Method for determining a peeling force prediction model and related device

Abstract

The application provides a determination method of a peeling force prediction model and related equipment. The method comprises the following steps: obtaining historical process data of a plurality of batches of electrode sheets in a production process and corresponding real values of peeling force; performing feature extraction on the historical process data to obtain a first feature set; obtaining physical mechanism features corresponding to the first feature set to obtain a second feature set; performing fusion processing on the first feature set and the second feature set to obtain a reference feature vector; training a preset initial model according to the reference feature vector and the real values of the peeling force to obtain a target prediction model; and the target prediction model is used for real-time prediction of the peeling force after electrode sheet rolling. Real-time and accurate prediction of the peeling force in the electrode sheet rolling process can be realized, thereby improving the production quality and safety level of the battery electrode sheet.

Description

Technical Field

[0001] This application relates to the field of battery electrode technology, and in particular to a method for determining a peel force prediction model and related equipment. Background Technology

[0002] With the rapid development of the new energy vehicle industry, the requirements for energy density, cycle life, and safety of lithium-ion batteries are constantly increasing. As a core component of lithium-ion batteries, the adhesion strength between the coating and the current collector of the electrode is usually characterized by peel force, which directly determines the charging and discharging stability and safety of the battery. If the peel force is not up to standard, it can easily lead to problems such as coating peeling and active material powder shedding. In severe cases, it can cause internal short circuits in the battery, leading to major safety accidents such as thermal runaway.

[0003] Currently, industry monitoring of peel force mainly relies on offline sampling and destructive testing. This involves cutting electrode samples after the rolling process and performing 180° or 90° peel tests using a tensile testing machine. This method has a significant time lag; when an abnormal peel force is detected, the corresponding batch of electrodes may have already been produced or even entered the next process. It cannot achieve real-time feedback and control of the production process. At the same time, offline sampling has limited coverage, which can easily lead to missed detection of non-conforming products, increasing production costs and quality risks.

[0004] Therefore, how to achieve real-time and accurate prediction of the peeling force in the electrode rolling process, so as to improve the production quality and safety level of battery electrodes, is an urgent problem to be solved. Summary of the Invention

[0005] This application provides a method and related equipment for determining a peel force prediction model. By integrating the basic process characteristics and physical mechanism characteristics of electrode production, the model prediction accuracy and generalization ability are effectively improved, enabling real-time and accurate prediction of peel force in the electrode rolling process. This avoids the drawbacks of traditional offline detection, such as strong lag and low sampling coverage, thereby improving the production quality and safety level of battery electrodes.

[0006] In a first aspect, embodiments of this application provide a method for determining a peel force prediction model, the method comprising: Obtain historical process data of multiple batches of electrode sheets during the production process and their corresponding actual peel force values; Feature extraction is performed on the historical process data to obtain a first feature set; Obtain the physical mechanism features corresponding to the first feature set to obtain the second feature set; The first feature set and the second feature set are fused to obtain a reference feature vector. Based on the reference feature vector and the true value of the peel force, a preset initial model is trained to obtain a target prediction model; the target prediction model is used for real-time prediction of the peel force after the electrode is rolled.

[0007] Secondly, embodiments of this application provide an intelligent peeling force prediction system, wherein the system is deployed with a target prediction model determined by the method described in the first aspect, and the system includes a data acquisition module, a feature extraction module, a fusion processing module, and a real-time prediction module, wherein: The data acquisition module is used to acquire real-time process datasets; The feature extraction module is used to extract features from the real-time process dataset to obtain a third feature set; and to obtain the physical mechanism features corresponding to the third feature set to obtain a fourth feature set. The fusion processing module is used to fuse the third feature set and the fourth feature set to obtain the target feature vector; The real-time prediction module is used to input the target feature vector into the target prediction model and output the target peeling force.

[0008] Thirdly, embodiments of this application provide an electronic device, including a processor, a memory, a communication interface, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor, and the programs include instructions for performing steps in any method of the first aspect of this application.

[0009] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program for electronic data interchange, wherein the computer program causes a computer to perform some or all of the steps described in any method of the first aspect of this application.

[0010] Fifthly, embodiments of this application provide a computer program product, wherein the computer program product includes a non-transitory computer-readable storage medium storing a computer program operable to cause a computer to perform some or all of the steps described in any method of the first aspect of this application. The computer program product may be a software installation package.

[0011] It is evident that by integrating the basic process characteristics and physical mechanism characteristics of electrode production, the model's prediction accuracy and generalization ability can be effectively improved, enabling real-time and accurate prediction of the peeling force in the electrode rolling process. This avoids the drawbacks of traditional offline testing, such as strong lag and low sampling coverage, thereby improving the production quality and safety level of battery electrodes. Attached Figure Description

[0012] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a system architecture diagram of an intelligent peeling force prediction system provided in an embodiment of this application; Figure 2 This is a front-end data source architecture diagram of a data acquisition module provided in an embodiment of this application; Figure 3 This is an application scenario diagram of an intelligent peeling force prediction system provided in an embodiment of this application; Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application; Figure 5 This is a flowchart illustrating a method for determining a peel force prediction model provided in an embodiment of this application; Figure 6 This is a schematic flowchart illustrating a method for determining a first physical mechanism feature, provided in an embodiment of this application. Figure 7 This is a schematic flowchart illustrating the process of determining a second physical mechanism feature, provided in an embodiment of this application. Figure 8 This is a block diagram of the functional units of a device for determining a peel force prediction model provided in an embodiment of this application. Detailed Implementation

[0014] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0015] The terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.

[0016] It should be understood that the term "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this document indicates that the preceding and following related objects are in an "or" relationship. In the embodiments of this application, "multiple" refers to two or more.

[0017] In the embodiments of this application, "at least one item" or its similar expression refers to any combination of these items, including any combination of a single item or a plurality of items. "One or more" means one or more, while "multiple" means two or more. For example, "at least one item" of a, b, or c can represent the following seven cases: a, b, c; a and b; a and c; b and c; a, b, and c. Each of a, b, and c can be an element or a set containing one or more elements.

[0018] In this application embodiment, "connection" refers to various connection methods such as direct connection or indirect connection to realize communication between devices. This application embodiment does not limit this in any way.

[0019] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0020] With the rapid development of the new energy vehicle industry, the requirements for energy density, cycle life, and safety of lithium-ion batteries are constantly increasing. As a core component of lithium-ion batteries, the adhesion strength between the coating and the current collector of the electrode is usually characterized by peel force, which directly determines the charging and discharging stability and safety of the battery. If the peel force is not up to standard, it can easily lead to problems such as coating peeling and active material powder shedding. In severe cases, it can cause internal short circuits in the battery, leading to major safety accidents such as thermal runaway.

[0021] Currently, industry monitoring of peel force mainly relies on offline sampling and destructive testing. This involves cutting electrode samples after the rolling process and performing 180° or 90° peel tests using a tensile testing machine. This method has a significant time lag; when an abnormal peel force is detected, the corresponding batch of electrodes may have already been produced or even entered the next process. It cannot achieve real-time feedback and control of the production process. At the same time, offline sampling has limited coverage, which can easily lead to missed detection of non-conforming products, increasing production costs and quality risks.

[0022] Therefore, how to achieve real-time and accurate prediction of the peeling force in the electrode rolling process, so as to improve the production quality and safety level of battery electrodes, is an urgent problem to be solved.

[0023] To address the aforementioned issues, this application provides a method and related equipment for determining a peel force prediction model. The method involves acquiring historical process data and corresponding actual peel force values ​​for multiple batches of electrode sheets during the production process; extracting features from the historical process data to obtain a first feature set; acquiring the physical mechanism features corresponding to the first feature set to obtain a second feature set; fusing the first and second feature sets to obtain a reference feature vector; and training a preset initial model based on the reference feature vector and the actual peel force values ​​to obtain a target prediction model. The target prediction model is used for real-time prediction of the peel force after electrode sheet rolling.

[0024] It is evident that by integrating the basic process characteristics and physical mechanism characteristics of electrode production, the model's prediction accuracy and generalization ability can be effectively improved, enabling real-time and accurate prediction of the peeling force in the electrode rolling process. This avoids the drawbacks of traditional offline testing, such as strong lag and low sampling coverage, thereby improving the production quality and safety level of battery electrodes.

[0025] For easier understanding, please refer to Figure 1 , Figure 1 This is a system architecture diagram of an intelligent peel force prediction system provided in an embodiment of this application. The intelligent peel force prediction system includes a data acquisition module, a feature extraction module, a fusion processing module, and a real-time prediction module. Each module is connected in sequence, and data flows unidirectionally and works together to complete the online intelligent prediction of electrode peel force. The specific functions of each module are as follows: The data acquisition module is used to establish communication connections with the coating equipment, surface density detection equipment, rolling equipment and production line control system of the electrode production line. It collects real-time process datasets of online operation, including process parameters of the coating process, coating surface density detection data and process parameters of the rolling process. It realizes synchronous acquisition, time alignment and stable transmission of multi-source, multi-process and multi-dimensional data, providing a raw and accurate data foundation for subsequent feature processing and model prediction.

[0026] For easier understanding, please refer to Figure 2 , Figure 2This is a front-end data source architecture diagram of a data acquisition module provided in an embodiment of this application. The components include: a coating machine (providing coating process data such as belt speed, oven temperature, coating gap, etc.); a surface density meter (providing surface density data (online detection values ​​of the dry film surface density of the electrode coating); and a rolling mill (providing rolling process data such as roll gap width, rolling pressure, rolling force, etc.). These three types of data (coating process data, surface density data, and rolling process data) are simultaneously imported into the data acquisition module as the system's raw input data, providing a foundation for subsequent feature extraction and model prediction.

[0027] The feature extraction module is used to perform basic feature extraction and physical mechanism feature calculation. On the one hand, it cleans, normalizes, and extracts key data from the real-time process dataset input by the data acquisition module to obtain the third feature set. On the other hand, based on the coating process data, areal density data, and rolling process data in the third feature set, it calculates the physical features according to the preset physical mechanism formula to obtain the fourth feature set, so that the system simultaneously possesses process data features and physical prior knowledge features.

[0028] The fusion processing module integrates the third and fourth feature sets output by the feature extraction module. Through feature dimension concatenation, redundant feature removal, feature dimension optimization, and vector standardization, the two types of features are integrated into a target feature vector with unified structure, consistent dimensions, and optimal information density. This ensures that the vector meets the input format requirements of the target prediction model, guaranteeing the stability and accuracy of the model's inference.

[0029] The real-time prediction module loads and calls the pre-trained target prediction model, inputs the target feature vector output by the fusion processing module into the model, performs fast online inference calculation, and outputs the target peel force prediction value after the current batch of electrode sheets is rolled in real time. This enables non-destructive, full-batch, second-level online prediction of peel force before or during rolling, providing a basis for production line quality monitoring and process adjustment.

[0030] It is evident that this intelligent peel force prediction system, by integrating basic process characteristics and physical mechanism characteristics, achieves online, non-destructive, and accurate prediction of electrode peel force, effectively reducing testing costs and improving the efficiency of production line quality control.

[0031] For easier understanding, please refer to Figure 3 , Figure 3This diagram illustrates an application scenario of an intelligent peel force prediction system provided in this application. The real-time process dataset serves as the system's data source, representing comprehensive online data for electrode coating and rolling processes. This data includes the coating machine's belt speed, oven temperature, coating gap, the areal density of the dry film measured by the areal density meter, and key process parameters such as the roll gap width and rolling pressure of the rolling mill. This real-time process dataset can be transmitted synchronously and in real-time to the intelligent peel force prediction system via the production line control system or equipment communication interface, providing raw and reliable production condition data for subsequent feature extraction and model inference.

[0032] It should be noted that the intelligent peel force prediction system not only integrates all functional modules such as data acquisition, feature extraction, fusion processing, and real-time prediction, but also integrates visualization and alarm modules. Internally, the system converts the input process data into a peel force prediction result, i.e., the target peel force, through data cleaning, mechanism feature calculation, multi-feature fusion, and model inference. Then, the visualization and alarm module outputs the target peel force to the target user, typically a production line process engineer, quality control personnel, or production line control system. The target user can use this target peel force to predict electrode quality in advance and adjust process parameters promptly when the peel force exceeds the acceptable range, avoiding batch quality defects. Simultaneously, the visualization and alarm module visually presents the prediction results, process trends, and abnormal warning information, assisting users in making rapid decisions and achieving proactive quality control on the production line.

[0033] In one possible embodiment, the peel force prediction curve in the visualization and alarm module is updated in real time with production line data, intuitively displaying the peel force change trend of the current batch of electrodes. The preset qualified alarm threshold of the visualization and alarm module can be 0.1–1.0 N / m. The predicted value within the normal range is presented as a green curve. For example, when the predicted value of the current batch is 0.43 N / m, the curve remains green, indicating that the product quality is stable and under control. When the predicted value suddenly drops to 0.09 N / m at a certain moment, below the preset lower limit threshold, the visualization and alarm module immediately triggers an audible and visual alarm. At the same time, the abnormal data point is highlighted in red on the visualization interface, and an alarm message is sent to the process personnel's mobile phone via the message push module: "Batch A001, predicted peel force 0.09 N / m, below the lower limit threshold, it is recommended to check the rolling pressure or coating surface density," realizing real-time early warning of abnormalities and process intervention guidance.

[0034] As can be seen, by displaying real-time curves, judging thresholds, and pushing alarms through multiple channels, the peel force prediction results are presented intuitively and anomalies are quickly alerted, providing process personnel with accurate quality control basis.

[0035] The following is combined Figure 4 The electronic devices in the embodiments of this application will be described. Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application, such as... Figure 4 As shown, the electronic device includes one or more processors, a memory, a communication interface, and one or more programs. The processor is connected to the memory and the communication interface via an internal communication bus.

[0036] The processor can be used for: Obtain historical process data of multiple batches of electrode sheets during the production process and their corresponding actual peel force values; Feature extraction is performed on the historical process data to obtain a first feature set; Obtain the physical mechanism features corresponding to the first feature set to obtain the second feature set; The first feature set and the second feature set are fused to obtain a reference feature vector. Based on the reference feature vector and the true value of the peel force, a preset initial model is trained to obtain a target prediction model; the target prediction model is used for real-time prediction of the peel force after the electrode is rolled.

[0037] The one or more programs are stored in the aforementioned memory and configured to be executed by the aforementioned processor, and the one or more programs include instructions for performing any step in the above method embodiments.

[0038] The processor can be a central processing unit (CPU), a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, cells, and circuits described in conjunction with the disclosure of this application. The processor can also be a combination that implements computational functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc. The communication unit can be a communication interface, transceiver, transceiver circuit, etc., and the storage unit can be a memory.

[0039] The memory can be volatile or non-volatile, or a combination of both. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0040] It is understood that electronic devices may include more or fewer structural elements than those shown in the above block diagram, such as power modules, physical buttons, Wi-Fi modules, speakers, Bluetooth modules, sensors, display modules, etc., without limitation.

[0041] It should be noted that the intelligent peel force prediction system also includes a prediction model determination module. This module can collect and clean multiple batches of historical process data and corresponding true peel force values ​​to construct a training sample set. Then, through basic feature extraction, physical mechanism feature calculation, and feature fusion processing, it completes iterative model training, hyperparameter tuning, and performance verification, ultimately outputting a deployable target prediction model. It can be understood that this electronic device can be equipped with this prediction model determination module.

[0042] After understanding the software and hardware architecture of this application, the following will be combined with... Figure 5 This application describes a method for determining a peel force prediction model in an embodiment. Figure 5 This is a flowchart illustrating a method for determining a peel force prediction model provided in an embodiment of this application, specifically including the following steps: Step S501: Obtain historical process data of multiple batches of electrode sheets during the production process and their corresponding actual peel force values.

[0043] Each batch of electrode sheets corresponds to a historical process dataset with a true peel force value; the historical process dataset is used to characterize the process parameters and test data of the corresponding batch of electrode sheets in the coating and rolling processes.

[0044] Specifically, based on preset batch filtering conditions (such as batch number / roll number), historical process datasets and multiple true peel force values ​​corresponding to different production batches can be retrieved in batches. The historical process datasets include, but are not limited to: coating process parameters, coating process inspection data, and rolling process parameters. Coating process parameters include coating gap, coating belt speed, oven temperature in each zone, and oven airflow. Coating process inspection data includes the dry film surface density of the electrode coating (which can be measured using an online surface density meter). Rolling process parameters include rolling pressure, roll gap width, rolling belt speed, rolling force, and electrode tension. Each true peel force value is the true peel force value measured by an offline peel force tester after rolling of the corresponding production batch of electrodes. This value can be used as a true label for model training and supervised learning.

[0045] Step S502: Extract features from the historical process data to obtain a first feature set.

[0046] The specific steps for extracting features from the historical process data to obtain the first feature set include: A1. Preprocess the historical process data to obtain preprocessed data; the preprocessing includes at least outlier removal and normalization. A2. Extract the process parameters related to peeling force from the preprocessed data to form the first feature set.

[0047] In a specific embodiment, firstly, the data in each historical process dataset is traversed batch by batch, and outliers (i.e., invalid values, missing values, and abrupt changes that significantly deviate from the normal production range) are identified and removed, thereby obtaining multiple first intermediate datasets. These outliers include, but are not limited to: null values ​​caused by equipment shutdown / roll change / malfunction, sudden increases or decreases in data caused by instantaneous sensor interference, and illegal values ​​exceeding the process window; no specific limitations are imposed here. The determination can be made using the quartile method or threshold filtering, directly deleting outlier rows or smoothly filling them with adjacent normal data; no specific limitations are imposed here either.

[0048] Then, the data from the cleaned first intermediate datasets are normalized to eliminate the interference of different parameter units and numerical ranges on model training, resulting in multiple second intermediate datasets. Min-Max normalization can be used to linearly map all data to the [0, 1] interval to unify the data scale, avoid large-value features suppressing small-value features, and improve model convergence speed and prediction stability.

[0049] Next, coating process data, areal density data, and rolling process data are extracted from multiple second intermediate datasets by batch, and aligned and combined using the batch number as a unique identifier to form a first feature set with uniform dimensions, complete structure, and usable for subsequent mechanism feature calculation.

[0050] It is evident that by removing outliers, normalizing, and extracting key data from historical process data, high-quality and standardized basic feature data are provided for subsequent feature calculation and model training, effectively improving data reliability and model training stability.

[0051] Step S503: Obtain the physical mechanism features corresponding to the first feature set to obtain the second feature set.

[0052] The first feature set includes a first process parameter (i.e., areal density data), a second process parameter (i.e., roll forming process data), and a third process parameter (i.e., coating process data).

[0053] The specific steps for obtaining the physical mechanism features corresponding to the first feature set to obtain the second feature set include: B1. Determine the first physical mechanism characteristic based on the first process parameter and the second process parameter; the first mechanism characteristic characterizes the compaction state of the electrode material. B2. Determine the second physical mechanism characteristic based on the third process parameters; the second physical mechanism characteristic characterizes the drying state; B3. Determine the second feature set based on the first physical mechanism feature and the second physical mechanism feature.

[0054] In a specific embodiment, firstly, taking any batch's first feature set as a reference object (i.e., the reference first feature set), data positioning and splitting are completed based on the batch number corresponding to the reference first feature set. The coating process data, areal density detection data, and rolling process data corresponding to that batch are clearly extracted, ensuring that the three types of data are from the same batch, in the same sequence, and without misalignment. Then, the first physical mechanism feature is obtained by calculation based on the areal density data and the rolling process data. Finally, the second physical mechanism feature is obtained by calculation based on the coating process data.

[0055] Next, the first and second physical mechanism features corresponding to the same reference first feature set are dimensionally concatenated and integrated to form the second feature set corresponding to that batch. Finally, all first feature sets are traversed, and the mechanism features are calculated and combined sequentially to generate multiple second feature sets that correspond one-to-one with each batch.

[0056] It is evident that by introducing physical mechanism features based on areal density, rolling parameters, and coating parameters, the combination of prior process knowledge and data features is achieved, thereby improving the subsequent model's ability to characterize and predict the formation law of peel force.

[0057] For easier understanding, please refer to Figure 6 , Figure 6 This is a flowchart illustrating a method for determining a first physical mechanism feature according to an embodiment of this application. The specific steps for determining the first mechanism feature based on the first process parameters and the second process parameters include: C1. Determine the first target data based on the preset springback coefficient and the second process parameters; C2. Determine the first physical mechanism characteristics based on the first process parameters and the first target data.

[0058] In a specific embodiment, firstly, the roll gap width data is extracted from the second process parameter (i.e., the roll pressing process data). This roll gap width data is a key process parameter preset by the roll press or collected in real time, and it is the basic input data for calculating the compacted thickness of the electrode sheet. Then, based on the batch number, product model, material system, and other identification information carried by the reference first feature set, the electrode sheet (i.e., the reference electrode sheet) corresponding to the batch data is located and determined. Next, based on the material system, formulation, areal density range, and process characteristics of the reference electrode sheet, a pre-calibrated or industry-standard springback coefficient is selected. This springback coefficient is used to compensate for the thickness deviation of the electrode sheet caused by elastic recovery after roll pressing.

[0059] Next, the roll gap width data is multiplied by the springback coefficient to obtain the thickness data of the rolled electrode. This thickness data takes into account the springback effect of the electrode material and can more accurately reflect the actual size of the electrode after compaction. Then, the ratio between the areal density data and the thickness data of the rolled electrode is calculated to obtain the first physical mechanism characteristic (i.e., compaction density). This first physical mechanism characteristic directly characterizes the degree of compaction of the electrode coating and the tightness of the interfacial bonding.

[0060] It is evident that by combining the roll gap width and the electrode springback coefficient to calculate the thickness after roll pressing and introducing the areal density ratio to obtain the compaction density characteristics, the physical mechanism of the roll pressing process is effectively incorporated, providing the model with key inputs that are more in line with the actual interface bonding state, and significantly improving the accuracy of peel force prediction.

[0061] For easier understanding, please refer to Figure 7 , Figure 7 This is a flowchart illustrating a method for determining a second physical mechanism feature according to an embodiment of this application. The specific steps for determining the second physical mechanism feature based on the third process parameters include: D1. Extract the second target data and the third target data from the third process parameters; D2. Determine the second physical mechanism characteristics based on the second target data and the third target data.

[0062] In a specific embodiment, firstly, the conveyor belt speed data (i.e., the second target data) and multiple baking temperature data (i.e., the third target data) are extracted from the third process parameters (i.e., coating process data). The conveyor belt speed data represents the real-time operating rate of the electrode sheet continuously transported within the coating oven, reflecting the duration of a single heating cycle. The multiple baking temperature data represent the actual temperature control parameters for each segmented temperature zone within the coating oven. Different temperature zones are set with differentiated temperatures according to the drying gradient requirements, used to characterize the segmented heating conditions of the electrode sheet.

[0063] Then, the baking temperature data is compared with the corresponding conveyor speed data of the same batch to obtain multiple reference ratios. These reference ratios can quantify the heat intensity per unit conveyor stroke and intuitively reflect the drying effect level of a single temperature zone.

[0064] Finally, the arithmetic mean of multiple reference ratios is taken to weaken the characteristic deviation caused by local temperature differences in each temperature zone, and to comprehensively characterize the overall drying and curing capability of the entire oven for the electrode sheet. This yields the second physical mechanism feature (i.e., drying intensity) used for model training. This second physical mechanism feature integrates the drying effect of all temperature zones in the oven and the overall influence of the conveyor belt speed, and can stably characterize the overall drying sufficiency of the coating, the magnitude of internal residual stress, and the quality of the bonding interface between the coating and the current collector.

[0065] It can be seen that by taking the average of the ratio of baking temperature to conveyor speed in each temperature zone to obtain the drying strength characteristics, the comprehensive influence of coating and drying process on coating adhesion performance is quantified, providing the model with key inputs with clear physical meaning and effectively improving the reliability of peel force prediction.

[0066] Step S504: The first feature set and the second feature set are fused to obtain a reference feature vector.

[0067] The specific steps of fusing the first feature set and the second feature set to obtain the reference feature vector include: E1. Merge the first feature set with the features in the same batch of the second feature set to obtain a fused feature set; E2. Optimize the feature dimensions and standardize the vectors of the fused feature set to obtain the reference feature vector.

[0068] In a specific embodiment, firstly, the first and second feature sets corresponding to the same batch are spliced ​​together in a fixed dimensional order, merging the basic process features and physical mechanism features to form a first intermediate feature set containing coating process features, areal density features, rolling process features, compaction density features, and drying strength features, thus achieving unified integration of multi-source features. Then, feature correlation analysis and redundancy detection are performed on this first intermediate feature set to remove redundant features with highly overlapping information, strong linear correlation, or no contribution to peel force prediction, thereby reducing feature dimensionality and data noise, decreasing model computation, and improving feature effectiveness, resulting in a fused feature set.

[0069] Next, based on feature importance ranking, variance filtering, or preset process weighting rules, the fusion feature set is dimensionally optimized, retaining key features strongly correlated with the formation mechanism of peeling force. The feature structure and information density are further optimized to make the feature set more suitable for the input requirements of the prediction model, resulting in a second intermediate feature set. This second intermediate feature set is then standardized to unify the numerical scale and dimensions of each feature, eliminating the impact of numerical differences between different features on model training and ensuring a balanced contribution from each feature. Finally, the optimized second intermediate feature set is converted into a standardized reference feature vector that can be directly input into the model.

[0070] It is evident that by feature concatenation, redundancy removal, dimensionality optimization, and standardization, efficient fusion and standardization of multiple features can be achieved, reducing data interference and improving model training efficiency and prediction generalization ability.

[0071] Step S505: Based on the reference feature vector and the true value of the peeling force, train the preset initial model to obtain the target prediction model.

[0072] The target prediction model is used for real-time prediction of the peeling force after the electrode is rolled.

[0073] The step of training a preset initial model based on the reference feature vector and the true value of the peeling force to obtain a target prediction model includes the following specific steps: F1. Determine the sample set based on the fused feature vector and the true value of the peeling force; F2. Divide the sample set into a training set, a validation set, and a test set according to a preset ratio; F3. Train the initial model using the training set, and adjust the model parameters according to the validation set until the validation results of the test set meet the preset conditions, thereby obtaining the target prediction model.

[0074] In a specific embodiment, firstly, the true peeling force values ​​corresponding to each batch are used as model label data and combined according to the batch correspondence to form a complete target sample set containing input features and supervision labels. Then, the target sample set is randomly divided into a training set, a validation set, and a test set according to a preset ratio (e.g., 7:2:1). The training set is used for model parameter learning, the validation set is used for model hyperparameter adjustment and overfitting monitoring, and the test set is used for independent evaluation of the final model's generalization ability, ensuring that the data distribution of each subset is consistent and there is no cross-contamination.

[0075] Then, a pre-set initial prediction model (such as a random forest model) is loaded, and the training set data is input into the model. Iterative training is performed based on gradient boosting or ensemble learning strategies, allowing the model to gradually learn the mapping relationship between reference feature vectors and the true values ​​of the stripping force. After basic fitting, a first prediction model with preliminary training is obtained. The first prediction model is then evaluated using validation set data. Based on the validation error, hyperparameters such as the model's learning rate, tree depth, and sampling ratio are adaptively adjusted to reduce the risk of overfitting and improve generalization ability, resulting in a second prediction model with optimized parameters.

[0076] Next, the second prediction model is independently tested using test set data. The coefficient of determination and root mean square error (RMSE) are calculated between the model's predicted values ​​and the actual peel force values ​​to obtain validation results. If the validation results meet preset conditions (e.g., the coefficient of determination and RMSE meet preset thresholds), the model performance is deemed satisfactory, training is immediately stopped, and the current second prediction model is identified as the target prediction model for online prediction. If the validation results do not meet preset conditions, the iterative training step is returned to further adjust the model structure and parameters, continuously optimizing the training. When the validation results meet the standards or the preset maximum number of training iterations is reached, training is terminated, and the final target prediction model is output.

[0077] It is evident that by reasonably dividing the sample set and combining iterative training, parameter tuning, and multiple rounds of performance verification, the model parameters can be effectively optimized, overfitting can be suppressed, and a peeling force target prediction model with excellent accuracy and generalization performance can be obtained.

[0078] In one possible embodiment, historical process data for 100 production batches can be collected. The specific process and testing data for batch 50 are as follows: The coating process data includes: coating belt speed 25m / min, oven zone 1 temperature 85℃, oven zone 2 temperature 90℃, oven zone 3 temperature 88℃, and coating gap 180μm; the areal density data is: dry film areal density 180.5g / m²; the rolling process data includes: rolling pressure 1200kN, roll gap width 120μm, rolling belt speed 30m / min, rolling force 150kN, and electrode tension 200N; the corresponding true peel force value is: 0.43N / m obtained through offline detection. Among these, after anomaly removal, normalization, and key feature extraction of the 50th batch of raw data, the first feature vector corresponding to this batch was constructed: X. 50 =[25,85,90,88,180,180.5,1200,120,30,150,200].

[0079] Then, for the 50th batch of data, physical mechanism features were extracted: Given that the dry film surface density of this batch is 180.5 g / m² and the roll gap width is 120 μm, and considering the rebound characteristics of the electrode material, the rebound coefficient can be preset to 1.05. Using the formula for calculating the thickness after rolling, the electrode thickness after rolling is: Roll gap width × Rebound coefficient = 120 μm × 1.05 = 126 μm = 0.126 mm. Then, the ratio between the dry film surface density and the electrode thickness after rolling is calculated to obtain the first physical mechanism characteristic, namely, the compaction density. This compaction density = surface density / thickness after rolling. To adapt to the model input specifications, the unit can be uniformly converted to g / cm³. 3 Then, the compaction density was calculated to be 1.432 g / cm³. 3 .

[0080] Given that the coating belt speed is 25 m / min and the three temperature zones of the drying oven are 85℃, 90℃, and 88℃, the ratio of temperature to belt speed in each zone is calculated, yielding multiple ratios (3.4, 3.6, and 3.52). The average of these ratios is taken as the second physical mechanism characteristic, i.e., the drying intensity, which is calculated to be 3.51.

[0081] Then, the second eigenvector is determined based on the compaction density and drying strength: Y 50 =[1.432, 3.51]. Concatenating the dimensions of the first and second feature vectors yields the 50th batch of enhanced feature vectors: Z. 50=[25,85,90,88,180,180.5,1200,120,30,150,200,1.432,3.51]. Mechanistic feature calculations and feature fusion were performed sequentially for all 100 batches to obtain an enhanced feature matrix. This enhanced feature matrix was then used in conjunction with the true labels corresponding to the true peeling force values ​​of each batch to train an optimized target prediction model.

[0082] The above primarily describes the solutions of the embodiments of this application from the perspective of the method execution process. It is understood that, in order to achieve the above functions, the electronic device includes corresponding hardware structures and / or software modules for executing each function. Those skilled in the art should readily recognize that, in conjunction with the units and algorithm steps of the various examples described in the embodiments provided herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed by hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0083] This application embodiment can divide the electronic device into functional units according to the above method example. For example, each function can be divided into a separate functional unit, or two or more functions can be integrated into one processing unit. The integrated unit can be implemented in hardware or as a software functional unit. It should be noted that the unit division in this application embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods.

[0084] When dividing each functional unit according to its corresponding function. Figure 8 This is a functional unit block diagram of a peel force prediction model determination device 800 provided in an embodiment of this application. The peel force prediction model determination device 800 includes a historical data acquisition unit 810, a first feature extraction unit 820, a second feature extraction unit 830, a feature fusion processing unit 840, and a prediction model training unit 850, wherein: The historical data acquisition unit 810 is used to acquire historical process data of multiple batches of electrode sheets during the production process and their corresponding true peel force values. The first feature extraction unit 820 is used to extract features from the historical process data to obtain a first feature set; The second feature extraction unit 830 is used to obtain the physical mechanism features corresponding to the first feature set to obtain the second feature set; The feature fusion processing unit 840 is used to fuse the first feature set and the second feature set to obtain a reference feature vector. The prediction model training unit 850 is used to train a preset initial model based on the reference feature vector and the true value of the peeling force to obtain a target prediction model; the target prediction model is used for real-time prediction of the peeling force after the electrode is rolled.

[0085] Optionally, in the step of extracting features from the historical process data to obtain a first feature set, the first feature extraction unit 820 is specifically used for: The historical process data is preprocessed to obtain preprocessed data; the preprocessing includes at least outlier removal and normalization. The process parameters associated with peeling force are extracted from the preprocessed data to form the first feature set.

[0086] Optionally, the first feature set includes a first process parameter, a second process parameter, and a third process parameter. In obtaining the physical mechanism features corresponding to the first feature set to obtain the second feature set, the second feature extraction unit 830 is specifically used for: The first physical mechanism feature is determined based on the first process parameter and the second process parameter; the first mechanism feature characterizes the compaction state of the electrode material. The second physical mechanism characteristic is determined based on the third process parameters; the second physical mechanism characteristic characterizes the drying state. The second feature set is determined based on the first physical mechanism feature and the second physical mechanism feature.

[0087] Optionally, in determining the first mechanistic feature based on the first process parameters and the second process parameters, the second feature extraction unit 830 is further specifically used for: The first target data is determined based on the preset springback coefficient and the second process parameters; The first physical mechanism characteristics are determined based on the first process parameters and the first target data.

[0088] Optionally, in determining the second physical mechanism feature based on the third process parameters, the second feature extraction unit 830 is further specifically used for: Extract the second target data and the third target data from the third process parameters; The second physical mechanism feature is determined based on the second target data and the third target data.

[0089] Optionally, in the process of fusing the first feature set and the second feature set to obtain a reference feature vector, the feature fusion processing unit 840 is specifically used for: The first feature set and the features from the same batch in the second feature set are fused to obtain a fused feature set; The fused feature set is optimized in terms of feature dimension and standardized in terms of vector to obtain the reference feature vector.

[0090] Optionally, in the step of training a preset initial model based on the reference feature vector and the true value of the peeling force to obtain a target prediction model, the prediction model training unit 850 is specifically used for: The sample set is determined based on the fused feature vector and the true value of the peeling force; The sample set is divided into a training set, a validation set, and a test set according to a preset ratio; The initial model is trained using the training set, and the model parameters are adjusted according to the validation set until the validation results of the test set meet the preset conditions, thereby obtaining the target prediction model.

[0091] It is evident that by integrating the basic process characteristics and physical mechanism characteristics of electrode production, the model's prediction accuracy and generalization ability can be effectively improved, enabling real-time and accurate prediction of the peeling force in the electrode rolling process. This avoids the drawbacks of traditional offline testing, such as strong lag and low sampling coverage, thereby improving the production quality and safety level of battery electrodes.

[0092] It should be noted that the specific implementation of each operation can be described in the corresponding description of the method embodiment shown above. The peel force prediction model determination device 800 can be used to execute the method embodiment of this application, and will not be described again here.

[0093] This application also provides a computer-readable storage medium storing a computer program for electronic data interchange, which causes a computer to perform some or all of the steps of any of the methods described in the above method embodiments, wherein the computer includes an electronic device.

[0094] This application also provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program operable to cause a computer to perform some or all of the steps of any of the methods described in the above method embodiments. The computer program product may be a software installation package, and the computer may include an electronic device.

[0095] It should be noted that, for the sake of simplicity, the above embodiments are all described as a series of actions. Those skilled in the art should understand that this application is not limited to the described order of actions, as some steps in the embodiments of this application can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions, steps, modules, or units involved are not necessarily essential to the embodiments of this application.

[0096] In the above embodiments, the descriptions of each embodiment in this application have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0097] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

[0098] The steps of the methods or algorithms described in the embodiments of this application can be implemented in hardware or by a processor executing software instructions. The software instructions can consist of corresponding software modules, which can be stored in RAM, flash memory, ROM, EPROM, electrically erasable programmable read-only memory (EEPROM), registers, hard disk, portable hard disk, read-only optical disk (CD-ROM), or any other form of storage medium well known in the art. An exemplary storage medium is coupled to a processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and storage medium can reside in an ASIC. Furthermore, the ASIC can reside in a terminal device or management device. Alternatively, the processor and storage medium can exist as discrete components in the terminal device or management device.

[0099] Those skilled in the art will recognize that, in one or more of the examples above, the functions described in the embodiments of this application can be implemented, in whole or in part, by software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. This computer program product includes one or more computer instructions. When these computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., digital video discs (DVDs)), or semiconductor media (e.g., solid-state disks (SSDs)).

[0100] The modules / units included in the various devices and products described in the above embodiments can be software modules / units, hardware modules / units, or a combination of both. For example, for devices and products applied to or integrated into a chip, all modules / units can be implemented using hardware methods such as circuits, or at least some modules / units can be implemented using software programs that run on a processor integrated within the chip, while the remaining (if any) modules / units can be implemented using hardware methods such as circuits. For devices and products applied to or integrated into a chip module, all modules / units can be implemented using hardware methods such as circuits. Different modules / units can be located in the same component (e.g., chip, circuit module, etc.) or different components of the chip module, or at least some modules / units can be implemented using hardware methods such as circuits. The implementation is achieved through a software program that runs on the processor integrated within the chip module. The remaining modules / units (if any) can be implemented using hardware methods such as circuits. For various devices and products applied to or integrated into terminal equipment, each of their modules / units can be implemented using hardware methods such as circuits. Different modules / units can be located in the same component (e.g., chip, circuit module, etc.) or different components within the terminal equipment. Alternatively, at least some modules / units can be implemented through a software program that runs on the processor integrated within the terminal equipment, while the remaining modules / units (if any) can be implemented using hardware methods such as circuits.

[0101] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the embodiments of this application. It should be understood that the above descriptions are merely specific embodiments of the embodiments of this application and are not intended to limit the protection scope of the embodiments of this application. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solutions of the embodiments of this application should be included within the protection scope of the embodiments of this application.

Claims

1. A method for determining a peel force prediction model, characterized in that, The method includes: Obtain historical process data of multiple batches of electrode sheets during the production process and their corresponding actual peel force values; Feature extraction is performed on the historical process data to obtain a first feature set; Obtain the physical mechanism features corresponding to the first feature set to obtain the second feature set; The first feature set and the second feature set are fused to obtain a reference feature vector. Based on the reference feature vector and the true value of the peel force, a preset initial model is trained to obtain a target prediction model; the target prediction model is used for real-time prediction of the peel force after the electrode is rolled.

2. The method as described in claim 1, characterized in that, The step of extracting features from the historical process data to obtain a first feature set includes: The historical process data is preprocessed to obtain preprocessed data; the preprocessing includes at least outlier removal and normalization. The process parameters associated with peeling force are extracted from the preprocessed data to form the first feature set.

3. The method as described in claim 2, characterized in that, The first feature set includes a first process parameter, a second process parameter, and a third process parameter. Obtaining the physical mechanism features corresponding to the first feature set to obtain the second feature set includes: The first physical mechanism feature is determined based on the first process parameter and the second process parameter; the first mechanism feature characterizes the compaction state of the electrode material. The second physical mechanism characteristic is determined based on the third process parameters; the second physical mechanism characteristic characterizes the drying state. The second feature set is determined based on the first physical mechanism feature and the second physical mechanism feature.

4. The method as described in claim 3, characterized in that, Determining the first mechanism feature based on the first process parameters and the second process parameters includes: The first target data is determined based on the preset springback coefficient and the second process parameters; The first physical mechanism characteristics are determined based on the first process parameters and the first target data.

5. The method as described in claim 3, characterized in that, The determination of the second physical mechanism characteristic based on the third process parameters includes: Extract the second target data and the third target data from the third process parameters; The second physical mechanism feature is determined based on the second target data and the third target data.

6. The method according to any one of claims 1-5, characterized in that, The process of fusing the first feature set and the second feature set to obtain a reference feature vector includes: The first feature set and the features from the same batch in the second feature set are fused to obtain a fused feature set; The fused feature set is optimized in terms of feature dimension and standardized in terms of vector to obtain the reference feature vector.

7. The method according to any one of claims 1-5, characterized in that, The step of training a preset initial model based on the reference feature vector and the true value of the peeling force to obtain a target prediction model includes: The sample set is determined based on the reference feature vector and the true value of the peeling force; The sample set is divided into a training set, a validation set, and a test set according to a preset ratio; The initial model is trained using the training set, and the model parameters are adjusted according to the validation set until the validation results of the test set meet the preset conditions, thereby obtaining the target prediction model.

8. A peeling force intelligent prediction system, said system being deployed with a target prediction model determined by the method as described in any one of claims 1-7, characterized in that, The system includes a data acquisition module, a feature extraction module, a fusion processing module, and a real-time prediction module, wherein: The data acquisition module is used to acquire real-time process datasets; The feature extraction module is used to extract features from the real-time process dataset to obtain a third feature set; and to obtain the physical mechanism features corresponding to the third feature set to obtain a fourth feature set. The fusion processing module is used to fuse the third feature set and the fourth feature set to obtain the target feature vector; The real-time prediction module is used to input the target feature vector into the target prediction model and output the target peeling force.

9. An electronic device, characterized in that, include: Processor, memory, communication interface, and one or more programs; The one or more programs are stored in the memory and configured to be executed by the processor, the programs including instructions for performing the steps of the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, the computer program including program instructions that, when executed by a processor, cause the processor to perform the method as described in any one of claims 1-7.

Images