Electrode material dislocation adjusting method, device and equipment and storage medium

Abstract

The invention provides an electrode material dislocation adjusting method, device and equipment and a storage medium, and relates to the technical field of battery production. The method comprises the following steps: acquiring pre-process data of a to-be-wound electrode material, wherein the pre-process data is a process parameter value of the to-be-wound electrode material before a winding process; processing the pre-process data through a process matching model to obtain a target process parameter corresponding to an ideal alignment state in the to-be-wound electrode material; processing the target process parameters and the pre-process data through a tab dislocation prediction model to obtain a tab dislocation prediction value in the to-be-wound electrode material; and if the dislocation prediction value meets a preset condition, generating a recommendation instruction based on the target process parameter, thereby realizing closed-loop control of winding and a pre-process, and effectively realizing tab dislocation adjustment.

Description

Technical Field

[0001] This disclosure relates to the field of battery manufacturing technology, and in particular to a method for adjusting electrode material misalignment, an electrode material misalignment adjustment device, an electronic device, and a computer-readable storage medium. Background Technology

[0002] Battery winding is a crucial step in battery production, directly impacting battery performance, safety, and lifespan. During the winding process, positive and negative electrode sheets, along with the separator, are alternately stacked and wound into a core structure. The tabs are the connection points between the electrodes and external circuitry; their alignment accuracy is critical to the overall battery performance.

[0003] In related technologies, the adjustment of tab misalignment mainly relies on the automated control capabilities of the winding machine. However, the above adjustment method is prone to problems such as large fluctuations in product yield and low overall process controllability. Summary of the Invention

[0004] This disclosure provides a method, apparatus, device, and storage medium for adjusting electrode material misalignment, which at least to some extent overcomes the problems of large fluctuations in product yield and low overall process controllability in related technologies for adjusting electrode tab misalignment.

[0005] Other features and advantages of this disclosure will become apparent from the following detailed description, or may be learned in part from practice of this disclosure.

[0006] According to one aspect of this disclosure, a method for adjusting misalignment in the production of electrode materials is provided, comprising: acquiring preliminary process data of the electrode material to be wound, wherein the preliminary process data is the process parameter values ​​of the electrode material to be wound in processes prior to the winding process; processing the preliminary process data using a process matching model to obtain target process parameters corresponding to the ideal alignment state of the electrode material to be wound; processing the target process parameters and the preliminary process data using a tab misalignment prediction model to obtain a predicted misalignment value of the tabs in the electrode material to be wound; and generating a recommended instruction based on the target process parameters if the predicted misalignment value meets preset conditions.

[0007] In one embodiment of this disclosure, the target process parameters include pressure roller pressure and / or variable roll diameter servo position; when the target process parameters include pressure roller pressure, the preceding process data includes first process data, and the process matching model includes a first matching model, which is trained using historical first process data of the target tab and corresponding pressure roller pressure labels; when the target process parameters include variable roll diameter servo position, the preceding process data includes second process data, and the process matching model includes a second matching model, which is trained using historical second process data of the target tab and corresponding variable roll diameter servo position labels.

[0008] In one embodiment of this disclosure, before processing the pre-process data using a process matching model to obtain the target process parameters corresponding to the ideal alignment state of the tabs in the electrode material to be wound, the method further includes: acquiring historical first process data of the target tabs and corresponding process parameter labels; constructing a training set and a validation set based on a first preset ratio; iteratively training the process matching model to be trained based on the training set to obtain an initial process matching model; evaluating the initial process matching model based on the validation set to obtain a model evaluation index; if the model evaluation index does not meet preset performance conditions, adjusting the model parameters of the initial process matching model; if the model evaluation index meets the preset performance conditions, determining the initial process matching model as the process matching model.

[0009] In one embodiment of this disclosure, before iteratively training the process matching model to be trained based on the training set to obtain an initial process matching model, the method further includes: initializing the hyperparameters of the process matching model to be trained; after determining the initial process matching model as the process matching model if the model evaluation index meets the preset performance condition, the method further includes: based on the validation set, performing hyperparameter tuning on the hyperparameters of the initial process matching model to determine the optimal hyperparameter combination, and configuring the process matching model according to the optimal hyperparameter combination.

[0010] In one embodiment of this disclosure, the method further includes: calculating the feature contribution of each historical first process data based on the process matching model; sorting the feature contribution, filtering the preceding process parameters corresponding to the historical first process data whose feature contribution is greater than a preset contribution threshold, and / or filtering the preceding process parameters corresponding to a preset number of historical first process data that are ranked first, as input to the process matching model.

[0011] In one embodiment of this disclosure, the model evaluation index includes at least one of mean squared error, mean absolute error, and coefficient of determination; wherein the model evaluation index satisfies the preset performance conditions, including at least one of the following: the mean squared error is less than or equal to a first error threshold; the mean absolute error is less than or equal to a second error threshold; and the coefficient of determination is greater than or equal to a preset coefficient threshold.

[0012] In one embodiment of this disclosure, the preprocess data includes second process data; the electrode misalignment prediction model is trained using the target process parameters, the second process data, and the corresponding electrode misalignment amount label.

[0013] In one embodiment of this disclosure, the process matching model and the tab misalignment prediction model include at least one gradient boosting decision tree model selected from lightweight gradient boosting machine, extreme gradient boosting decision tree model, random forest, and support vector machine.

[0014] In one embodiment of this disclosure, the first process data includes at least one of the following: average surface density of both sides of cathode coating, average surface density of both sides of anode coating, winding and feeding-anode roll forming pre-splitting in-process time, average anode roll forming thickness, anode roll forming speed, winding and feeding-cathode roll forming pre-splitting in-process time, average cathode roll forming thickness, and cathode roll forming speed.

[0015] In one embodiment of this disclosure, the second process data includes at least one of: anode die-cutting-anode roll forming pre-slitting in-process time, cathode die-cutting-cathode roll forming pre-slitting in-process time, average cathode roll forming thickness, cathode roll forming speed, average cathode double-sided surface density, average anode roll forming thickness, and average anode double-sided surface density.

[0016] In one embodiment of this disclosure, the preset condition includes a preset misalignment threshold; wherein, generating a recommendation instruction based on the target process parameter if the misalignment prediction value meets the preset condition includes: if the misalignment prediction value is less than or equal to the preset misalignment threshold, determining that the misalignment prediction value meets the preset condition; if the misalignment prediction value is greater than the preset misalignment threshold, determining that the misalignment prediction value does not meet the preset condition.

[0017] In one embodiment of this disclosure, the misalignment prediction value includes at least one of the following: electrode misalignment amount, electrode misalignment probability, or electrode misalignment level.

[0018] In one embodiment of this disclosure, the method further includes: if the misalignment prediction value does not meet the preset condition, then the target process parameters are not recommended.

[0019] According to another aspect of this disclosure, an electrode material misalignment adjustment device is provided, comprising: a data acquisition module for acquiring preliminary process data of the electrode material to be wound, wherein the preliminary process data is the process parameter value of the electrode material to be wound in the process prior to the winding process; a parameter determination module for processing the preliminary process data through a process matching model to obtain target process parameters corresponding to the ideal alignment state of the electrode material to be wound; a misalignment detection module for processing the target process parameters and the preliminary process data through a tab misalignment prediction model to obtain a predicted misalignment value of the tab in the electrode material to be wound; and a parameter recommendation module for generating a recommendation instruction based on the target process parameters if the predicted misalignment value meets preset conditions.

[0020] According to another aspect of this disclosure, an electronic device is also provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to perform the electrode material misalignment adjustment method described in any one of the preceding claims by executing the executable instructions.

[0021] According to another aspect of this disclosure, a computer-readable storage medium is also provided, on which a computer program is stored, which, when executed by a processor, implements the electrode material misalignment adjustment method described in any of the preceding claims.

[0022] According to another aspect of this disclosure, a computer program product is also provided, comprising a computer program that, when executed by a processor, implements the electrode material misalignment adjustment method described in any of the preceding claims.

[0023] In this embodiment, by deeply exploring the relationship between the pre-process data of the electrode material to be wound and the process parameters of the winding process, a bridge is built between the pre-process and the winding process, realizing the full-link connection of process data. The target process parameters predicted by the process matching model provide strong support for subsequent tab misalignment adjustment. The tab misalignment prediction model performs intelligent analysis on the target process parameters and pre-process data to predict the tab misalignment prediction value, which greatly improves the accuracy and timeliness of misalignment risk prediction. When the misalignment prediction value meets expectations, the recommended instructions generated based on the target process parameters can directly empower the winding equipment. This not only realizes the dynamic optimization of the winding process parameters, but also builds a closed-loop adjustment mechanism between the pre-process and the winding process, significantly improving the stability and reliability of battery product quality.

[0024] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0025] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure. It is obvious that the drawings described below are merely some embodiments of this disclosure, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0026] Figure 1 This diagram illustrates the structure of an energy storage system provided in an embodiment of the present disclosure.

[0027] Figure 2 The diagram shows a flowchart of an electrode material misalignment adjustment method provided in an embodiment of this disclosure.

[0028] Figure 3 The flowchart illustrates a process matching model training method provided in an embodiment of this disclosure.

[0029] Figure 4 A flowchart of another process matching model training method provided in an embodiment of this disclosure is shown.

[0030] Figure 5 A flowchart of a feature selection method provided in an embodiment of this disclosure is shown.

[0031] Figure 6 This diagram illustrates another electrode material misalignment adjustment method provided in an embodiment of the present disclosure.

[0032] Figure 7 This diagram illustrates the structure of an electrode material misalignment adjustment device provided in an embodiment of the present disclosure.

[0033] Figure 8 This diagram illustrates a structural block diagram of an electronic device provided in an embodiment of the present disclosure. Detailed Implementation

[0034] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that this disclosure will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0035] Furthermore, the accompanying drawings are merely illustrative of this disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0036] Because the energy we need is highly time- and space-dependent, in order to utilize energy rationally and improve energy efficiency, it is necessary to store one form of energy in the same way or by converting it into another, and then release it in a specific energy form based on future application needs. Currently, the main way to generate green electricity is to develop green energy sources such as photovoltaics and wind power to replace fossil fuels.

[0037] Currently, the generation of green electricity generally relies on solar, wind, and hydropower. However, wind and solar power are generally characterized by strong intermittency and large fluctuations, which can cause grid instability, insufficient power during peak demand periods, and excessive power during off-peak periods. Unstable voltage can also damage the power grid. Therefore, insufficient electricity demand or insufficient grid capacity may lead to the problem of "wind and solar curtailment." Solving these problems requires energy storage. This involves converting electrical energy into other forms of energy through physical or chemical means and storing it. When needed, this energy can be converted back into electrical energy and released. Simply put, energy storage is like a large "power bank," storing electrical energy when solar and wind power are abundant and releasing the stored electricity when needed.

[0038] Taking electrochemical energy storage as an example, this solution provides an energy storage device 110, which is applied to the energy storage system 100. The energy storage device 110 is equipped with a set of chemical batteries, which mainly use the chemical elements in the batteries as energy storage medium. The charging and discharging process is accompanied by the chemical reaction or change of the energy storage medium. Simply put, the electrical energy generated by wind and solar energy is stored in the chemical batteries. When the use of external electrical energy reaches its peak, the stored electrical energy is released for use, or transferred to places with a shortage of electricity for use.

[0039] Current energy storage applications are quite widespread, including generation-side energy storage, grid-side energy storage, and consumption-side energy storage. The corresponding types of energy storage devices 110 include: (1) Large-scale energy storage power stations (including multiple prefabricated energy storage modules) applied to wind power and photovoltaic power stations can help renewable energy power generation meet grid connection requirements and improve the utilization rate of renewable energy. As a high-quality active / reactive power regulation power source on the power supply side, energy storage power stations can achieve load matching of power in time and space, enhance the renewable energy absorption capacity, reduce instantaneous power changes, reduce the impact on the power grid, improve the absorption of new energy power generation, and are of great significance in power grid system backup, alleviating peak load power supply pressure and peak regulation and frequency regulation. (2) The energy storage prefabricated cabin applied on the grid side mainly functions as peak regulation, frequency regulation and grid congestion relief. In terms of peak regulation, it can realize peak shaving and valley filling of electricity load, that is, charging the energy storage battery when the electricity load is low and releasing the stored electricity during the peak electricity load period, thereby achieving a balance between power production and consumption. (3) Small energy storage cabinets applied to the electricity consumption side mainly function as self-consumption of electricity, peak-valley price arbitrage, capacity cost management, and improvement of power supply reliability. Depending on the application scenario, electricity consumption side energy storage can be divided into industrial and commercial energy storage cabinets, household energy storage devices, energy storage charging piles, etc., which are generally used in conjunction with distributed photovoltaics. Industrial and commercial users can use energy storage for peak-valley price arbitrage and capacity cost management. In the electricity market implementing peak-valley pricing, by charging the energy storage system when the electricity price is low and discharging the energy storage system when the electricity price is high, peak-valley price arbitrage can be achieved, reducing electricity costs. In addition, industrial enterprises subject to two-part tariffs can use energy storage systems to store energy during off-peak hours and discharge during peak loads, thereby reducing peak power and the maximum demand declared, achieving the goal of reducing capacity costs. Household photovoltaics with energy storage can improve the level of self-consumption of electricity. Due to high electricity prices and poor power supply stability, the demand for household photovoltaic installations is driven. Given that photovoltaic power generation occurs during the day, while user load is generally higher at night, configuring energy storage can better utilize photovoltaic power, improve self-consumption levels, and reduce electricity costs. Furthermore, energy storage is needed in areas such as communication base stations and data centers for backup power.

[0040] In some embodiments, see Figure 1 , Figure 1 This is a schematic diagram of the structure of an energy storage system 100 according to an embodiment of this application. Figure 1 And this application Figure 1 The embodiments are illustrated using a shared energy storage scenario on the generation / distribution side as an example. The energy storage device in this application is not limited to a prefabricated energy storage module in a generation / distribution energy storage scenario.

[0041] This application provides an energy storage system 100, which includes: a high-voltage cable 120, a first power conversion device 130, a second power conversion device 140, and an energy storage device 110 provided in this application. In some embodiments of the power generation scenario, the second power conversion device 140 can be a wind power conversion device. Since the electricity generated by wind power conversion is volatile, random, and intermittent, the unstable electricity output by the wind power conversion device can be stored in the energy storage device through grid connection. The energy storage device is connected to the high-voltage cable 120 and outputs smooth electricity to the power consumption side of the distribution network, realizing peak shaving and frequency regulation, and ensuring stable grid operation; or, the wind power conversion device... The device is always connected to the high-voltage cable 120. Under normal power generation conditions, the power output of the wind power conversion device is supplied to the power consumption side of the distribution network through the high-voltage cable. When the current power load is low and the wind power conversion device generates excess power, the excess power is first stored in the energy storage device 110 to reduce wind and solar curtailment rates and improve the problem of new energy power generation consumption. When the power load is high, the power grid issues an instruction to transmit the power stored in the energy storage device 110 together with the high-voltage cable 120 in grid-connected mode to supply power to the power consumption side. This provides the power grid with various services such as peak shaving, frequency regulation, and backup, giving full play to the peak shaving role of the power grid, promoting peak shaving and valley filling, and alleviating the power supply pressure of the power grid.

[0042] In some embodiments on the distribution network side, the first power conversion device 130 can be a photovoltaic panel, and the energy storage device 110 is connected to the high-voltage cable 120 and installed downstream of the high-voltage cable 120 between the user load and the user load. The electrical energy output by the photovoltaic panel is stored in the energy storage device 110, which can respond in a timely manner to act as a backup power source when the power grid / distribution network fails; or, it can provide power supply support to alleviate line congestion when the high-voltage cable 120 transmission line is blocked, and to delay the economic pressure caused by the expansion of the power grid / distribution capacity when the power grid is planned to be expanded.

[0043] Optionally, the first power conversion device 130 may include, but is not limited to, a photovoltaic panel, and the second power conversion device 140 may include, but is not limited to, a wind power conversion device. The first power conversion device 130 and the second power conversion device 140 can convert at least one of solar energy, light energy, wind energy, thermal energy, tidal energy, biomass energy, and mechanical energy into electrical energy.

[0044] Optionally, the energy storage device 110 may include, but is not limited to, energy storage applications such as energy storage power stations, hydropower / thermal / wind power generation systems, solar power generation systems, mobile power systems, smart home systems, or temporary power supply systems, and may also be applied in multiple fields such as data centers, military equipment, aerospace, charging piles, and electric vehicles.

[0045] Optionally, the energy storage device 110 may include battery modules, battery packs, battery clusters, mobile power supplies, energy storage cabinets / prefabricated energy storage compartments, and other battery integrated systems composed of individual batteries. The actual application form of the energy storage device 110 provided in this application embodiment may be, but is not limited to, the listed products, and may also be other application forms. This application embodiment does not strictly limit the application form of the energy storage device 110.

[0046] Alternatively, the single cell is not limited to at least one of cylindrical, square, prismatic, or other shaped cells.

[0047] Optionally, the single cell can be a rechargeable battery, which refers to a single cell that can be recharged after discharge to activate the active materials and continue to be used. The single cell can be a lithium-ion battery, sodium-ion battery, sodium-lithium-ion battery, lithium metal battery, sodium metal battery, lithium-sulfur battery, magnesium-ion battery, nickel-metal hydride battery, nickel-cadmium battery, lead-acid battery, etc., and this application does not specifically limit it.

[0048] Battery manufacturing processes mainly include electrode preparation, die cutting, electrode slitting, and winding. Among these, the winding process is a critical step in battery production, affecting battery performance and lifespan. During the battery winding process, the positive electrode, negative electrode, and separator are wound together by the precision winding needle mechanism of the winding machine to form a core structure. This ensures that adjacent positive and negative electrode sheets are effectively isolated by the separator, preventing potential short-circuit risks.

[0049] The tab is the connection point between the electrode and the external circuit. Tab misalignment directly affects the battery's charging and discharging stability and safety performance. Addressing tab misalignment in battery production typically relies on the automated control capabilities of the winding machine, for example, by adjusting the pressure of the clamping roller (also known as the P-roller) to reduce the degree of tab misalignment. This adjustment mechanism can intervene in the tab position to some extent, and its effects gradually become apparent after the equipment reaches a stable operating stage, achieving a certain degree of expected control. However, this adjustment method is limited to the winding process, forming a relatively closed control system.

[0050] Although the automatic adjustment of the P-roller pressure can play a role after the equipment is running stably, in the initial stage of equipment startup, the battery production system has not reached a stable state, and the electrode misalignment cannot be effectively controlled, resulting in large fluctuations in product yield during the startup phase.

[0051] Current closed-loop control systems are limited to the winding process and lack the ability to adjust process parameters in conjunction with upstream and downstream processes (such as electrode preparation, material processing before stacking / winding, etc.). This makes the response of the production process to electrode misalignment anomalies passive and isolated. Once an electrode misalignment anomaly occurs, temporary remedial measures can only be taken after the problem has occurred, resulting in low overall process controllability, serious delays in adjustment measures, long problem investigation cycles, and difficulty in achieving efficient and stable battery production.

[0052] To address at least some of the aforementioned technical problems, the electrode material misalignment adjustment method provided in this disclosure deeply mines the relationship between the pre-process data of the electrode material to be wound and the process parameters of the winding process, constructing a bridge between the pre-process and the winding process, thus achieving full-link connectivity of process data. The target process parameters predicted by the process matching model provide strong support for subsequent electrode misalignment adjustment; the electrode misalignment prediction model intelligently analyzes the target process parameters and pre-process data to predict the misalignment value of the electrode, significantly improving the accuracy and timeliness of misalignment risk prediction; when the misalignment prediction value meets expectations, the recommended instructions generated based on the target process parameters can directly empower the winding equipment, not only realizing the dynamic optimization of the winding process parameters, but also constructing a closed-loop adjustment mechanism between the pre-process and the winding process, significantly improving the stability and reliability of battery product quality.

[0053] This disclosure provides a method for adjusting electrode material misalignment, which can be executed by any electronic device with computing capabilities, such as a controller. The controller may include a programmable logic controller (PLC) or a microcontroller, as long as it can perform the corresponding functions.

[0054] Figure 2 This diagram illustrates a flowchart of an electrode material misalignment adjustment method according to an embodiment of the present disclosure, as follows: Figure 2 As shown, the electrode material misalignment adjustment method provided in this embodiment includes the following steps S202~S208. Wherein: S202. Obtain the preliminary process data of the electrode material to be wound. The preliminary process data is the process parameter value of the electrode material before the winding process.

[0055] The electrode material to be wound includes electrode materials used to form a core, and may include a positive electrode sheet, a negative electrode sheet, and a separator. In the battery manufacturing process, the electrode material to be wound is formed in a pre-process; in the winding process, the electrode material is wound and shaped by a winding machine to form a core structure.

[0056] In one embodiment, the process preceding the winding process is called the pre-process, which may include processes such as electrode preparation, electrode cutting, and tab preparation.

[0057] The electrode preparation process includes a stirring process, a coating process, and a rolling process. In the stirring process, positive electrode active materials (such as ternary materials, lithium iron phosphate, etc.) and negative electrode active materials (such as graphite) are mixed with binders, conductive agents, and solvents in specific proportions to form uniform positive and negative electrode slurries. In the coating process, the positive and negative electrode slurries are uniformly applied to the surface of the current collector to form a continuous electrode layer. In the rolling process, the coated electrodes are rolled using a rolling mill to compact the electrode coating to the target thickness.

[0058] The electrode cutting and tab preparation process is used to cut the rolled electrode into narrow electrode rolls by a slitting machine, and to prepare tabs at both ends of the slitting electrode.

[0059] The preliminary process data for the electrode material to be wound includes the time for anode die-cutting-anode roll forming pre-cutting in-process (WIP), time for cathode die-cutting-cathode roll forming pre-cutting in-process, time for winding and feeding-anode roll forming pre-cutting in-process, time for winding and feeding-cathode roll forming pre-cutting in-process, average cathode roll forming thickness, cathode roll forming speed, average cathode coating surface density on both sides, average anode roll forming thickness, anode roll forming speed, and average anode coating surface density on both sides.

[0060] The time between anode die-cutting and anode roll forming pre-cutting in-process and cathode die-cutting and cathode roll forming pre-cutting in-process is the period from when the electrode material, including the positive and negative electrode sheets, enters the die-cutting process after roll forming pre-cutting until the material completes the die-cutting process and is transferred out of the die-cutting process.

[0061] The winding and feeding-anode roll pre-cutting work-in-process time and the winding and feeding-cathode roll pre-cutting work-in-process time are the time interval from when the electrode material after roll pre-cutting enters the winding and feeding stage until the material is successfully loaded onto the winding machine's material shaft and ready to start winding.

[0062] The average thickness of the cathode and the average thickness of the anode are obtained by measuring the thickness of the electrode multiple times within the detection area after the electrode has undergone rolling treatment and calculating the average thickness.

[0063] The average thickness of cathode roll forming, the average thickness of anode roll forming, the cathode roll forming speed, and the anode roll forming speed are all key process parameters used to measure the processing quality of the roll forming process.

[0064] The average surface density of the cathode coating and the average surface density of the anode coating refer to the average surface density of the electrode current collector after the slurry is applied to the front and back sides of the electrode multiple times within a specified testing area. This average value is used to evaluate the uniformity of the coating process and ensure the stability of subsequent processes.

[0065] In one embodiment, after obtaining the preceding process data, the preceding process data can be preprocessed. Preprocessing includes, but is not limited to, data cleaning, format conversion, normalization or standardization, feature selection or extraction, etc., so as to improve data quality, eliminate noise interference, make it meet the requirements of subsequent model input, and provide highly reliable data support for electrode misalignment adjustment.

[0066] First, data cleaning removes erroneous data and fills in missing information in the preceding process data, ensuring basic accuracy and consistency. Next, the cleaned preceding process data undergoes format conversion and normalization to unify the expression of data from different sources, ensuring compatibility. Following this is data integration, which merges multi-source data to obtain fused data, forming a more comprehensive view and resolving potential data conflicts. Finally, through fused data reshaping, the structure of the fused data is adjusted to conform to the data format of the process matching model and the tab misalignment prediction model.

[0067] S204. Process the preceding process data through the process matching model to obtain the target process parameters corresponding to the ideal alignment state of the electrode material to be wound.

[0068] In one embodiment, the process matching model can be trained based on historical pre-process data and process parameter labels. It is used to characterize the correlation between the pre-process data and the target process parameters corresponding to the tabs in the electrode material to be wound under ideal alignment. That is, the pre-process data is input into the process matching model, and the target process parameters are output. It can be understood that the predicted target process parameters indicate that the tabs in the electrode material to be wound are in an ideal alignment state.

[0069] The ideal alignment of the electrode material to be wound can be characterized by the alignment of the tabs in the electrode material. Furthermore, the ideal alignment of the electrode material to be wound means that the misalignment of the positive and negative tabs in the electrode width direction is less than a preset misalignment threshold, the misalignment in the electrode length direction is less than a preset misalignment threshold, and the thickness of the positive and negative tabs in the electrode thickness direction matches the corresponding electrode thickness.

[0070] The aforementioned preset misalignment threshold can be pre-configured in the controller. The value of the preset misalignment threshold can be determined according to actual needs. For example, the preset misalignment threshold can be 0.05mm, 0.1mm, etc.

[0071] In one embodiment, the process matching model is a trained process matching model or a fully trained process matching model, meaning the accuracy of the process matching model meets expectations. The process matching model includes at least one gradient boosting decision tree (GBDT) model selected from Lightweight Gradient Boosting Machine (LightGBM), eXtreme Gradient Boosting (XGBoost), Random Forest, and Support Vector Machine (SVM).

[0072] LightGBM is a framework based on the GBDT algorithm, supporting efficient and scalable parallel training. XGBoost is a model optimization method that uses Gradient Boosting Decision Tree (GBDT) as a base learner, continuously adding new decision trees to fit the residuals of previous models.

[0073] S206. The target process parameters and the preceding process data are processed by the tab misalignment prediction model to obtain the predicted value of the tab misalignment in the electrode material to be wound.

[0074] In one embodiment, the tab misalignment prediction model is used to characterize the correlation between the target process parameters and the preceding process data and the predicted value of tab misalignment in the electrode material to be wound.

[0075] The tab misalignment prediction model can be trained using historical target process parameters, historical preceding process data, and corresponding misalignment prediction value labels. Specifically, the historical target process parameters and historical preceding process data are input into the tab misalignment prediction model, which outputs the predicted misalignment value of the tab in the electrode material to be wound.

[0076] The above misalignment prediction value is used to measure the degree of misalignment between the positive and negative tabs in the electrode material to be wound.

[0077] In one embodiment, the tab misalignment prediction model includes at least one gradient boosting decision tree model selected from lightweight gradient boosting machine, extreme gradient boosting decision tree model, random forest, and support vector machine. The training method of the tab misalignment prediction model is similar to that of the process matching model in the aforementioned embodiment, with the only difference being the input and output data of the model; the similarities will not be elaborated further.

[0078] S208. If the misalignment prediction value meets the preset conditions, a recommended instruction based on the target process parameters is generated.

[0079] In one embodiment, preset conditions can be pre-configured in the controller, and the preset conditions are determined according to actual needs.

[0080] When the misalignment prediction value meets the preset conditions, it indicates that the misalignment prediction value obtained after winding the electrode material to be wound based on the target process parameters can meet the process requirements. That is, it is determined that the production risk under the target process parameters is within a controllable range. At this time, a recommendation instruction is generated. The recommendation instruction is used to push the target process parameters to the user. The pushed target process parameters can pop up a prompt window on the equipment control terminal of the winding machine. The prompt window displays the target process parameters and recommends that operators or automation systems use the target process parameters for subsequent production operations.

[0081] When the misalignment prediction value does not meet the preset conditions, it indicates that the misalignment prediction value obtained after winding the electrode material based on this target process parameter cannot meet the process requirements. In other words, it is determined that the target process parameter may lead to an excessively high risk of electrode misalignment. At this time, no recommended instruction is generated, that is, no prompt window is displayed on the equipment control terminal of the winding machine or no operation based on the target process parameter is performed.

[0082] In this embodiment, by deeply exploring the relationship between the pre-process data of the electrode material to be wound and the process parameters of the winding process, a bridge is built between the pre-process and the winding process, realizing the full-link connection of process data. The target process parameters predicted by the process matching model provide strong support for subsequent tab misalignment adjustment. The tab misalignment prediction model performs intelligent analysis on the target process parameters and pre-process data to predict the tab misalignment prediction value, which greatly improves the accuracy and timeliness of misalignment risk prediction. When the misalignment prediction value meets expectations, the recommended instructions generated based on the target process parameters can directly empower the winding equipment. This not only realizes the dynamic optimization of the winding process parameters, but also builds a closed-loop adjustment mechanism between the pre-process and the winding process, significantly improving the stability and reliability of battery product quality.

[0083] In one embodiment, the target process parameters include pressure roller pressure and / or variable roll diameter servo position; when the target process parameters include pressure roller pressure, the preceding process data includes first process data, and the process matching model includes a first matching model, which is trained using the historical first process data of the target tab and the corresponding pressure roller pressure label; when the target process parameters include variable roll diameter servo position, the preceding process data includes second process data, and the process matching model includes a second matching model, which is trained using the historical second process data of the target tab and the corresponding variable roll diameter servo position label.

[0084] In battery winding equipment, the variable roll diameter servo system is a core component that compensates for tension and positional deviations caused by the increase in cell diameter during winding by adjusting the position of the servo motor of the winding needle in real time, ensuring that the tabs are always in an ideal alignment state. The variable roll diameter servo position corresponding to the ideal alignment state of the tabs in the electrode material to be wound refers to the servo position corresponding to the ideal alignment state of the tabs during the winding process by real-time detection of roll diameter and position feedback, and staged position compensation.

[0085] In battery winding equipment, the pressure of the clamping roller is the core process parameter of the winding process. By changing the bonding state between the electrode and the separator, the tension transmission law, and the stability of the winding, it affects the alignment accuracy of the tabs.

[0086] In one embodiment, the first matching model and the second matching model may include at least one gradient boosting decision tree model selected from lightweight gradient boosting machine, extreme gradient boosting decision tree model, random forest, and support vector machine.

[0087] In one embodiment, the first process data includes at least one of the following: average surface density of both sides of cathode coating, average surface density of both sides of anode coating, winding and feeding-anode roll forming pre-splitting in-process time, average anode roll forming thickness, anode roll forming speed, winding and feeding-cathode roll forming pre-splitting in-process time, average cathode roll forming thickness, and cathode roll forming speed.

[0088] The second process data includes at least one of the following: in-process time for anode die-cutting-anode roll forming pre-splitting, in-process time for cathode die-cutting-cathode roll forming pre-splitting, mean cathode roll forming thickness, mean cathode roll forming speed, mean cathode double-sided areal density, mean anode roll forming thickness, and mean anode double-sided areal density. The second process model includes at least one gradient boosting decision tree model selected from the following: lightweight gradient boosting machine, extreme gradient boosting decision tree model, random forest, and support vector machine.

[0089] It should be noted that the training method of the second matching model is the same as that of the first matching model in the aforementioned embodiments, with the only difference being the input and output data. The training method can be referred to the aforementioned embodiments, and will not be repeated here.

[0090] In this embodiment of the disclosure, the second process parameters are processed by the second matching model to predict the variable winding diameter servo position, thereby realizing a closed loop of the entire process, breaking down the information barrier between processes, achieving deep collaboration between the pre-process and winding control, improving the winding control accuracy, ensuring the stability of the ideal alignment state of the tabs, enhancing the robustness of the production process, and improving the consistency and reliability of battery products.

[0091] In this embodiment of the disclosure, by appropriately selecting the preceding process data and predicting the corresponding target process parameters through the process matching model, a closed-loop adjustment mechanism between the preceding process and the winding process is effectively constructed, thereby improving the stability and reliability of battery production.

[0092] It should be noted that the training and inference processes of the first and second matching models are similar, differing only in the input and output quantities. The training process of the process matching model will be used as an example for explanation below.

[0093] Figure 3 This diagram illustrates a flowchart of a process matching model training method provided in an embodiment of this disclosure. Figure 3 As shown, in one embodiment, before processing the preceding process data through a process matching model to obtain the target process parameters corresponding to the ideal alignment state of the tabs in the electrode material to be wound, the method further includes: S302. Obtain the historical pre-process data and corresponding process parameter labels of the target electrode, and construct a training set and a validation set based on the first preset ratio; S304. Iteratively train the process matching model to be trained based on the training set to obtain the initial process matching model; S306. Evaluate the initial process matching model based on the validation set to obtain model evaluation indicators; S308. If the model evaluation index does not meet the preset performance conditions, adjust the model parameters of the initial process matching model. S310. If the model evaluation index meets the preset performance conditions, then the initial process matching model is determined as the process matching model.

[0094] In S302, the historical preceding process data and the preceding process data have at least some of the same process parameters, and the process parameter labels are the actual values ​​corresponding to the target process parameters.

[0095] The aforementioned first preset ratio can be determined according to actual needs, for example, dividing the training set and validation set into an 8:2 or 7:3 ratio. The data in the training set is used for iterative model training, and the data in the validation set is used for model performance evaluation.

[0096] In S304, the data from the training set is sequentially input into the process matching model to be trained, and the predicted values ​​of process parameters are output. The gradient of the loss function is calculated based on the predicted values ​​and labels of the process parameters. Based on the gradient of the loss function, the process matching model to be trained is guided to gradually add weak classifiers, thereby minimizing the model error. When the iteration stopping condition is met, the initial process matching model is obtained. Meeting the iteration stopping condition may include the loss function value of the initial process matching model being less than a preset loss function threshold, or the initial process matching model having converged.

[0097] In S306~S310, the validation set is not used in model training but is used to monitor the training process and adjust parameters.

[0098] In one embodiment, model evaluation metrics include mean squared error (MSE), mean absolute error (MAE), and coefficient of determination (R²). 2 At least one of the following (Score); wherein the model evaluation index meets the preset performance conditions, including at least one of the following: mean square error is less than or equal to a first error threshold; mean absolute error is less than or equal to a second error threshold; and the coefficient of determination is greater than or equal to a preset coefficient threshold.

[0099] MSE measures the squared mean of the differences between the model's predicted values ​​(pressure roller pressure) and actual values ​​(pressure roller pressure labels), and is sensitive to outliers. MAE calculates the average of the absolute differences between the predicted and actual values ​​for all data points, providing a more intuitive measure of error. The coefficient of determination measures the model's ability to explain data variability, ranging from 0 to 1. It reveals the model's performance on training data and its generalization ability on unseen data, providing a more intuitive measure of error and assessing the model's ability to explain data variability. By comprehensively analyzing these three indicators, the model's performance can be fully evaluated, and further optimization and adjustments can be made based on the evaluation results. Ultimately, an accurate and stable first-match model can be constructed, providing support for quality control and optimization of the battery cell production line.

[0100] It should be noted that the values ​​of the first error threshold, the second error threshold, and the preset coefficient threshold can be determined according to actual needs, and this disclosure does not impose specific limitations on them.

[0101] In this embodiment, by constructing a training set and a validation set, the model is iteratively trained based on the training set, enabling the model to grasp the core relationship between the preceding process data and the target process parameters; the model is evaluated based on the validation set, and finally a process matching model is obtained, avoiding overfitting, selecting the optimal parameters, thereby improving the accuracy of the process matching model and enhancing its generalization ability.

[0102] Figure 4 This diagram illustrates another process matching model training method provided in an embodiment of the present disclosure. Figure 4 As shown, in one embodiment, before S304 iteratively trains the process matching model to be trained based on the training set to obtain the initial process matching model, the method further includes: S303. Initialize the hyperparameters of the process matching model to be trained; After determining the initial process matching model as the process matching model in S310 if the model evaluation index meets the preset performance conditions, the method further includes: S312. Based on the validation set, perform hyperparameter tuning on the hyperparameters of the initial process matching model, determine the optimal hyperparameter combination, and configure the process matching model according to the optimal hyperparameter combination.

[0103] In S303, the hyperparameters of the process matching model to be trained may include learning rate, tree depth, number of leaf nodes, regularization coefficient, etc. In one feasible implementation, during model initialization, the tree depth can be set to 40000, the maximum number of leaf nodes can be set to 50, the minimum number of leaf nodes can be set to 20, the learning rate can be set to 0.005, the L1 regularization coefficient can be set to 0.1, and the L2 regularization coefficient can be set to 0.5.

[0104] During grid search, a hyperparameter grid is constructed and an evaluation system is defined. The value range of each hyperparameter is discretized according to a preset step size to form all possible combinations of hyperparameters. The evaluation system may include evaluation indicators such as the average tab alignment deviation, the average dynamic adjustment delay, and the alignment pass rate of multiple consecutive cells. Each evaluation indicator is assigned a corresponding weight value.

[0105] In S312, the hyperparameter combinations are iterated to control the simulation effect. Offline simulation testing is used to perform simulation tests on each hyperparameter combination, and the comprehensive evaluation score for each hyperparameter combination is calculated. The comprehensive evaluation score is the weighted sum of the various evaluation indicators.

[0106] In one embodiment, hyperparameters of the initial process matching model can be tuned using hyperparameter tuning methods such as grid search, random search, and Bayesian optimization.

[0107] After screening for the optimal hyperparameter combination, the comprehensive evaluation scores are sorted from highest to lowest. The top 5 parameter combinations with the highest comprehensive evaluation scores are selected, and production tests are conducted based on these selected combinations to determine the optimal hyperparameter combination. The optimal hyperparameter combination is then fixed as the configuration parameters of the process matching model. This optimal hyperparameter combination is used to configure the process matching model for subsequent matching and prediction of target process parameters.

[0108] In this embodiment, the pre-process data is processed by a process matching model to predict the clamping roller pressure and variable winding diameter servo position corresponding to the ideal alignment of the tabs. Hyperparameters are then tuned using methods such as grid search, random search, and Bayesian optimization to determine the optimal combination of hyperparameters. This achieves a closed-loop process across the entire chain, breaks down information barriers between processes, enables deep collaboration between pre-processes and winding control, improves winding control accuracy, ensures the stability of the ideal alignment of the tabs, enhances the robustness of the production process, and improves the consistency and reliability of battery products.

[0109] Figure 5 A flowchart illustrating a feature selection method provided in an embodiment of this disclosure is shown. Figure 5 As shown, in one embodiment, the method further includes: S502. Calculate the feature contribution of each historical first process data based on the process matching model; S504. Sort the feature contribution, filter the preceding process parameters corresponding to the first historical process data whose feature contribution is greater than the preset contribution threshold, and / or filter the preceding process parameters corresponding to the first historical process data with the highest sorting value, as the input of the process matching model.

[0110] In one embodiment, when the process matching model is LightGBM or XGBoost, the feature contribution can include at least one of gain importance, split importance, and coverage importance. Gain importance represents the total information gain contributed by the preceding process parameters across all decision trees; split importance represents the total number of times the preceding process parameters are used to split the decision tree; and coverage importance represents the sum of the sample weights covered when the preceding process parameters split. It should be noted that different types of process matching models use different parameters to measure feature contribution, which can be determined based on the specific process matching model selected.

[0111] The aforementioned preset contribution threshold and preset quantity can be determined according to actual needs and pre-configured in the controller; their values ​​are not specifically limited. For example, the preset contribution threshold is the average contribution threshold, and the preset quantity accounts for 70% of the total number of preceding process parameters.

[0112] In this embodiment of the disclosure, invalid and redundant features are eliminated based on the feature importance analysis results, which simplifies the model and improves generalization ability.

[0113] It should be noted that, in addition to using feature contribution to filter the input of the process matching model, Pearson correlation coefficient or mutual information can also be used to remove strongly correlated pre-process parameters.

[0114] In one embodiment, in addition to dividing the training set and validation set, a second preset ratio is used to divide the training set, validation set, and test set, with the test set simulating the prediction scenario of the target process parameters. The second preset ratio can be set to 8:1:1 or 7:2:1, thereby balancing the needs of model training, parameter tuning, and evaluation while ensuring data distribution consistency.

[0115] The following section uses LightGBM as an example to introduce the model training and deployment of the first matching model. The model building phase, using LightGBM for predictive modeling, mainly includes dataset preparation, dataset partitioning, model initialization, model training, model evaluation, model optimization, and model deployment.

[0116] In the dataset preparation stage, based on the range of values ​​for the pressure roller pressure, the pressure roller pressure corresponding to the ideal state of the electrode tab and the first process data are selected for the first matching model. The pressure roller pressure is used as the label of the first matching model, i.e., the output of the first matching model. The historical first process data is used as the input of the first matching model. The dataset is constructed, and the first matching model is trained based on the dataset.

[0117] When partitioning the dataset, a stratified sampling method is used to divide it into a training set, a validation set, and a test set, with a partition ratio of 8:1:1 or 7:2:1. This ensures the consistency of data distribution and balances the needs of model training, parameter tuning, and model evaluation. It should be noted that this disclosure does not specify a particular ratio for the partitioning of the training set, validation set, and test set.

[0118] During model initialization, the LightGBM library is imported, providing a wealth of adjustable model parameters. LightGBM model parameters are configured as follows: a gradient boosting decision tree algorithm is used, with the objective function set to regression to predict continuous values. The maximum number of leaf nodes in the decision tree is set to 50, and the minimum number of data samples per leaf node is 20 to prevent overfitting. The number of decision trees used in the model (i.e., the number of iterations) is set to 40,000; this parameter, combined with early stopping, prevents overfitting. The learning rate is set to 0.005, indicating that each tree contributes less to the final result, requiring more iterations to achieve the same training effect. In each iteration, 80% of the features and 80% of the samples are randomly selected, with random sample selection every 10 iterations to increase the model's generalization ability. The model's output log level is set to -1, disabling log output. Finally, GPU acceleration is specified, significantly improving training speed. This configuration tends to use a larger number of leaf nodes and deeper trees, while controlling model complexity and overfitting risk through a smaller learning rate, features, and samples. It should be noted that during the model initialization phase, the hyperparameters of the model can also be set according to actual needs, and this disclosure does not impose any restrictions on this.

[0119] During the model training phase, the `lgb.Dataset` function can be used to create dedicated data structures for LightGMB, creating separate structures for training and validation data. The target variable, such as the pressure roller pressure, can be specified using the `label` parameter. The `lgb.train` function is then called to begin model training, passing in a predefined dictionary of model parameters, training data, validation data, and the specified number of boost iterations, `num_boost_round`. The `valid_sets` parameter is set to the validation set, and the early stopping mechanism `early_stopping_round` is enabled. This mechanism stops training if the evaluation metrics on the validation set do not improve within a specified number of iterations, effectively preventing overfitting. Through these operations, an initial LightGMB model predicting the pressure roller pressure based on the first process data can be trained.

[0120] During the model evaluation phase, the performance of the initial LightGMB model needs to be evaluated using test set data to ensure that the trained LightGMB model can accurately predict the quality indicators of the battery cell production line. To comprehensively evaluate the model's predictive ability, three evaluation metrics are used: mean squared error, mean absolute error, and coefficient of determination.

[0121] During the model optimization phase, hyperparameter tuning and early stopping can be performed based on the validation set data to evaluate the model's performance on new data. Based on the feature importance analysis results, features with lower importance are removed to simplify the model and improve generalization ability.

[0122] The first matching model was tested on the test set to evaluate its generalization ability in an unbiased manner, verify its engineering practicality, and expose its defects.

[0123] During the model deployment phase, the first matched model that has been trained is saved in the controller for subsequent model loading and use, while also providing prediction functionality.

[0124] It should be noted that the model construction and training of the second matching model are similar to those of the first matching model, with the only difference being the model input and output. The training process of the second matching model can be referred to the training process of the first matching model, and will not be repeated here.

[0125] In one embodiment, the preceding process data includes the second process data; the electrode misalignment prediction model is trained using the target process parameters, the second process data, and the corresponding electrode misalignment amount labels.

[0126] The second process data includes at least one of the following: anode die-cutting-anode roll forming pre-slitting in-process time, cathode die-cutting-cathode roll forming pre-slitting in-process time, average cathode roll forming thickness, cathode roll forming speed, average cathode double-sided surface density, average anode roll forming thickness, and average anode double-sided surface density.

[0127] In one embodiment, the electrode misalignment prediction model includes at least one gradient boosting decision tree model selected from lightweight gradient boosting machine, extreme gradient boosting decision tree model, random forest, and support vector machine.

[0128] It should be noted that the training method of the tab misalignment prediction model is the same as that of the process matching model, with the only difference being the input and output quantities. The similarities will not be elaborated further.

[0129] In this embodiment, the target process parameters and pre-process data are intelligently analyzed by the tab misalignment prediction model to predict the misalignment value, which greatly improves the accuracy and timeliness of misalignment risk prediction. When the misalignment prediction value meets expectations, the recommended instructions generated based on the target process parameters can directly empower the winding equipment. This not only realizes the dynamic optimization of the winding process parameters, but also builds a closed-loop adjustment mechanism between the pre-process and the winding process, which significantly improves the stability and reliability of battery product quality.

[0130] In one embodiment, the preset conditions include a preset misalignment threshold; wherein, if the misalignment prediction value meets the preset conditions in step S208, a recommendation instruction based on the target process parameters is generated, including: If the misalignment prediction value is less than or equal to the preset misalignment threshold, the misalignment prediction value is determined to meet the preset condition; if the misalignment prediction value is greater than the preset misalignment threshold, the misalignment prediction value is determined to not meet the preset condition.

[0131] The above misalignment prediction value is used to measure the degree of misalignment between the positive and negative tabs in the electrode material to be wound.

[0132] In one embodiment, the misalignment prediction value includes at least one of the following: electrode misalignment amount, electrode misalignment probability, and electrode misalignment level.

[0133] The misalignment of the tabs can include lateral misalignment, longitudinal misalignment, height misalignment, and angular misalignment. Lateral misalignment refers to the positional deviation of the positive and negative tabs along the width of the electrode, that is, the deviation of the center line of the tab from the center line of the pre-set tab window of the separator in the width direction. Longitudinal misalignment refers to the positional deviation of the positive and negative tabs along the length of the electrode, or the spacing deviation between multiple tabs on the same electrode. Height deviation is the deviation of the positive and negative tabs along the thickness of the electrode, that is, the adhesion deviation between the tab and the electrode body and the separator in the thickness direction. Angular misalignment refers to the skewing or twisting deviation of the tabs caused by the previous process or winding tension.

[0134] The tab misalignment probability refers to the percentage of battery cells produced where the actual position of the tab deviates from the ideal alignment within a certain production batch or time range. For example, in a batch of 1000 battery cells, if 5 cells have a lateral misalignment exceeding 0.05mm, 3 cells have a longitudinal misalignment exceeding 0.1mm, and 2 cells have both lateral and longitudinal misalignments, then the overall tab misalignment probability is (5+3-2) / 1000×100%=0.6%.

[0135] The tab misalignment level refers to a tiered standard for classifying tab alignment based on the specific numerical value of the tab misalignment, the probability of misalignment, and its impact on battery performance. For example, a cell with a lateral misalignment ≤ 0.05mm is defined as level 0, a cell with a lateral misalignment between 0.05mm and 0.08mm is defined as level 1, and a cell with a lateral misalignment between 0.08mm and 0.12mm is defined as level 3. The higher the tab misalignment level, the greater the tab misalignment.

[0136] In one embodiment, when the misalignment prediction value includes the tab misalignment amount, the preset misalignment threshold may include a longitudinal misalignment threshold, a lateral misalignment threshold, a height misalignment threshold, an angle misalignment threshold, etc. The value of the preset misalignment threshold can be determined according to actual needs. For example, the longitudinal misalignment threshold and the lateral misalignment threshold can be 0.05mm, the height misalignment threshold is 0.02mm, and the angle misalignment threshold is 0.1mm / m.

[0137] In one embodiment, when the misalignment prediction value includes the electrode misalignment probability, the preset misalignment threshold may include a single-dimensional preset misalignment probability threshold or a comprehensive preset misalignment probability threshold. The value of the preset misalignment probability threshold can be determined according to actual needs. For example, the preset misalignment probability threshold for the horizontal dimension is 1%, the preset misalignment probability threshold for the vertical dimension is 1.5%, and the comprehensive preset misalignment probability threshold is 2%.

[0138] When the predicted misalignment value includes the electrode misalignment level, the predicted misalignment threshold includes a preset misalignment level threshold. For example, the preset misalignment level threshold can be level 1.

[0139] It should be noted that the aforementioned preset misalignment threshold can be a fixed value or a range of values, and this disclosure does not impose any specific limitations on it.

[0140] In this embodiment of the disclosure, by comparing the relationship between the misalignment prediction value and the preset misalignment threshold, it is determined whether the misalignment prediction value meets the preset conditions, thereby identifying misalignment risks in advance, triggering the adjustment of winding process parameters in advance, nipping misalignment risks in the bud in the preceding process, reducing the amount of tab misalignment, accurately matching the misalignment level, ensuring battery quality consistency, avoiding cost waste caused by excessive pursuit of ideal alignment, and preventing quality loss due to insufficient control.

[0141] Figure 6 This diagram illustrates a flowchart of another electrode material misalignment adjustment method provided by an embodiment of the present disclosure. In one embodiment, such as Figure 6 As shown, the method also includes: S210. If the misalignment prediction value does not meet the preset conditions, the target process parameters are not recommended.

[0142] In one embodiment, when the tab misalignment amount is greater than a preset misalignment amount threshold, the tab misalignment probability is greater than a preset probability threshold, or the tab misalignment level is greater than a preset level threshold, the misalignment prediction value is determined to not meet the preset conditions. That is, the system determines that the target process parameters may lead to an excessively high risk of tab misalignment, and does not generate operation instructions to recommend the target process parameters, nor does it recommend pop-ups to the equipment control terminal or perform any operation based on the target process parameters.

[0143] In this disclosure, to more accurately control tab misalignment, a collaborative decision-making mechanism based on a process matching model and a tab misalignment prediction model is employed. First, the process matching model comprehensively analyzes the preceding process data to generate an optimal hyperparameter combination for the two key parameters directly affecting tab positioning: clamping roller pressure and variable roll diameter servo position. However, relying solely on these two parameters is insufficient, as tab misalignment is also influenced by multiple factors such as material state and process time in upstream processes. Therefore, this optimal hyperparameter combination is not directly applied but integrated with second-stage process data. The second-stage process data covers everything from the winding and feeding stage, including the WIP (work-in-process) time of the anode and cathode during the pre-slitting stage of the roll forming, as well as key material characteristics such as the average roll forming thickness, roll forming speed, and average surface density of the coated sides for both the cathode and anode. By deeply integrating the optimal hyperparameter combination with this rich second-stage process data from preceding processes, a comprehensive dataset is formed.

[0144] Next, this integrated dataset is uniformly input into a pre-trained tab misalignment prediction model. This model can quantitatively predict the amount of tab misalignment in the current and short-term future based on historical second-stage process data, optimal hyperparameter combinations, and algorithmic relationships. The prediction results are compared with preset tab misalignment upper and lower limits (also known as tab misalignment thresholds): if the predicted tab misalignment value exceeds the acceptable range, it indicates that the target process parameter combination may not guarantee quality, and the system will abandon recommending this set of P-roller pressure and variable roll diameter servo positions to the equipment control end. Conversely, if the predicted tab misalignment value is within the acceptable range, it indicates that this parameter combination has a high probability of controlling the tab misalignment within the ideal range, and the system will formally recommend this optimized target process parameter set to the equipment control end. At this time, the equipment will issue a pop-up reminder to the operator or automation system to confirm the parameter change suggestion, thereby achieving prediction-based, more proactive, and refined process control.

[0145] Based on the same inventive concept, this disclosure also provides an electrode material misalignment adjustment device, as shown in the following embodiment. Since the principle by which this device solves the problem is similar to that of the method embodiment described above, the implementation of this device embodiment can refer to the implementation of the method embodiment described above, and repeated details will not be elaborated further.

[0146] Figure 7 This diagram illustrates an electrode material misalignment adjustment device according to an embodiment of the present disclosure. Figure 7 As shown, in one embodiment, the device includes a data acquisition module 710, a parameter determination module 720, a misalignment detection module 730, and a parameter recommendation module 740.

[0147] Among them, the data acquisition module 710 is used to acquire the pre-process data of the electrode material to be wound. The pre-process data is the process parameter value of the electrode material to be wound in the process before the winding process. The parameter determination module 720 is used to process the preceding process data through the process matching model to obtain the target process parameters corresponding to the ideal alignment state of the electrode material to be wound. The misalignment detection module 730 is used to process the target process parameters and the preceding process data through the tab misalignment prediction model to obtain the predicted value of the tab misalignment in the electrode material to be wound. The parameter recommendation module 740 is used to generate a recommendation instruction based on the target process parameters if the misalignment prediction value meets the preset conditions.

[0148] It should be noted that the pre-process data includes the first process data, and the process matching model includes the first matching model, which is trained using the historical first process data of the target electrode and the corresponding pressure label of the pressure roller; when the target process parameter includes the variable roll diameter servo position, the pre-process data includes the second process data, and the process matching model includes the second matching model, which is trained using the historical second process data of the target electrode and the corresponding variable roll diameter servo position label.

[0149] In one embodiment, the device further includes a first model training module (not shown in the figures). The first model training module is used to: acquire historical first process data and corresponding process parameter labels of the target electrode tabs before processing the preceding process data through the process matching model to obtain the target process parameters corresponding to the ideal alignment state of the tabs in the electrode material to be wound; construct a training set and a validation set based on a first preset ratio; iteratively train the process matching model to be trained based on the training set to obtain an initial process matching model; evaluate the initial process matching model based on the validation set to obtain a model evaluation index; if the model evaluation index does not meet preset performance conditions, adjust the model parameters of the initial process matching model; if the model evaluation index meets the preset performance conditions, determine the initial process matching model as the process matching model.

[0150] In one embodiment, the first model training module is used to initialize the hyperparameters of the process matching model to be trained before iteratively training the process matching model to be trained based on the training set to obtain the initial process matching model; after determining the initial process matching model as the process matching model if the model evaluation index meets the preset performance conditions, the module performs hyperparameter tuning on the hyperparameters of the initial process matching model based on the validation set to determine the optimal hyperparameter combination, and configures the process matching model according to the optimal hyperparameter combination.

[0151] In one embodiment, the first model training module is used to calculate the feature contribution of each historical first process data based on the process matching model; sort the feature contribution, filter the preceding process parameters corresponding to the historical first process data whose feature contribution is greater than a preset contribution threshold, and / or filter the preceding process parameters corresponding to a preset number of historical first process data that are ranked first, as input to the process matching model.

[0152] It should be noted that the model evaluation index includes at least one of the following: mean squared error, mean absolute error, and coefficient of determination; wherein, the model evaluation index meets the preset performance conditions, including at least one of the following: mean squared error is less than or equal to a first error threshold; mean absolute error is less than or equal to a second error threshold; and coefficient of determination is greater than or equal to a preset coefficient threshold.

[0153] It should be noted that the preceding process data includes the second process data; the electrode misalignment prediction model is trained using the target process parameters, the second process data, and the corresponding electrode misalignment amount labels.

[0154] In one embodiment, the process matching model and the tab misalignment prediction model include at least one gradient boosting decision tree model selected from lightweight gradient boosting machine, extreme gradient boosting decision tree model, random forest, and support vector machine.

[0155] It should be noted that the first process data includes at least one of the following: the average surface density of the cathode coating on both sides, the average surface density of the anode coating on both sides, the in-process time of winding and feeding to anode roll forming and pre-cutting, the average thickness of anode roll forming, the anode roll forming speed, the in-process time of winding and feeding to cathode roll forming and pre-cutting, the average thickness of cathode roll forming, and the cathode roll forming speed.

[0156] In one embodiment, the second process data includes at least one of: anode die-cutting-anode roll forming pre-slitting in-process time, cathode die-cutting-cathode roll forming pre-slitting in-process time, average cathode roll forming thickness, cathode roll forming speed, average cathode double-sided surface density, average anode roll forming thickness, and average anode double-sided surface density.

[0157] In one embodiment, the preset conditions include a preset misalignment threshold; the parameter recommendation module 740 is used to determine that the misalignment prediction value meets the preset conditions if the misalignment prediction value is less than or equal to the preset misalignment threshold; and to determine that the misalignment prediction value does not meet the preset conditions if the misalignment prediction value is greater than the preset misalignment threshold.

[0158] In one embodiment, the misalignment prediction value includes at least one of the following: electrode misalignment amount, electrode misalignment probability, or electrode misalignment level.

[0159] In one embodiment, the parameter recommendation module 740 is further configured to not recommend target process parameters if the misalignment prediction value does not meet preset conditions.

[0160] In this embodiment, by deeply exploring the relationship between the pre-process data of the electrode material to be wound and the process parameters of the winding process, a bridge is built between the pre-process and the winding process, realizing the full-link connection of process data. The target process parameters predicted by the process matching model provide strong support for subsequent tab misalignment adjustment. The tab misalignment prediction model performs intelligent analysis on the target process parameters and pre-process data to predict the tab misalignment prediction value, which greatly improves the accuracy and timeliness of misalignment risk prediction. When the misalignment prediction value meets expectations, the recommended instructions generated based on the target process parameters can directly empower the winding equipment. This not only realizes the dynamic optimization of the winding process parameters, but also builds a closed-loop adjustment mechanism between the pre-process and the winding process, significantly improving the stability and reliability of battery product quality.

[0161] Those skilled in the art will understand that various aspects of this disclosure can be implemented as a system, method, or program product. Therefore, various aspects of this disclosure can be specifically implemented in the following forms: a completely hardware implementation, a completely software implementation (including firmware, microcode, etc.), or a combination of hardware and software aspects, collectively referred to herein as a "circuit," "module," or "system."

[0162] The following reference Figure 8 To describe an electronic device 800 according to such an embodiment of the present disclosure. Figure 8 The electronic device 800 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments disclosed herein.

[0163] In one embodiment, the electronic device 800 includes a processor and a memory for storing executable instructions of the processor, wherein the processor is configured to perform the above-described electrode material misalignment adjustment method by executing the executable instructions.

[0164] like Figure 8 As shown, the electronic device 800 is manifested in the form of a general-purpose computing device. The components of the electronic device 800 may include, but are not limited to: at least one processing unit 810, at least one storage unit 820, and a bus 830 connecting different system components (including storage unit 820 and processing unit 810).

[0165] The storage unit stores program code, which can be executed by the processing unit 810, causing the processing unit 810 to perform the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of this disclosure. For example, the processing unit 810 can perform the following steps of the above method embodiment: obtaining the pre-process data of the electrode material to be wound, wherein the pre-process data is the process parameter value of the electrode material to be wound in the process before the winding process; processing the pre-process data through a process matching model to obtain the target process parameters corresponding to the ideal alignment state of the electrode material to be wound; processing the target process parameters and the pre-process data through a tab misalignment prediction model to obtain the tab misalignment prediction value in the electrode material to be wound; if the misalignment prediction value meets the preset conditions, generating a recommended instruction based on the target process parameters.

[0166] Storage unit 820 may include a readable medium in the form of a volatile storage unit, such as random access memory (RAM) 8201 and / or cache memory 8202, and may further include a read-only memory (ROM) 8203.

[0167] The storage unit 820 may also include a program / utility 8204 having a set (at least one) of program modules 8205, such program modules 8205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each or some combination of these examples may include an implementation of a network environment.

[0168] Bus 830 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the various bus structures.

[0169] Electronic device 800 can also communicate with one or more external devices 840 (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable a user to interact with the electronic device 800, and / or with any device that enables the electronic device 800 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 850. Furthermore, electronic device 800 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 860. Figure 8 As shown, network adapter 860 communicates with other modules of electronic device 800 via bus 830. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with electronic device 800, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0170] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, terminal device, or network device, etc.) to execute the methods according to the embodiments of this disclosure.

[0171] In particular, according to embodiments of the present disclosure, the processes described above with reference to the flowcharts can be implemented as a computer program product, which includes a computer program that, when executed by a processor, implements the steps described in the "Exemplary Methods" section above according to various exemplary implementations of the present disclosure.

[0172] In exemplary embodiments of this disclosure, a computer-readable storage medium is also provided, having a computer program stored thereon that, when executed by a processor, implements the steps described in the "Exemplary Methods" section above according to various exemplary embodiments of this disclosure. This computer-readable storage medium may be a readable signal medium or a readable storage medium.

[0173] In some possible implementations, various aspects of this disclosure can also be implemented as a program product comprising a computer program that, when executed by a processor, implements the electrode material misalignment adjustment method of any of the above embodiments. In one possible embodiment, the program product includes program code that, when run on a terminal device, causes the terminal device to perform the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of this disclosure.

[0174] More specific examples of computer-readable storage media in this disclosure may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0175] In this disclosure, a computer-readable storage medium may include a data signal propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of transmitting, propagating, or transmitting a program for use by or in connection with an instruction execution system, apparatus, or device.

[0176] Optionally, the program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0177] In practice, program code for performing the operations of this disclosure can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0178] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0179] Furthermore, although the steps of the method in this disclosure are described in a specific order in the accompanying drawings, this does not require or imply that the steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps may be omitted, multiple steps may be combined into one step, and / or a step may be broken down into multiple steps.

[0180] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, mobile terminal, or network device, etc.) to execute the methods according to the embodiments of this disclosure.

[0181] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the appended claims.

Claims

1. A method for adjusting the misalignment of electrode materials, characterized in that, include: Obtain the pre-process data of the electrode material to be wound, wherein the pre-process data is the process parameter value of the electrode material to be wound in the process before the winding process; The preceding process data is processed by a process matching model to obtain the target process parameters corresponding to the ideal alignment state of the electrode material to be wound. The target process parameters and the preceding process data are processed by the electrode misalignment prediction model to obtain the predicted value of the electrode misalignment in the electrode material to be wound. If the misalignment prediction value meets the preset conditions, a recommended instruction based on the target process parameters is generated.

2. The method according to claim 1, characterized in that, The target process parameters include pressure roller pressure and / or variable roll diameter servo position; When the target process parameter includes the pressure of the clamping roller, the preceding process data includes first process data, and the process matching model includes a first matching model. The first matching model is trained by the historical first process data of the target electrode and the corresponding pressure of the clamping roller. When the target process parameters include variable roll diameter servo position, the preceding process data includes second process data, and the process matching model includes a second matching model, which is trained using the historical second process data of the target electrode and the corresponding variable roll diameter servo position label.

3. The method according to claim 1, characterized in that, Before processing the pre-process data using a process matching model to obtain the target process parameters corresponding to the ideal alignment state of the tabs in the electrode material to be wound, the method further includes: Obtain the historical first process data and corresponding process parameter labels of the target electrode, and construct a training set and a validation set based on a first preset ratio; Based on the training set, the process matching model to be trained is iteratively trained to obtain the initial process matching model; The initial process matching model is evaluated based on the validation set to obtain model evaluation indicators. If the model evaluation index does not meet the preset performance conditions, then the model parameters of the initial process matching model are adjusted. If the model evaluation index meets the preset performance conditions, then the initial process matching model is determined as the process matching model.

4. The method according to claim 3, characterized in that, Before iteratively training the process matching model to be trained based on the training set to obtain the initial process matching model, the method further includes: Initialize the hyperparameters of the process matching model to be trained; After determining the initial process matching model as the process matching model if the model evaluation index meets the preset performance conditions, the method further includes: Based on the validation set, the hyperparameters of the initial process matching model are tuned to determine the optimal combination of hyperparameters, and the process matching model is configured according to the optimal combination of hyperparameters.

5. The method according to claim 3, characterized in that, The method further includes: The feature contribution of each historical first process data is calculated based on the process matching model. The feature contribution is sorted, and the preceding process parameters corresponding to the first historical process data with a feature contribution greater than a preset contribution threshold are selected, and / or the preceding process parameters corresponding to a preset number of the first historical process data are selected as the input of the process matching model.

6. The method according to claim 3, characterized in that, The model evaluation index includes at least one of mean squared error, mean absolute error, and coefficient of determination. Wherein, the model evaluation index satisfies the preset performance conditions, including at least one of the following: The mean square error is less than or equal to the first error threshold; The mean absolute error is less than or equal to the second error threshold; The determination coefficient is greater than or equal to a preset coefficient threshold.

7. The method according to claim 1, characterized in that, The preceding process data includes the second process data; the electrode misalignment prediction model is trained using the target process parameters, the second process data, and the corresponding electrode misalignment amount labels.

8. The method according to claim 1, characterized in that, The process matching model and the tab misalignment prediction model include at least one gradient boosting decision tree model selected from lightweight gradient boosting machine, extreme gradient boosting decision tree model, random forest, and support vector machine.

9. The method according to claim 2, characterized in that, The first process data includes at least one of the following: average surface density of both sides of cathode coating, average surface density of both sides of anode coating, winding and feeding-anode roll forming pre-splitting in-process time, average anode roll forming thickness, anode roll forming speed, winding and feeding-cathode roll forming pre-splitting in-process time, average cathode roll forming thickness, and cathode roll forming speed.

10. The method according to claim 2 or 8, characterized in that, The second process data includes at least one of the following: anode die-cutting-anode roll forming pre-slitting in-process time, cathode die-cutting-cathode roll forming pre-slitting in-process time, average cathode roll forming thickness, cathode roll forming speed, average cathode double-sided surface density, average anode roll forming thickness, and average anode double-sided surface density.

11. The method according to claim 1, characterized in that, The preset conditions include a preset misalignment threshold; Wherein, if the misalignment prediction value meets the preset conditions, a recommendation instruction based on the target process parameters is generated, including: If the misalignment prediction value is less than or equal to the preset misalignment threshold, then the misalignment prediction value is determined to meet the preset condition. If the predicted misalignment value is greater than the preset misalignment threshold, then the predicted misalignment value is determined not to meet the preset condition.

12. The method according to claim 11, characterized in that, The misalignment prediction value includes at least one of the following: electrode misalignment amount, electrode misalignment probability, or electrode misalignment level.

13. The method according to claim 11, characterized in that, The method further includes: If the misalignment prediction value does not meet the preset conditions, the target process parameters are not recommended.

14. An electrode material misalignment adjustment device, characterized in that, include: The data acquisition module is used to acquire the pre-process data of the electrode material to be wound, wherein the pre-process data is the process parameter value of the electrode material to be wound in the process before the winding process. The parameter determination module is used to process the preceding process data through a process matching model to obtain the target process parameters corresponding to the ideal alignment state of the electrode material to be wound. The misalignment detection module is used to process the target process parameters and the preceding process data through the tab misalignment prediction model to obtain the predicted misalignment value of the tab in the electrode material to be wound. The parameter recommendation module is used to generate a recommendation instruction based on the target process parameters if the misalignment prediction value meets the preset conditions.

15. An electronic device, characterized in that, include: processor; Memory for storing the executable instructions of the processor; The processor is configured to execute the electrode material misalignment adjustment method according to any one of claims 1 to 13 by executing the executable instructions.

16. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the electrode material misalignment adjustment method according to any one of claims 1 to 13.

Images