A dry electrode delamination defect suppression method based on in-situ monitoring

Abstract

The application provides a dry-process electrode delamination defect inhibition method based on in-situ monitoring, comprising the following steps: S1, collecting the calendering pressure data and the powder fluidity data in real time from the dry-process electrode preparation process, classifying the interaction mode of the calendering pressure level and the powder fluidity index, and determining the potential delamination defect risk area; S2, simulating the material bonding force distribution through a neural network model, judging the interlayer stress difference value, and if the interlayer stress difference value exceeds a preset stress difference threshold, adjusting the calendering pressure parameter to the powder fluidity matching range to obtain an optimized process parameter set; S3, running a structure stability test in a simulation environment to obtain the delamination defect occurrence probability of the electrode material; S4, extracting the performance consistency index from the delamination defect occurrence probability, comparing and matching with historical data through iteration to determine the final defect control scheme; and S5, updating the real-time monitoring module of the preparation process by using the final defect control scheme to obtain a stable output result for battery industry application.

Description

Technical Field

[0001] This invention relates to the field of dry electrode technology, and in particular to a method for suppressing delamination defects in dry electrodes based on in-situ monitoring. Background Technology

[0002] In the field of new energy materials, dry electrode fabrication technology has attracted much attention due to its environmental friendliness and high efficiency, and is considered one of the key technologies for promoting the green development of the battery industry. Especially, fabricating electrode materials at room temperature not only reduces energy consumption but also simplifies the process, which is of great significance for achieving large-scale production. However, how to effectively suppress delamination defects and ensure the structural stability and performance consistency of electrode materials during this process has become a key issue that the industry urgently needs to overcome.

[0003] Currently, although various methods have been attempted to address the delamination defect problem in dry electrode fabrication, most solutions often struggle to adapt to the dynamic changes of multiple variables in complex process environments. Particularly at room temperature, the interactions between process parameters are often overlooked, leading to poor defect control. Existing methods focus more on optimizing individual steps, lacking in-depth exploration of the synergistic effects of different factors throughout the fabrication process. This results in frequent defect occurrences under certain specific conditions, impacting the overall quality of the electrode material.

[0004] Focusing on the technical challenges, a core difficulty in dry electrode fabrication lies in the complexity and uncertainty of process parameters. The calendering pressure, in particular, is a crucial factor, directly affecting the bonding force between material particles. Improper pressure settings can lead to weak interlayer bonding, inducing delamination. Furthermore, the choice of calendering pressure is closely related to powder flowability. Powders with poor flowability are prone to uneven distribution under pressure, further exacerbating interlayer stress differences and ultimately making delamination defects difficult to avoid. These two factors are intertwined, jointly constituting a challenge for process control. For example, in actual production, if the calendering pressure is too high while the powder flowability is insufficient, localized accumulation of material may occur during compaction, leading to uneven interlayer bonding and ultimately causing peeling or cracking during electrode use.

[0005] Therefore, accurately identifying and controlling the interactions between these key process parameters at room temperature to avoid delamination defects caused by improper parameter settings has become a critical issue that dry electrode fabrication technology urgently needs to address. Solving this problem not only affects the performance stability of electrode materials but also directly impacts the reliability and economy of the entire battery manufacturing process. Summary of the Invention

[0006] To address the technical problems mentioned in the background section, this invention provides a method for suppressing delamination defects in dry electrodes based on in-situ monitoring, the method comprising:

[0007] S1. Real-time acquisition of calendering pressure data and powder flowability data during the dry electrode fabrication process; classification of the interaction patterns between calendering pressure levels and powder flowability indicators; identification of potential delamination defect risk areas. S2. Simulation of material bonding force distribution using a neural network model; determination of interlayer stress difference values; adjustment of calendering pressure parameters to match powder flowability range if the interlayer stress difference value exceeds a preset stress difference threshold; obtaining an optimized set of process parameters. S3. Structural stability testing in a simulated environment to obtain the probability of delamination defects in the electrode material. S4. Extraction of performance consistency indicators from the probability of delamination defects; determination of the final defect control scheme through iterative comparison and matching with historical data. S5. Update of the real-time monitoring module of the fabrication process using the final defect control scheme; obtaining stable output results for battery industry applications.

[0008] Furthermore, step S1 includes:

[0009] Step S11: Real-time acquisition of pressure data and powder flowability data during the dry electrode calendering process using a sensor array to generate an initial set of process parameters;

[0010] Step S12: Based on the initial set of process parameters, the support vector machine algorithm is used for classification to obtain the interaction pattern between the calendering pressure level and the powder flowability index, and to determine the potential risk area of ​​delamination defects.

[0011] Step S13: Extract stress data sequences and liquidity data sequences for the corresponding time periods for potential stratified defect risk areas, and calculate the standard deviation of stress data and the coefficient of variation of liquidity data.

[0012] Step S14: If the standard deviation of the pressure data is higher than the preset pressure difference threshold and the coefficient of variation of the liquidity data is higher than the preset coefficient of variation threshold, then mark the current potential stratification defect risk area as a high-risk stratification defect area.

[0013] Step S15: Based on the set of process parameters corresponding to the high-risk stratified defect areas, cluster them into multiple defect pattern subsets, and use the support vector machine algorithm to retrain the defect pattern subsets, update the interaction pattern classification boundary, and obtain the optimized risk area division.

[0014] Furthermore, step S14 includes setting the overpressure difference threshold to 15 N / cm and the coefficient of variation threshold to 0.12.

[0015] Furthermore, step S2 includes:

[0016] Step S21: Obtain layered defect data inside the material through scanning technology, mark potential risk areas, and obtain preliminary defect distribution information;

[0017] Step S22: Based on the defect distribution information, a neural network model is used to simulate the material bonding force distribution, calculate the stress values ​​between each layer, and determine the distribution state of interlayer stress.

[0018] Step S23: Calculate the stress difference between adjacent layers based on the distribution of interlayer stress. If the stress difference exceeds the preset stress difference threshold, record the stress anomaly points in the corresponding area to obtain a set of anomaly areas.

[0019] Step S24: Extract key location data from the abnormal area set, analyze the influence of calendering pressure on powder flowability, determine whether parameters need to be adjusted, and obtain the pressure adjustment requirements.

[0020] Step S25: Based on the pressure adjustment requirements, calculate the matching range between powder flowability and calendering pressure, adjust the relevant parameters to the target range, and determine the optimized pressure parameter values.

[0021] Step S26: Obtain the optimized pressure parameter values, perform an overall verification in conjunction with the process parameter set, and determine whether the production conditions are met. If not, iteratively adjust until the requirements are met to obtain the final process parameter set.

[0022] Step S27: Based on the final set of process parameters, generate corresponding control commands, transmit them to the production equipment for parameter updates, and complete the process optimization process.

[0023] Furthermore, step S3 includes:

[0024] Step S31: Based on the optimized process parameter set, perform structural stability testing in the simulated environment by loading the parameter set, and obtain stress distribution data of the electrode material in the simulated environment;

[0025] Step S32: Use the finite element analysis method to determine the stress concentration region inside the electrode material;

[0026] Step S33: If the stress value in the stress concentration area exceeds the preset stress threshold, mark the corresponding parameter set as a high-risk parameter set and obtain the potential location of the layered defect.

[0027] Step S34: Run multiple random perturbation tests on the high-risk parameter set in a simulation environment to obtain the probability of occurrence of layered defects, determine the correspondence between the probability values ​​of occurrence of layered defects and the high-risk parameter set, and determine the probability distribution of defect occurrence.

[0028] Furthermore, step S4 includes:

[0029] Step S41: Obtain the original dataset for probability analysis from the data records of layered defects, classify and organize the defect distribution for each layer, and obtain a preliminary defect probability set.

[0030] Step S42: Based on the preliminary defect probability set, use a consistent performance evaluation standard to compare the defect probability of each level with the preset defect probability threshold range. If the defect probability of a certain level exceeds the defect probability threshold, it is marked as an abnormal level, and the distribution list of abnormal levels is determined.

[0031] Step S43: For the distribution list of the anomaly level, obtain the relevant data groups in the historical records, and match the data of the current anomaly level with the historical records one by one through iterative comparison to obtain the reference dataset with the highest matching degree.

[0032] Step S44: Based on the reference dataset with the highest matching degree, extract the strategy content related to defect control. If the strategy in the reference dataset is applicable to the current anomaly level, directly use the strategy to determine the initial control strategy combination.

[0033] Step S45: For the initial control strategy combination, the logical consistency between strategies is detected by data correlation analysis tools. If a strategy conflicts with other strategies, the strategy is removed to obtain the optimized control strategy set.

[0034] Step S46: Based on the optimized control strategy set and the processing flow of the scheme optimization, apply the strategy set to the current layered defect scenario, and determine the applicability of the final scheme by simulating execution to determine the final scheme for defect control.

[0035] Step S47: For the output of the final solution, the support vector machine algorithm is used to predict and analyze the execution effect of the solution. By comparing the prediction results, the adaptability distribution of the solution in different levels of defects is obtained, and the comprehensive coverage capability of the solution is judged.

[0036] Furthermore, step S5 includes:

[0037] Step S51: Obtain sensor data and equipment operating parameters collected in real time during the production process;

[0038] Step S52: Determine whether a defect signal exists from real-time data using an anomaly detection algorithm;

[0039] Step S53: If a defect signal exists, determine the defect type and severity.

[0040] Step S54: Obtain the control parameter adjustment rules for the corresponding defect type according to the final defect control scheme;

[0041] Step S55: Update the threshold and feedback coefficient of the real-time monitoring module using the control parameter adjustment rules;

[0042] Step S56: Process subsequent production process data through the updated real-time monitoring module to obtain intermediate output after defect suppression;

[0043] Step S57: Determine the degree of deviation between the intermediate output and the preset stable output standard;

[0044] In step S58, if the deviation exceeds the range, a supplementary correction instruction is obtained from the final defect control scheme to obtain the corrected monitoring module parameters.

[0045] The technical solution provided by this invention has the following beneficial effects:

[0046] This invention discloses a method for suppressing delamination defects in dry-process electrodes based on in-situ monitoring. By real-time acquisition of rolling pressure and powder flowability data, an initial set of process parameters is constructed. Combined with a support vector machine (SVM) algorithm for classification and interaction, potential delamination defect risk areas are accurately identified. Furthermore, this invention uses a neural network model to simulate the material bonding force distribution and calculate the interlayer stress difference. When the difference exceeds the standard, the rolling pressure parameters are dynamically adjusted to match the powder flowability, thus optimizing the process parameters. Subsequently, the invention tests the structural stability in a simulated environment, extracts performance consistency indicators, and determines the final defect control scheme through iterative comparison with historical data, updating the real-time monitoring module of the preparation process. The core innovation of this invention lies in the integration of multi-source data analysis and intelligent algorithms to achieve closed-loop control from defect risk prediction to process optimization, ultimately significantly reducing the probability of delamination defects in electrode materials, ensuring stable output for battery industry applications, and demonstrating efficient and intelligent process improvement capabilities. Attached Figure Description

[0047] Figure 1 This is a flowchart of a dry electrode delamination defect suppression method based on in-situ monitoring according to the present invention.

[0048] Figure 2 This is a schematic diagram of a dry electrode delamination defect suppression method based on in-situ monitoring according to the present invention. Detailed Implementation

[0049] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] like Figures 1-2This invention provides a method for suppressing delamination defects in dry electrodes based on in-situ monitoring, which may specifically include:

[0051] S1 collects calendering pressure data and powder flowability data in real time during the dry electrode preparation process, classifies the interaction pattern between calendering pressure level and powder flowability index, and identifies potential delamination defect risk areas.

[0052] Optionally, this step also includes:

[0053] Step S11: Real-time acquisition of pressure data and powder flowability data during the dry electrode calendering process using a sensor array to generate an initial set of process parameters.

[0054] Step S12: Based on the initial set of process parameters, the support vector machine algorithm is used for classification to obtain the interaction pattern between calendering pressure level and powder flowability index, and to determine the potential delamination defect risk area.

[0055] Step S13: Extract the stress data sequence and liquidity data sequence within the corresponding time period for the potential stratified defect risk area, and calculate the standard deviation of the stress data and the coefficient of variation of the liquidity data.

[0056] Step S14: If the standard deviation of the pressure data is higher than the preset pressure difference threshold and the coefficient of variation of the liquidity data is higher than the preset coefficient of variation threshold, then mark the current potential stratification defect risk area as a high-risk stratification defect area.

[0057] Step S15: Based on the set of process parameters corresponding to the high-risk stratified defect areas, cluster them into multiple defect pattern subsets, and use the support vector machine algorithm to retrain the defect pattern subsets, update the interaction pattern classification boundary, and obtain the optimized risk area division.

[0058] For example, in the dry electrode calendering process, data can be collected in real time using multi-point pressure sensors and powder flow monitoring devices to generate an initial set of process parameters, including calendering speed of 20-40 m / min, roll gap of 0.1-0.3 mm, and corresponding pressure distribution and flowability indicators. This real-time acquisition helps to capture process fluctuations and ensures that the data reflects the actual production status.

[0059] Specifically, using the support vector machine algorithm to classify the initial parameter set can effectively identify the interaction pattern between the calendering pressure level and the powder flowability index.

[0060] In one embodiment, when the pressure is above 500 N / cm and the flowability index is below 0.8, the classifier categorizes such parameters as a highly coupled mode, thereby revealing the nonlinear relationship between the two. This classification helps to quickly delineate normal and abnormal interaction regions, improving defect prediction accuracy.

[0061] It is understandable that the potential layered defect risk area is determined based on the interaction mode, mainly by mapping the mode boundary to the process parameter space.

[0062] For example, if a high-pressure, low-flow interaction pattern corresponds to an area with a roll gap of less than 0.15 mm, it is marked as a potential delamination defect risk area. This method of identification can pinpoint the potential defect parameter range in advance, avoiding the time wasted by full-batch inspection.

[0063] For example, stress data sequences and liquidity data sequences for the corresponding time periods are extracted for areas with potential stratification defects, and the stress standard deviation and liquidity coefficient of variation are calculated.

[0064] In one embodiment, if the pressure standard deviation exceeds the pressure difference threshold of 15 N / cm and the coefficient of variation is higher than the coefficient of variation threshold of 0.12, it is marked as a high-risk stratified defect area.

[0065] Specifically, during the rolling of lithium battery electrode sheets, the pressure is usually between 10-200 MPa (far higher than 15 N / cm, about 1.5 MPa). However, local pressure fluctuations (such as uneven roller gaps and fluctuations in powder feeding) are the main cause of delamination. If the pressure difference threshold is set to 15 N / cm (1.5 MPa), then a fluctuation of 18 N / cm (1.8 MPa) (exceeding the threshold by 20%) is sufficient to trigger local stress concentration, leading to particle displacement, which meets the conditions for delamination.

[0066] In one embodiment, a flow rate coefficient of variation greater than the coefficient of variation threshold of 0.12 is considered "unstable." In one possible implementation, a 10-second data sequence showing pressure fluctuations from 480 N / cm to 520 N / cm, with a standard deviation of 18 N / cm, and a flowability coefficient of variation of 0.15, indicates a tendency for stratification due to uneven powder distribution. This labeling can accurately pinpoint high-risk periods, improving the efficiency of defect tracing.

[0067] Specifically, multiple defect pattern subsets are formed by clustering the process parameters corresponding to high-risk areas. For example, parameters with excessively small roll gaps and excessively high speeds are clustered into one subset, while parameters with uneven pressure and abrupt changes in flowability are clustered into another subset. This clustering helps to separate defect patterns with different causes, facilitating targeted optimization.

[0068] Understandably, retraining the defect pattern subset using a support vector machine can update the interaction pattern classification boundary and obtain an optimized risk region division.

[0069] For example, the initial boundary might misclassify some edge parameters as risks. After retraining, the boundary better matches the actual data distribution, and the risk area is reduced by 20%. This optimization process significantly improves the model's robustness, reduces false alarms, and enhances the early warning capability of layered defects, ultimately reducing the scrap rate and stabilizing electrode quality.

[0070] S2 simulates the material bonding force distribution through a neural network model, determines the interlayer stress difference value, and if the interlayer stress difference value exceeds the preset stress difference threshold, adjusts the calendering pressure parameter to match the powder flowability range to obtain the optimized process parameter set.

[0071] Optionally, this step also includes:

[0072] Step S21: Obtain layered defect data inside the material through scanning technology, mark potential risk areas, and obtain preliminary defect distribution information.

[0073] Step S22: Based on the defect distribution information, a neural network model is used to simulate the material bonding force distribution, calculate the stress values ​​between each layer, and determine the distribution state of interlayer stress.

[0074] Step S23: Calculate the stress difference between adjacent layers based on the distribution of interlayer stress. If the stress difference exceeds the preset stress difference threshold, record the stress anomaly points in the corresponding area to obtain a set of anomaly areas.

[0075] Step S24: Extract key location data from the abnormal area set, analyze the influence of calendering pressure on powder flowability, determine whether parameters need to be adjusted, and obtain the pressure adjustment requirements.

[0076] Step S25: Based on the pressure adjustment requirements, calculate the matching range between powder flowability and calendering pressure, adjust the relevant parameters to the target range, and determine the optimized pressure parameter values.

[0077] Step S26: Obtain the optimized pressure parameter values, perform an overall verification in conjunction with the process parameter set, and determine whether the production conditions are met. If not, iteratively adjust until the requirements are met to obtain the final process parameter set.

[0078] Step S27: Based on the final set of process parameters, generate corresponding control commands, transmit them to the production equipment for parameter updates, and complete the process optimization process.

[0079] For example, after dry electrode fabrication, scanning technology can be used to acquire data on layered defects within the material. This allows for precise marking of previously identified potential risk areas, thus providing preliminary defect distribution information. Such scanning typically employs ultrasonic or X-ray tomography techniques to directly detect density differences and interface separation within the electrode sheet. In one embodiment, risk areas are scanned layer by layer; when a density abrupt change exceeds 5%, it is marked as a potential layering point, thereby forming a defect distribution map. This marking method helps transform abstract risks into visual information, facilitating subsequent analysis.

[0080] Understandably, by using a neural network model to simulate the distribution of material bonding forces based on defect distribution information, the stress values ​​between layers can be calculated, thereby determining the distribution of interlayer stress. The neural network simulates the distribution of bond strength between powder particles by inputting defect locations and material properties.

[0081] In one possible implementation, the model output shows that the bonding force between the active material layer and the current collector is 120 MPa in the central region, while it drops to 90 MPa at the edges, thus revealing the overall state of uneven stress. This simulation can accurately reflect the actual interlayer stress conditions and support the tracing of defect causes.

[0082] For example, based on the distribution of interlayer stress, the stress difference between adjacent layers is calculated. If the stress difference exceeds a preset stress difference threshold, such as 30 MPa, the stress anomaly points in the corresponding area are recorded, and finally a set of anomaly areas is obtained.

[0083] In one embodiment, areas with stress differences of up to 45 MPa between adjacent layers are marked as abnormal areas. This cluster is concentrated on the side with smaller roll gaps, indicating that uneven pressure directly leads to stress concentration. This recording method can quickly identify high-risk areas and improve defect prevention efficiency.

[0084] Understandably, by extracting key location data from the set of abnormal areas and analyzing the impact of calendering pressure on powder flowability, it is possible to determine whether parameters need to be adjusted and obtain the pressure adjustment requirements.

[0085] Specifically, by comparing historical pressure records corresponding to outliers, when the pressure peak exceeds 550 N / cm, the fluidity decreases significantly, with the analysis showing an impact of up to 65%, thus determining that the pressure needs to be reduced. This analysis helps to clarify the direction of parameter adjustments and avoid blind optimization.

[0086] For example, based on the pressure adjustment requirements, calculate the matching range between powder flowability and calendering pressure, adjust the relevant parameters to the target range, such as keeping the flowability index above 0.85 and controlling the pressure at 450-500 N / cm, and determine the optimized pressure parameter values.

[0087] In one embodiment, reducing the pressure from 520 N / cm to 480 N / cm stabilized the matching range and improved powder uniformity. This adjustment significantly reduced the probability of delamination, ensuring electrode consistency.

[0088] Understandably, after obtaining the optimized pressure parameter values, a comprehensive check is performed in conjunction with the set of process parameters to determine whether the production conditions are met. If not, the parameters are iteratively adjusted until they meet the requirements, ultimately yielding the final set of process parameters.

[0089] For example, if the calibration reveals that the roll gap of 0.12mm does not match the new pressure, it is iteratively increased to 0.18mm until all indicators meet the standards. This calibration process ensures the coordination between parameters and improves production stability.

[0090] For example, the final set of process parameters is used to generate corresponding control commands, which are then transmitted to the production equipment for parameter updates, thereby completing the process optimization process.

[0091] In one possible implementation, the instructions include setting a pressure of 480 N / cm and a speed of 30 m / min. After the equipment executes these instructions in real time, the defect rate decreases significantly. This closed-loop optimization can continuously improve electrode quality, reduce scrap, and stabilize mass production.

[0092] S3, run the structural stability test in the simulation environment to obtain the probability of delamination defects in the electrode material.

[0093] Optionally, this step also includes:

[0094] Step S31: Based on the optimized process parameter set, perform structural stability testing in the simulated environment by loading the parameter set, and obtain stress distribution data of the electrode material in the simulated environment.

[0095] Step S32: Use the finite element analysis method to determine the stress concentration region inside the electrode material.

[0096] Step S33: If the stress value in the stress concentration area exceeds the preset stress threshold, the corresponding parameter set is marked as a high-risk parameter set to obtain the potential location of the layered defect.

[0097] Step S34: Run multiple random perturbation tests on the high-risk parameter set in a simulation environment to obtain the probability of occurrence of layered defects, determine the correspondence between the probability values ​​of occurrence of layered defects and the high-risk parameter set, and determine the probability distribution of defect occurrence.

[0098] In one embodiment, structural stability tests are performed in a simulated environment by loading these optimized process parameters according to the optimized set of process parameters, which can verify the overall mechanical strength of the electrode material after calendering.

[0099] For example, by optimizing parameters through virtual loading, the deformation behavior of electrodes under standard charge-discharge cycles can be simulated to obtain key stability indicators. This type of testing helps to identify potential structural weaknesses in advance and improve the long-term durability of electrodes.

[0100] Specifically, structural stability tests are used to obtain stress distribution data of the electrode material under simulated conditions. For example...

[0101] In one possible implementation, test results show that the electrode surface stress is uniformly distributed in the range of 50 to 150 MPa, while the edge region is slightly higher, reaching 180 MPa. This data reflects the internal equilibrium state of the material after the rolling pressure is adjusted, which is beneficial for reducing damage caused by volume expansion during cycling.

[0102] For example, based on stress distribution data, the finite element analysis method can be used to determine the stress concentration regions inside the electrode material.

[0103] In one embodiment, after the electrode coating is meshed using the finite element method, analysis shows that the stress value in the stress concentration region at the particle interface can reach 200 MPa. If it exceeds a preset stress threshold, such as 150 MPa, it indicates a potential risk of cracking. This method can accurately locate high-stress points, supporting subsequent risk marking and thus improving the mechanical reliability of the electrode.

[0104] It should be noted that if the stress value in the stress concentration area exceeds the preset stress threshold, the corresponding parameter set is marked as a high-risk parameter set.

[0105] For example, when the stress value in the stress concentration area exceeds 180 MPa, the parameter set is identified as high risk. This helps to screen out parameters that are not conducive to interlayer bonding and avoid coating peeling in actual production.

[0106] In one possible implementation, the potential locations of layered defects are obtained based on a set of high-risk parameters.

[0107] For example, high-risk areas are mainly concentrated in the middle layer along the electrode thickness direction, with potential locations corresponding to areas of uneven particle packing. This method of data acquisition provides targeted input for subsequent simulations, reducing the probability of delamination defects.

[0108] For example, by using the Monte Carlo simulation method, multiple random perturbation tests are run in a simulation environment for a set of high-risk parameters to obtain the probability of layered defects occurring.

[0109] In one embodiment, after 1000 random perturbations, the probability of delamination defects ranges from 5% to 15%, with an average of 8%. This probability assessment takes into account material variability and process fluctuations, which is beneficial for quantifying the level of risk.

[0110] Specifically, for the probability of occurrence of layered defects, the correspondence between the probability values ​​of layered defects and the set of high-risk parameters is determined, and the probability distribution of defect occurrence is determined.

[0111] For example, a high-risk parameter set with a probability greater than 10% corresponds to higher rolling pressure, revealing a correlation between excessive pressure and flow mismatch. This assessment allows for parameter optimization to reduce the overall defect rate and improve battery consistency and safety.

[0112] For example, from multiple perspectives, structural stability testing and finite element analysis support each other. The former provides macroscopic stress data, while the latter accurately locates the concentration area. Monte Carlo simulation introduces randomness to assess uncertainty. The combination of the three forms a complete risk prediction chain, effectively reducing the capacity decay risk caused by layered defects.

[0113] S4 extracts performance consistency indicators from the probability of occurrence of layered defects, and determines the final defect control scheme by iterative comparison and matching with historical data.

[0114] Optionally, this step also includes:

[0115] Step S41: Obtain the original dataset for probability analysis from the data records of layered defects, classify and organize the defect distribution for each layer, and obtain a preliminary defect probability set.

[0116] Step S42: Based on the preliminary defect probability set, use a consistent performance evaluation standard to compare the defect probability of each level with the preset probability threshold range. If the defect probability of a certain level exceeds the defect probability threshold, it is marked as an abnormal level, and the distribution list of abnormal levels is determined.

[0117] Step S43: For the distribution list of the anomaly level, obtain the relevant data groups in the historical records, and match the data of the current anomaly level with the historical records one by one through iterative comparison to obtain the reference dataset with the highest matching degree.

[0118] Step S44: Based on the reference dataset with the highest matching degree, extract the strategy content related to defect control. If the strategy in the reference dataset is applicable to the current anomaly level, directly use the strategy to determine the initial control strategy combination.

[0119] Step S45: For the initial control strategy combination, the logical consistency between strategies is detected by data correlation analysis tools. If a strategy conflicts with other strategies, the strategy is removed to obtain the optimized control strategy set.

[0120] Step S46: Based on the optimized control strategy set and the processing flow of the scheme optimization, apply the strategy set to the current layered defect scenario, and determine the applicability of the final scheme by simulating execution to determine the final scheme for defect control.

[0121] Step S47: For the output of the final solution, the support vector machine algorithm is used to predict and analyze the execution effect of the solution. By comparing the prediction results, the adaptability distribution of the solution in different levels of defects is obtained, and the comprehensive coverage capability of the solution is judged.

[0122] For example, by obtaining the original dataset for probability analysis from the data records of layered defects, and classifying and organizing the defect distribution for each layer, a foundation for systematically constructing defect features can be built.

[0123] In one embodiment, the data recording includes defect logs after multiple simulation runs. By extracting defect counts and location information along the thickness direction of each coating layer, the data is stratified and classified to obtain a preliminary defect probability set. This set reflects the frequency of defect occurrence at different depths of the electrode, which helps in subsequent precise intervention.

[0124] Specifically, based on the initial defect probability set, a consistent evaluation standard is used to compare the defect probability of each level with a preset defect probability threshold range. If the defect probability of a certain level exceeds the defect probability threshold, it is marked as an abnormal level, and a distribution list of abnormal levels is determined.

[0125] In one possible implementation, a defect probability threshold of 8% is set. When the defect probability in the intermediate layer reaches 12%, that layer is marked as abnormal. This list clearly identifies all abnormal locations, supporting targeted analysis. This marking method can quickly identify areas of concentrated risk, preventing defect propagation from affecting the overall electrode stability.

[0126] For example, for a distribution list of anomaly levels, relevant data groups in the historical records are obtained. Through iterative comparison, the current anomaly level data is matched one by one with the historical records to obtain the reference dataset with the highest degree of matching.

[0127] In one embodiment, the comparison process considers defect type, location, and probabilistic similarity, with historical datasets showing a match rate of over 90% selected as references. This iterative matching ensures the reliability of the reference data and provides experience to control current defects.

[0128] Specifically, based on the reference dataset with the highest matching degree, the strategy content related to defect control is extracted. If the strategy is applicable to the current anomaly level, it is directly referenced to determine the initial combination of control strategies.

[0129] For example, if the reference data includes strategies for adjusting the surface roughness of the current collector, and the current anomaly level also involves interface bonding issues, these strategies are directly incorporated into the combination. This referencing method accelerates strategy formation and reduces redundant trials.

[0130] For example, for the initial combination of control strategies, data correlation analysis tools are used to check the logical consistency between strategies. If a strategy conflicts with others, it is removed, resulting in an optimized set of control strategies.

[0131] In one possible implementation, if a conflict is detected between increasing the binder ratio and decreasing the calendering speed, the latter is eliminated to prioritize interfacial strength. This optimization process ensures complementary strategies and improves the coordination of the control scheme.

[0132] Specifically, based on the optimized set of control strategies and the optimized processing flow of the solution, the strategies are applied to the current layered defect scenario. The applicability is judged through simulation execution, and the final defect control solution is determined.

[0133] For example, if the defect probability drops below 5% after simulated execution of the applied strategy, then the solution is confirmed as the final choice. This simulation verification helps to evaluate the effectiveness in advance and improves the practical value of the solution.

[0134] For example, the support vector machine algorithm is used to predict and analyze the execution effect of the final solution output. By comparing the prediction results, the adaptive distribution of the solution in different levels of defects is obtained, and the comprehensive coverage capability is judged.

[0135] In one embodiment, the predicted display scheme exhibits 95% adaptability to the surface layer and 85% adaptability to the intermediate layer, demonstrating good overall coverage. This predictive analysis quantifies the robustness of the scheme, supporting a comprehensive reduction in the risk of delamination defects and ensuring long-term mechanical reliability of the electrodes and consistency of battery performance.

[0136] S5 updates the real-time monitoring module of the manufacturing process using the final defect control scheme to obtain stable output results for battery industry applications.

[0137] Optionally, this step also includes:

[0138] Step S51: Obtain sensor data and equipment operating parameters collected in real time during the production process.

[0139] Step S52: Determine whether a defect signal exists from real-time data using an anomaly detection algorithm.

[0140] Step S53: If a defect signal exists, determine the defect type and severity.

[0141] Step S54: Obtain the control parameter adjustment rules for the corresponding defect type according to the final defect control scheme.

[0142] Step S55: Update the threshold and feedback coefficient of the real-time monitoring module using the control parameter adjustment rules.

[0143] Step S56: Process subsequent production process data through the updated real-time monitoring module to obtain intermediate output after defect suppression.

[0144] Step S57: Determine the degree of deviation between the intermediate output and the preset stable output standard.

[0145] In step S58, if the deviation exceeds the range, a supplementary correction instruction is obtained from the final defect control scheme to obtain the corrected monitoring module parameters.

[0146] For example, in the production process, real-time acquisition of sensor data and equipment operating parameters is fundamental to ensuring timely defect detection. Sensor data may include key indicators such as coating thickness, temperature changes, and pressure distribution, while equipment operating parameters encompass information such as roller speed and tension control. Real-time monitoring of this data provides the initial basis for subsequent defect detection. For instance, on an electrode production line, if a sensor collects coating thickness data every second and detects a range of thickness fluctuations exceeding the standard value, this could be an initial indication of a defect.

[0147] For example, in the application of anomaly detection algorithms, a specific threshold range can be set to determine whether a defect signal exists.

[0148] In one possible implementation, the algorithm analyzes the frequency and amplitude of abnormal fluctuations in real-time data and combines this with a defect feature library from historical data to determine whether the current signal is a defect signal. For example, if during the electrode coating process, the thickness data of a certain area is detected to exceed the normal range five times consecutively, with fluctuations exceeding 10%, the algorithm will classify it as a defect signal and further categorize it as a coating unevenness defect, with a severity level of medium.

[0149] For example, after determining the defect type and severity, it is crucial to obtain corresponding control parameter adjustment rules based on the final defect control scheme. Suppose the scheme's rules for addressing uneven coating defects involve adjusting the roller speed and coating amount ratio, specifically reducing the speed to 80% and increasing the coating amount by 5%. These rules will be directly used to update the thresholds and feedback coefficients of the real-time monitoring module, ensuring timely response to adjustment needs during subsequent production data processing.

[0150] For example, when processing subsequent production data, the updated real-time monitoring module dynamically adjusts equipment parameters based on new thresholds and feedback coefficients to obtain an intermediate output after defect suppression. Assuming the adjusted coating thickness fluctuation range is reduced to within 3% of the standard value, it indicates that the intermediate output has approached a stable state. This dynamic adjustment mechanism effectively reduces the likelihood of continued defect occurrence.

[0151] For example, when judging the degree of deviation between the intermediate output and the preset stable output standard, if the deviation is found to be outside the range, such as the thickness fluctuation still exceeding the standard value by 5%, then supplementary correction instructions need to be obtained from the final defect control scheme.

[0152] In one embodiment, the correction instruction might be to further reduce the roller speed to 75% and fine-tune the coating amount to an increase of 7%. These instructions update the monitoring module parameters to ensure that the production process gradually stabilizes.

[0153] For example, from multiple perspectives, the above method forms a closed-loop mechanism in defect suppression and parameter adjustment. Real-time data acquisition provides the foundation for defect detection, anomaly detection algorithms accurately locate problems, and control parameter adjustment rules and correction instructions ensure the targeted nature of intervention measures. Suppose that in electrode production, a batch of products frequently experiences coating unevenness issues. After adjusting the above process, the defect rate drops from 15% to 2%, significantly improving production stability. This closed-loop mechanism effectively ensures production continuity and product quality consistency.

[0154] Based on the embodiments of the present invention described above, and through the above description, those skilled in the art can make various changes and modifications without departing from the technical concept of the present invention. The technical scope of the present invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A method for suppressing delamination defects in dry electrodes based on in-situ monitoring, characterized in that, The method includes: S1. Real-time acquisition of calendering pressure data and powder flowability data during the dry electrode fabrication process; classification of the interaction patterns between calendering pressure levels and powder flowability indicators; identification of potential delamination defect risk areas. S2. Simulation of material bonding force distribution using a neural network model; determination of interlayer stress difference values; if the interlayer stress difference value exceeds a preset stress difference threshold, adjustment of calendering pressure parameters to match powder flowability range; obtaining an optimized set of process parameters. S3. Running structural stability tests in a simulated environment to obtain the probability of delamination defects in the electrode material. S4. Extracting performance consistency indicators from the probability of delamination defects; determining the final defect control scheme through iterative comparison and matching with historical data. S5. Updating the real-time monitoring module of the fabrication process using the final defect control scheme; obtaining stable output results for battery industry applications.

2. The method according to claim 1, characterized in that, Step S1 includes: Step S11: Real-time acquisition of pressure data and powder flowability data during the dry electrode calendering process using a sensor array to generate an initial set of process parameters; Step S12: Based on the initial set of process parameters, the support vector machine algorithm is used for classification to obtain the interaction pattern between the calendering pressure level and the powder flowability index, and to determine the potential risk area of ​​delamination defects. Step S13: Extract stress data sequences and liquidity data sequences for the corresponding time periods for potential stratified defect risk areas, and calculate the standard deviation of stress data and the coefficient of variation of liquidity data. Step S14: If the standard deviation of the pressure data is higher than the preset pressure difference threshold and the coefficient of variation of the liquidity data is higher than the preset coefficient of variation threshold, then mark the current potential stratification defect risk area as a high-risk stratification defect area. Step S15: Based on the set of process parameters corresponding to the high-risk stratified defect areas, cluster them into multiple defect pattern subsets, and use the support vector machine algorithm to retrain the defect pattern subsets, update the interaction pattern classification boundary, and obtain the optimized risk area division.

3. The method according to claim 2, characterized in that, Step S14 includes: setting the overpressure difference threshold to 15 N / cm and the coefficient of variation threshold to 0.

12.

4. The method according to claim 1, characterized in that, Step S2 includes: Step S21: Obtain layered defect data inside the material through scanning technology, mark potential risk areas, and obtain preliminary defect distribution information; Step S22: Based on the defect distribution information, a neural network model is used to simulate the material bonding force distribution, calculate the stress values ​​between each layer, and determine the distribution state of interlayer stress. Step S23: Calculate the stress difference between adjacent layers based on the distribution of interlayer stress. If the stress difference exceeds the preset stress difference threshold, record the stress anomaly points in the corresponding area to obtain a set of anomaly areas. Step S24: Extract key location data from the abnormal area set, analyze the influence of calendering pressure on powder flowability, determine whether parameters need to be adjusted, and obtain the pressure adjustment requirements. Step S25: Based on the pressure adjustment requirements, calculate the matching range between powder flowability and calendering pressure, adjust the relevant parameters to the target range, and determine the optimized pressure parameter values. Step S26: Obtain the optimized pressure parameter values, perform an overall verification in conjunction with the process parameter set, and determine whether the production conditions are met. If not, iteratively adjust until the requirements are met to obtain the final process parameter set. Step S27: Based on the final set of process parameters, generate corresponding control commands, transmit them to the production equipment for parameter updates, and complete the process optimization process.

5. The method according to claim 1, characterized in that, Step S3 includes: Step S31: Based on the optimized process parameter set, perform structural stability testing in the simulated environment by loading the parameter set, and obtain stress distribution data of the electrode material in the simulated environment; Step S32: Use the finite element analysis method to determine the stress concentration region inside the electrode material; Step S33: If the stress value in the stress concentration area exceeds the preset stress threshold, mark the corresponding parameter set as a high-risk parameter set and obtain the potential location of the layered defect. Step S34: Run multiple random perturbation tests on the high-risk parameter set in a simulation environment to obtain the probability of occurrence of layered defects, determine the correspondence between the probability values ​​of occurrence of layered defects and the high-risk parameter set, and determine the probability distribution of defect occurrence.

6. The method according to claim 1, characterized in that, Step S4 includes: Step S41: Obtain the original dataset for probability analysis from the data records of layered defects, classify and organize the defect distribution for each layer, and obtain a preliminary defect probability set. Step S42: Based on the preliminary defect probability set, use a consistent performance evaluation standard to compare the defect probability of each level with the preset defect probability threshold range. If the defect probability of a certain level exceeds the defect probability threshold, it is marked as an abnormal level, and the distribution list of abnormal levels is determined. Step S43: For the distribution list of the anomaly level, obtain the relevant data groups in the historical records, and match the data of the current anomaly level with the historical records one by one through iterative comparison to obtain the reference dataset with the highest matching degree. Step S44: Based on the reference dataset with the highest matching degree, extract the strategy content related to defect control. If the strategy in the reference dataset is applicable to the current anomaly level, directly use the strategy to determine the initial control strategy combination. Step S45: For the initial control strategy combination, the logical consistency between strategies is detected by data correlation analysis tools. If a strategy conflicts with other strategies, the strategy is removed to obtain the optimized control strategy set. Step S46: Based on the optimized control strategy set and the processing flow of the scheme optimization, apply the strategy set to the current layered defect scenario, and determine the applicability of the final scheme by simulating execution to determine the final scheme for defect control. Step S47: For the output of the final solution, the support vector machine algorithm is used to predict and analyze the execution effect of the solution. By comparing the prediction results, the adaptability distribution of the solution in different levels of defects is obtained, and the comprehensive coverage capability of the solution is judged.

7. The method according to claim 1, characterized in that, Step S5 includes: Step S51: Obtain sensor data and equipment operating parameters collected in real time during the production process; Step S52: Determine whether a defect signal exists from real-time data using an anomaly detection algorithm; Step S53: If a defect signal exists, determine the defect type and severity. Step S54: Obtain the control parameter adjustment rules for the corresponding defect type according to the final defect control scheme; Step S55: Update the threshold and feedback coefficient of the real-time monitoring module using the control parameter adjustment rules; Step S56: Process subsequent production process data through the updated real-time monitoring module to obtain intermediate output after defect suppression; Step S57: Determine the degree of deviation between the intermediate output and the preset stable output standard; In step S58, if the deviation exceeds the range, a supplementary correction instruction is obtained from the final defect control scheme to obtain the corrected monitoring module parameters.

Images