each set of thickness datasets and the thickness data in other thickness datasets at adjacent time points in the data space. The distance value includes Euclidean distance or Manhattan distance. The thickness data includes the average thickness value and the dispersion value of each region. The distance value also includes a first distance value calculated from the average value of each region in each set of thickness datasets and other thickness datasets at adjacent time points, and a second distance value calculated from the dispersion value. The comprehensive distance index is determined based on the first and second distance values. The formula for determining the comprehensive distance index is as follows: D = w 1 D 1+ w 2 D 2 D1 is the first distance value, and D2 is the second distance value. w 1 represents the first preset weight. w 2 represents the second preset weight. w 2 greater than w 1; The target thickness dataset is determined based on the comprehensive distance index, which includes the first comprehensive distance index of the thickness dataset at the current time point relative to the thickness dataset at the previous time point, and the second comprehensive distance index of the thickness dataset at the current time point relative to the thickness dataset at the next time point. When both the first and second comprehensive distance indices are less than or equal to preset threshold values, the thickness dataset at the current time point is determined as the target thickness dataset. Based on the average values ​​of each region in the target thickness dataset, determine the thickness variation of the raw material along the transverse and longitudinal directions; When the thickness variation characterizes the uniform thickness variation of the raw material in the same direction, compare whether the variation of the dispersion value of each region in the same direction matches the thickness variation. If not, repeat the steps to determine the target thickness dataset; If so, the equipment requiring process parameter correction will be located based on the thickness variation and dispersion values; Based on the target thickness dataset, determine the first thickness variation of the raw material from the center to the edge in the transverse direction and the second thickness variation of the raw material in the longitudinal direction; The clamping force of the clamping device and the rotation speed of the stretching device are adjusted based on the first and second thickness changes.

2. The self-calibration method for lithium battery separator process parameters based on big data analysis according to claim 1, characterized in that, The first thickness variation includes the lateral thickness deviation between two adjacent areas of the raw material in the transverse direction, and the second thickness variation includes the longitudinal thickness deviation between two adjacent areas of the raw material in the longitudinal direction. The step of combining the first and second thickness variations to correct the clamping force of the clamping device and the rotational speed of the stretching device includes: If the lateral thickness data deviation is greater than the preset first deviation threshold and the longitudinal thickness data deviation is less than or equal to the preset second deviation threshold, the clamping force of the clamping device is adjusted according to the lateral thickness data deviation. If the longitudinal thickness data deviation is greater than the second deviation threshold and the transverse thickness data deviation is less than or equal to the first deviation threshold, then the speed of the stretching equipment is adjusted according to the longitudinal thickness data deviation. If the lateral thickness data deviation is greater than the first deviation threshold and the longitudinal thickness data deviation is greater than the second deviation threshold, then the clamping force of the clamping device and the rotation speed of the stretching device are adjusted in conjunction with the lateral thickness data deviation and the longitudinal thickness data deviation.

3. The self-calibration method for lithium battery separator process parameters based on big data analysis according to claim 2, characterized in that, The step of correcting the clamping force of the clamping device based on the lateral thickness data deviation includes: Based on the lateral thickness data deviation, locate all areas of abnormal thickness in the raw material; Calculate the distance between abnormal regions among all regions with varying thicknesses; The initial value of the clamping force was determined based on the deviation value of the transverse thickness data and by using regression analysis. The initial value of the clamping force is corrected based on the distance to the abnormal area to obtain the target value of the clamping force; The clamping force of the clamping device is adjusted according to the target clamping force value.

4. The self-calibration method for lithium battery separator process parameters based on big data analysis according to claim 2, characterized in that, The method of combining lateral and longitudinal thickness deviation data to collaboratively correct the clamping force of the clamping device and the rotation speed of the tensioning device includes: The lateral thickness data deviation value and the longitudinal thickness data deviation value are input into the pre-built deviation coupling correction model. The deviation coupling correction model outputs the clamping force correction amplitude and the rotation speed correction amplitude. The deviation coupling correction model is constructed based on the gradient boosting regression tree. The clamping force of the clamping device and the rotation speed of the stretching device are corrected according to the clamping force correction range and the rotation speed correction range, respectively.

5. A self-calibration system for lithium battery separator process parameters based on big data analysis, characterized in that, The lithium battery separator process parameter self-calibration system based on big data analysis includes: One or more processors; Memory; And one or more applications, wherein the one or more applications are stored in memory and configured to be executed by a processor to implement the method of any one of claims 1 to 4.

Images