Lithium battery diaphragm whole-process production data transmission optimization method and system

Abstract

The invention discloses a lithium battery diaphragm whole-process production data transmission optimization method and system, and particularly relates to the technical field of data transmission. Sensor data of a lithium battery diaphragm whole process is collected through a sensor array, the sensor data is extracted to obtain actual process parameters, and the actual process parameters are compared with preset process parameters; outputting the abnormal data; based on a space-time topological structure of a sensor, calculating space-time-function coupling weights among abnormal data, constructing a structural coupling matrix, identifying error clustering candidate bodies, and extracting space-time distribution characteristics and abnormal combination modes of the error clustering candidate bodies; performing structural classification on the error clustering candidate bodies by utilizing a preset rule engine, and outputting a deviation report; and outputting a sensor calibration instruction and a process fine tuning instruction based on the deviation report, thereby solving the problems that the deviation source is difficult to accurately identify, the key abnormal data transmission is not timely and the false alarm rate is high.

Description

Technical Field

[0001] The present invention relates to the field of data transmission technology, and more specifically, to a method and system for optimizing data transmission throughout the entire production process of lithium battery separators. Background Art

[0002] In the manufacturing process of lithium battery separators, in order to ensure the thickness uniformity and process stability of the final product, key parameters in multiple links such as drying, stretching, and rolling need to be continuously monitored and precisely controlled. To this end, the production system usually deploys a large number of multi-type sensors such as temperature, pressure, tension, thickness, etc. to collect real-time data from each process section. With the improvement of the level of intelligent manufacturing, more and more edge computing nodes are used to process local data to support local early warning, process adaptive adjustment and quality traceability. However, in the context of the surge in sensor data, how to accurately identify the source of deviation in the production process has become a key factor affecting product quality and production line operation efficiency.

[0003] Traditional methods generally use residual analysis, single-point trend monitoring and other means to detect deviations in process parameters. These methods are somewhat effective when dealing with single anomalies under stable working conditions. However, when faced with the spatial coupling, temporal drift and functional linkage characteristics of multi-source data in complex working conditions, traditional methods are often unable to identify systematic deviations, and are particularly difficult to effectively distinguish between quality anomalies caused by sensor drift, process disturbances or recipe deviations. In addition, the existing technology lacks priority judgment and dynamic adaptation strategies for the transmission bandwidth scheduling mechanism of abnormal data, resulting in the possibility of delay or loss of key data under network constraints, further reducing the timeliness and accuracy of abnormal response. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a method and system for optimizing data transmission in the entire production process of lithium battery diaphragms. By constructing a spatiotemporal-functional coupling model of sensor data, introducing an anomaly scoring-driven bandwidth scheduling mechanism and an error clustering identification and classification strategy, the present invention solves the problems of difficulty in accurately identifying the source of deviation, untimely transmission of key abnormal data, and high false alarm rate raised in the above-mentioned background technology.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for optimizing data transmission in the entire production process of a lithium battery separator, comprising the following steps: Step 1: Sensor data from the entire lithium battery separator process is collected through a sensor array and transmitted to a local edge database through an edge node. Production data includes at least sensor data corresponding to the drying process and the stretching process. By setting temperature sensors and pressure sensors on the edge gateways of each thermal control zone and stretching section, the temperature-time curve of the thermal control zone and the pressure-time curve of the stretching zone with confidence annotations are output. Step 2: Extract sensor data to obtain actual process parameters, compare the actual process parameters with the preset process parameters, and output abnormal data; Step 3: Deviation Source Identification: Receive abnormal data and, based on the spatiotemporal topology of the sensors, calculate the spatial-temporal-functional coupling weights between the abnormal data. Construct a structural coupling matrix to reflect the correlation between sensor anomalies. Based on the structural coupling matrix, identify error cluster candidates, extract their spatiotemporal distribution characteristics and abnormal combination patterns. Use a pre-set rule engine to structurally classify the error cluster candidates and output a deviation report. Step 4: Output sensor calibration instructions and process fine-tuning instructions based on the deviation report, execute the sensor calibration instructions and process fine-tuning instructions, and monitor the calibration results and process adjustment results.

[0006] Preferably, when the cumulative amount or frequency of abnormal data exceeds the limit, an alarm message is issued. Based on the abnormal score of the sensor data and the current network bandwidth status, the network bandwidth is matched for the sensor data to ensure the transmission priority of the abnormal sensor data. Sensor data with a high abnormal score value is transmitted first.

[0007] Preferably, the abnormality score is quantified in a manner that: the abnormality score is comprehensively calculated based on the degree of deviation between the actual process parameters of the sensor data and the preset process parameters, the duration of the deviation, and the degree of impact of the deviation on the production quality.

[0008] Preferably, the sensor deviation source identification process includes the following steps: Step S101: Data preprocessing and feature extraction: After receiving abnormal data from each sensor data channel, a graph neural network model or clustering algorithm is input based on the sensor's spatiotemporal topological structure. The spatial-temporal-functional coupling weights between the abnormal data are calculated based on temporal synchrony, spatial proximity, and functional category similarity to form a structural coupling matrix. Based on the structural coupling matrix, candidate error clusters are screened. Step S102: Analyze error cluster candidates, extract their spatiotemporal distribution characteristics and abnormal combination patterns, perform structural classification on the error cluster candidates based on a rule engine and a machine learning classification model, and output error labels; Step S103: Output a deviation report including the deviation type and coverage. The deviation types include process-induced errors, recipe-induced errors, and sensor-induced errors. The coverage refers to the time range in which the deviation occurs, the work sections and sensor indexes involved, and the degree of impact of the thickness deviation.

[0009] Preferably, if the deviation tag indicates an error caused by the process, process adjustment is delayed, sensor data is continuously collected, and steps S101 to S103 are repeated for confirmation to avoid erroneous feedback that disturbs the process; If the deviation tag indicates that the sensor causes the error, a calibration instruction is sent to the edge node where the corresponding sensor is located to adjust the gain or bias coefficient of the corresponding sensor; If the process causes errors, a process fine-tuning instruction is sent to the edge node where the corresponding sensor is located.

[0010] Preferably, the sensor calibration instruction refers to adjusting the gain / offset calibration coefficient of the sensor in the abnormal section; the process fine-tuning instruction refers to adjusting the process parameters in the abnormal section, and the sensor calibration instruction is executed in the following manner: If the sensor's historical drift amplitude exceeds the threshold for a long period of stability or the residual fluctuation before calibration is abnormally severe, it is judged that the sensor performance has deteriorated or failed, and a prompt is given to replace the sensor component. If the replacement conditions are not met, continue the calibration process; Send sensor calibration instructions to the edge node where the corresponding sensor is located. The execution of the sensor calibration instructions includes the calibration target value and the calibration coefficient adjustment strategy. Depending on the sensor type, perform gain calibration or offset calibration. Gain calibration refers to correcting the slope coefficient of the sensor measurement output through linear regression, while offset calibration refers to compensating the measurement value for the baseline offset. Compare the sensor output within a certain time window before and after calibration with the preset standard parameters to calculate the calibration error residual; When the calibration error is lower than the set threshold, the calibration is considered successful and the calibration parameters and their effects are recorded.

[0011] Preferably, the method for executing the process fine-tuning instruction is: Step S201: Based on the identified error cluster data, using the time-aligned temperature, pressure, and thickness measurement data, a temperature-pressure-thickness coupling model with an adaptive weight adjustment mechanism is constructed. An ensemble learning algorithm is used to fuse ridge regression and decision tree regression to build corresponding sub-models for different production recipe scenarios. Dynamic switching is performed under different production recipe scenarios to improve prediction accuracy under nonlinear working conditions. Step S202: Introducing an online residual monitoring mechanism based on a sliding window to evaluate the residual between the predicted thickness and the actual thickness data in real time. When the residual continuously exceeds the threshold range or shows a deviation trend, triggering abnormal location reasoning, automatically locating the variable causing the residual and its corresponding work section; Step S203: generating a process fine-tuning instruction under the minimum disturbance principle according to the importance ranking and deviation direction of the variables; Step S204: After the correction instruction is issued, the execution segment data is collected in real time and fed back to the temperature-pressure-thickness coupling model to update the model parameters, achieve dynamic convergence of the adaptive working condition evolution, and fine-tune the process parameters based on the predicted thickness; Step S205: The update log and the results of each round of process fine-tuning instructions, thickness improvement, and residual fluctuation changes are all archived in the edge database for subsequent quality backtracking and process stability assessment to improve the system's steady-state operation capability.

[0012] Preferably, the method also includes: Step 5: Collect historical production data covering the temperature in the temperature control zone, the pressure in the stretching zone, and the thickness data, covering different production formulas and working conditions, analyze the historical production data, use the ridge regression equation, combine cross-validation to optimize the regression coefficient, and establish an initial thickness prediction model; select a drift-free verification data set to test the prediction performance, trigger the thickness prediction model update based on the abnormal data accumulation threshold, test the mean square error of the updated model under actual working conditions, and enable the new model if it meets the standard, otherwise roll back to the historical version and alarm.

[0013] Preferably, the thickness prediction model is constructed as follows: Step S301: collecting historical production data, including temperature data of the heat control zone, pressure data of the stretching zone, and actual thickness data of the downstream thickness gauge; Step S302: Correct the timing offset between the sensors in the thermal control zone and the stretching zone to ensure synchronization of temperature, pressure, and thickness data; remove random noise through sliding average or low-pass filtering; correct sensor drift based on historical calibration records; use the 3σ criterion or median absolute deviation to eliminate some abnormal data to reduce data deviation caused by drift or faults; normalize the temperature, pressure, and thickness data to the range of 0 to 1 to eliminate dimensional differences; Step S303: Using ridge regression to train the ridge regression equation, and using cross-validation to optimize the regression coefficient; training until the loss function meets the requirements, where the loss function is the mean square error between the actual thickness data and the predicted thickness; Step S304: After the training is completed, a set of known drift-free validation data sets is selected to test the prediction performance of the initial thickness prediction model; after the verification is passed, the application is deployed.

[0014] Preferably, the thickness prediction model is updated based on the accumulation of abnormal data and the updated model is output; the thickness prediction model is updated in the following manner: When the abnormal data accumulates to the threshold, the update of the thickness prediction model is started; Aggregate abnormal data by production batch and perform preprocessing operations, which at least include timestamp alignment, denoising and drift correction, and normalization. Receive normalized abnormal data, perform ridge regression, and output updated thickness prediction model coefficients to reduce the influence of historical data. Use weighted regression to give higher weight to recent abnormal data based on the exponential decay function. The residuals of the updated model are compared with the historical model. If the mean square error of the updated model is reduced, the historical model is replaced, and the verified model is stored in the edge database according to the timestamp, production recipe and sensor calibration status.

[0015] To achieve the above objectives, the present invention provides the following technical solutions: a lithium battery separator full-process production data transmission optimization system, comprising: The sensor data acquisition module collects sensor data of the entire process of lithium battery separators through the sensor array and transmits it to the local edge database through the edge node; The abnormal data screening module extracts sensor data to obtain actual process parameters, compares the actual process parameters with the preset process parameters, and outputs abnormal data; The deviation source identification module receives abnormal data and, based on the spatiotemporal topology of the sensors, calculates the spatial-temporal-functional coupling weights between the abnormal data, constructs a structural coupling matrix, and identifies error cluster candidates based on the structural coupling matrix, extracts their spatiotemporal distribution characteristics and abnormal combination patterns; structurally classifies the error cluster candidates using a preset rule engine and outputs a deviation report; The instruction execution module outputs sensor calibration instructions and process fine-tuning instructions based on the deviation report, executes the sensor calibration instructions and process fine-tuning instructions, and monitors the calibration results and process adjustment results; the way to execute the process fine-tuning instructions is: to build a coupling model that integrates temperature, pressure and thickness data, and realize dynamic switching prediction under multiple production recipe scenarios through an integrated learning algorithm; to introduce an online residual monitoring mechanism based on a sliding window to realize real-time diagnosis of prediction deviations and anomaly location; and to generate minimum disturbance process fine-tuning instructions based on the importance of key variables and the direction of deviation.

[0016] Technical effects and advantages of the present invention: The lithium battery separator full-process production data transmission optimization method provided by the present invention, by introducing the spatiotemporal topological structure and structural coupling matrix of the sensor, combined with graph neural network, rule engine and machine learning model, realizes cluster recognition and refined classification of sensor anomalies, and automatically generates response instructions according to the deviation type; it not only improves the accuracy and interpretation ability of abnormal source identification, but also can effectively distinguish different types of errors such as sensor drift, process deviation and recipe anomaly, reducing the risk of misjudgment and misoperation; it supports scoring of abnormal data and dynamically adjusts its transmission priority to ensure rapid response and reporting of high-risk anomalies.

[0017] The method proposed in this paper optimizes data transmission throughout the lithium battery separator production process. By combining a thickness prediction model, a process correction algorithm, and an adaptive sensor calibration mechanism, a closed-loop feedback control system based on the temperature-pressure-thickness coupling relationship is established. By dynamically constructing sub-models for specific recipe scenarios through an integrated learning approach, and integrating online residual monitoring and variable importance analysis, this method enables real-time fine-tuning of process parameters with minimal disturbance while ensuring product quality, effectively preventing the impact of significant process fluctuations on separator quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a flowchart for optimizing data transmission for the entire production process of the lithium battery separator of the present invention.

[0019] Figure 2 A flow chart is constructed for the thickness prediction model of the present invention.

[0020] Figure 3 This is a structural block diagram of the data transmission optimization system for the full process production of lithium battery separators of the present invention. DETAILED DESCRIPTION

[0021] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0022] At the same time, it should be understood that for the convenience of description, the sizes of the various parts shown in the drawings are not drawn according to the actual proportional relationship.

[0023] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the present disclosure, its application, or uses.

[0024] Technologies, methods, and equipment known to ordinary technicians in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and equipment should be considered part of the specification.

[0025] Example 1, see Figure 1 The present invention provides a flowchart for optimizing the data transmission of the whole process of lithium battery separator production. Figure 1 A method for optimizing data transmission in the entire production process of a lithium battery separator is shown, comprising the following steps: Step 1: Sensor data from the entire lithium battery separator process is collected through a sensor array and transmitted to a local edge database via an edge node. Production data includes at least sensor data corresponding to the drying process and the stretching process. By setting temperature sensors and pressure sensors on the edge gateways of each thermal control zone and stretching section, the temperature-time curve of the thermal control zone and the pressure-time curve of the stretching zone with confidence annotations are output. The preprocessing method for step 1 is: timestamp the sensor data; use the dynamic time warping algorithm to correct the timing offset between sensors in real time. Step 2: Extract sensor data to obtain actual process parameters, compare the actual process parameters with the preset process parameters, and output abnormal data; Step 3: Deviation Source Identification: Receive abnormal data and calculate the spatial-temporal-functional coupling weights between abnormal data based on the spatiotemporal topology of the sensors. The spatial-temporal-functional coupling weights are calculated based on temporal synchrony, spatial proximity, and functional category similarity. A structural coupling matrix is ​​constructed to reflect the correlation between sensor anomalies. Based on the structural coupling matrix, error cluster candidates are identified, and their spatiotemporal distribution characteristics and abnormal combination patterns are extracted. A preset rule engine is used to structurally classify the error cluster candidates and output a deviation report. Step 4: Output sensor calibration instructions and process fine-tuning instructions based on the deviation report, execute the sensor calibration instructions and process fine-tuning instructions, and monitor the calibration results and process adjustment results.

[0026] In one possible embodiment, when the cumulative amount or frequency of abnormal data exceeds a limit, an alarm is issued. Based on the abnormality score of the sensor data and the current network bandwidth, the network bandwidth is matched to the sensor data to ensure the transmission priority of the abnormal sensor data. Sensor data with a high abnormality score is transmitted first. Specific implementation methods include: To avoid bandwidth limitations during critical periods, which could lead to delays or loss of important data, edge nodes apply different compression ratios to temperature data at three levels: "sudden changes first, drift second, and normal data lowest" (for example, lossless compression for abnormal data, slight loss for drift data, and high compression for normal data). This significantly reduces bandwidth usage. Only high-priority data within a ±30s window is transmitted in real time, at the sub-second level. Non-abnormal data is accumulated to a specified time interval (e.g., every 5 minutes) and then uploaded in batches to minimize bandwidth peak contention. The available link bandwidth is measured periodically (e.g., every 10 seconds) and the results are sent to edge nodes. The nodes automatically adjust the number of parallel transmission threads and shard size accordingly, ensuring that critical data can still be transmitted first even when bandwidth is limited. To avoid the problem of insufficient communication reliability caused by dependence on a single path, the edge node is connected to two or more links at the same time, and a health detection probe is configured for each link. When the latency or packet loss rate of the main link exceeds the preset threshold, the node automatically switches high-priority data to an available backup link and reroutes it, logging and issuing alarms for the switching event. The sensor data with the highest anomaly score can be sent in parallel on both the main and backup channels. The results of the two channels are compared through CRC check, and the receipt is confirmed if they are consistent. If one channel fails, the other can still ensure data integrity.

[0027] In one possible embodiment, the anomaly score is quantified by comprehensively calculating the anomaly score based on the degree of deviation between the actual process parameters of the sensor data and the preset process parameters, the duration of the deviation, and the impact of the deviation on production quality. Specifically, for each sensor data point, the residual between its actual value and the preset value is calculated. The residual is defined as the ratio of the difference between the actual value and the preset value to the preset value to normalize the deviation. The cumulative mean and standard deviation of the residuals are calculated using a sliding window (e.g., a window size of 5 minutes) to reflect the persistence and volatility of the deviation. In addition, based on process knowledge, the impact of different sensor data on the thickness uniformity of the lithium battery separator is weighted. For example, the weight of the temperature sensor in the thermal control zone is higher than that of the pressure sensor in the stretching zone. The weight values ​​are determined based on historical production data and regression analysis of thickness deviations. The anomaly score is obtained through weighted summation. The weight coefficient can be dynamically adjusted according to the production recipe and operating conditions. The higher the anomaly score value, the greater the impact of the sensor data anomaly on production quality and the higher the priority.

[0028] Background information: In the production of lithium battery separators, identifying the source of deviation is a key link in ensuring thickness uniformity and process stability. Although traditional deviation identification methods (such as residual analysis) can locate sensor drift or process deviation, they may misjudge when faced with transient disturbances, leading to false alarms or production interruptions.

[0029] It should be further explained in the embodiment of the present invention that the sensor deviation source identification process includes the following steps: Step S101: Data preprocessing and feature extraction: After receiving abnormal data from each sensor data channel, a graph neural network model or clustering algorithm is input based on the sensor's spatiotemporal topological structure. The spatial-temporal-functional coupling weights between the abnormal data are calculated based on temporal synchrony, spatial proximity, and functional category similarity to form a structural coupling matrix. Based on the structural coupling matrix, candidate error clusters are screened. Explanation: The spatiotemporal topological structure of the sensor refers to the network structure formed based on the physical position of the sensor in the lithium battery separator production process (spatial topology, such as the distribution of thermal control zones and stretching zones) and the time series of data acquisition (temporal topology); through graph neural networks or clustering algorithms, combined with time synchronization, spatial proximity and functional category similarity, the coupling relationship between abnormal data is calculated to identify the source of deviation.

[0030] Explanation: The space-time-function coupling weight is a weighted indicator used to measure the similarity or correlation of abnormal data of multiple sensors in the dimensions of spatial position, temporal behavior and functional type, reflecting whether the abnormal data are correlated, and then used to discover whether the clustered abnormal behavior comes from systematic process problems rather than isolated sensor abnormalities; the structural coupling matrix is ​​a matrix constructed with sensors as nodes. Each element in the matrix represents the coupling strength between any two sensors (that is, the above-mentioned coupling weight), which is similar to the weighted form of the node adjacency matrix in the graph neural network, and is used to identify which sensors' data anomalies have group consistency (that is, clustering).

[0031] Step S102: Analyze error cluster candidates, extract their spatiotemporal distribution characteristics and abnormal combination patterns, perform structural classification on the error cluster candidates based on a rule engine and machine learning classification model, and output error labels. The rule engine, based on process knowledge and sensor configuration, matches the following conditions: whether the error cluster candidate is located in the stretching zone or thermal control zone, whether multiple sensor types (such as temperature, pressure, thickness, strain) are involved, and whether there is tension feedback delay or temperature control lag. If the conditions are met, the error cluster candidate is marked as a process-induced error. If only a single sensor is abnormal and the abnormality score is high, it is marked as a sensor-induced error; If the anomaly score is correlated with the recipe parameter deviation and has no structural coupling, it is marked as recipe-induced error; If the abnormal score is related to the deviation of process parameters in a specific work section, it is marked as process-induced error; The rule engine is explained as a conditional judgment based on process knowledge. For example, large temperature fluctuations in the heat control zone accompanied by abnormal tension indicate process-induced errors. It is used to determine whether the error has structural characteristics. The machine learning classification model is a trained classification algorithm model (such as a decision tree, SVM, random forest, or graph neural network) used to learn abnormal characteristic patterns from clustered data. When the rule engine cannot fully cover all abnormal patterns, the machine learning classification model can make data-driven judgments. It combines inputs such as spatiotemporal characteristics, deviation trends, and anomaly scores to output error labels. Step S103: Output a deviation report including the deviation type and coverage. The deviation types include process-induced errors, recipe-induced errors, and sensor-induced errors. The coverage refers to the time range in which the deviation occurs, the work sections and sensor indexes involved, and the degree of impact of the thickness deviation.

[0032] In one possible embodiment, if the deviation tag indicates an error caused by the process, process adjustment is delayed (for example, an alarm is set after 5 minutes), sensor data is continuously collected, and steps S101 to S103 are repeated for confirmation to avoid erroneous feedback that disrupts the process; If the deviation tag indicates that the sensor causes the error, a calibration instruction is sent to the edge node where the corresponding sensor is located to adjust the gain or bias coefficient of the corresponding sensor; If the process causes an error, a process fine-tuning instruction is sent to the edge node where the corresponding sensor is located. For example, the target heating rate of the thermal control zone or stretching zone is corrected, or it is recommended to check the equipment status.

[0033] It should be further explained in the embodiments of the present invention that the sensor calibration instruction refers to adjusting the gain / offset calibration coefficients of sensors (such as temperature sensors and pressure sensors) in the abnormal section; the process fine-tuning instruction refers to adjusting the process parameters in the abnormal section, correcting the target heating rate of each section during the next preheating or recipe switching, and correcting the pressure curve during the diaphragm rolling or winding process to minimize the fluctuation of the layer thickness and continuously converge the thickness prediction model with the on-site working conditions; the method for executing the sensor calibration instruction is: If the sensor's historical drift amplitude exceeds the threshold for a long period of stability or the residual fluctuation before calibration is abnormally severe, it is judged that the sensor performance has deteriorated or failed, and a prompt is given to replace the sensor component. If the replacement conditions are not met, continue the calibration process; Send sensor calibration instructions to the edge node where the corresponding sensor is located. The execution of the sensor calibration instructions includes the calibration target value and the calibration coefficient adjustment strategy. Depending on the sensor type, perform gain calibration or offset calibration. Gain calibration refers to correcting the slope coefficient of the sensor measurement output through linear regression, while offset calibration refers to compensating the measurement value for the baseline offset. Compare the sensor output within a certain time window before and after calibration with the preset standard parameters to calculate the calibration error residual; When the calibration error is lower than the set threshold, the calibration is considered successful and the calibration parameters and their effects are recorded; otherwise, the calibration process is repeated or manual maintenance suggestions are triggered.

[0034] It is necessary to further explain in the embodiment of the present invention that the method of executing the process fine-tuning instruction is: Step S201: Based on the identified error cluster data, using time-aligned temperature, pressure, and thickness measurement data, a temperature-pressure-thickness coupling model with an adaptive weight adjustment mechanism is constructed. An ensemble learning algorithm is used to fuse ridge regression and decision tree regression to build corresponding sub-models for different production recipe scenarios. Dynamic switching is performed under different production recipe scenarios to improve prediction accuracy under nonlinear working conditions. A thickness prediction model is constructed based on key factors. A multivariate regression analysis is performed on the temperature-time curve, pressure-time curve, and lithium battery separator thickness data from a downstream laser / β-ray thickness gauge to output a thickness prediction model. Step S202: Introduce an online residual monitoring mechanism based on a sliding window to perform real-time evaluation of the residual between the predicted thickness and the actual thickness data. When the residual continuously exceeds the threshold range or shows a deviation trend, trigger anomaly location reasoning to automatically locate the variable causing the residual (for example, through principal component analysis, key factors leading to thickness anomalies include temperature and pressure) and the corresponding process section; Step S203: Generate process fine-tuning instructions based on the minimum disturbance principle according to the importance ranking and deviation direction of the variables, including but not limited to fine-tuning the tension change rate in the stretching section, adjusting the heating slope in the heat control zone, and optimizing the temperature stabilization time in the preheating zone; Step S204: After the correction instruction is issued, the execution segment data is collected in real time and fed back to the temperature-pressure-thickness coupling model to update the model parameters, achieve dynamic convergence of the adaptive working condition evolution, and fine-tune the process parameters based on the predicted thickness; Step S205: The update log and the results of each round of process fine-tuning instructions, thickness improvement, and residual fluctuation changes are all archived in the edge database for subsequent quality backtracking and process stability assessment to improve the system's steady-state operation capability.

[0035] The production process of lithium battery separators involves extrusion, drying, stretching, thickness measurement and winding. The thermal control zone is used to control the crystallinity of the membrane or remove the solvent. The stretching zone forms a microporous structure through tension. The thickness measurement step ensures that the thickness of the separator is uniform. The thickness uniformity and quality stability of the separator directly affect the safety and performance of the battery. The production involves multiple thermal control zones and stretching zones. The temperature and pressure parameters of each zone need to be precisely controlled to ensure the uniformity of the separator thickness. However, in actual production, the sensor may drift after long-term operation. The switching of production formulas (such as the process requirements of different types of separators) will lead to changes in working conditions, which will cause the thickness prediction of the lithium battery separator to gradually deviate from the actual working conditions, resulting in thickness prediction deviations and quality fluctuations. In order to solve the problem that the thickness prediction model has not been calibrated for a long time and cannot reflect the changes in on-site working conditions, Example 2 is set.

[0036] Example 2, see Figure 2The thickness prediction model construction flow chart is provided, and the method also includes: Step 5: Collect historical production data covering the temperature in the temperature control zone, the pressure in the stretching zone, and the thickness data, covering different production recipes and working conditions, analyze the historical production data, use the ridge regression equation, combine cross-validation to optimize the regression coefficient, and establish an initial thickness prediction model; select a drift-free verification data set to test the prediction performance, trigger the thickness prediction model update based on the abnormal data accumulation threshold, test the mean square error of the updated model under actual working conditions, and enable the new model if it meets the standard, otherwise roll back to the historical version and issue an alarm.

[0037] It needs to be further explained in the embodiment of the present invention that the thickness prediction model is constructed in the following manner: Step S301: Collect historical production data, including temperature data of the heat control zone, pressure data of the stretching zone, and actual thickness data of the downstream thickness gauge; the collected data covers different production recipes (such as different thickness specifications) and operating conditions (such as normal operation and recipe switching); Step S302: Correct the timing offset between the sensors in the thermal control zone and the stretching zone to ensure synchronization of temperature, pressure, and thickness data; remove random noise through sliding average or low-pass filtering; correct sensor drift based on historical calibration records (such as adjusting bias or gain); use the 3σ criterion or median absolute deviation to eliminate some abnormal data to reduce data deviation caused by drift or faults; normalize the temperature, pressure, and thickness data to the range of 0 to 1 to eliminate dimensional differences; Step S303: Ridge regression is used (if there is multicollinearity between features, ridge regression is preferred) to train the ridge regression equation, and cross-validation (such as 5-fold cross-validation) is used to optimize the regression coefficient βi; the training is performed until the loss function meets the requirements, where the loss function is the mean square error between the actual thickness data and the predicted thickness; Step S304: After the training is completed, a set of known drift-free validation data sets (sensor data obtained through laboratory calibration or recent maintenance) is selected to test the prediction performance of the initial thickness prediction model; after verification, the application is deployed.

[0038] Explanation: In view of the multicollinearity of the temperature in the heat control zone, the pressure in the stretching zone and the downstream thickness data, the regularization parameter λ is dynamically adjusted, and the optimal λ value is determined through k-fold cross validation to balance the model deviation and generalization ability; to adapt to the switching of production recipes, the recent data is given a higher weight based on the exponential decay function, and the thickness prediction model is updated to reduce the interference of historical data on nonlinear working conditions.

[0039] Background information: The initial thickness prediction model is trained based on historical data. If it has not been updated for a long time, it cannot adapt to equipment wear, environmental changes or recipe switching, resulting in increased thickness prediction errors. The update of the thickness prediction model is initiated based on the accumulation of abnormal data, and the updated model is output.

[0040] It needs to be further explained in the embodiment of the present invention that the thickness prediction model is updated in the following manner: When the abnormal data accumulates to the threshold, the update of the thickness prediction model is started; Abnormal data is aggregated by production batch and preprocessed, including at least timestamp alignment, denoising and drift correction, noise removal through sliding average, drift correction and normalization based on gain / bias coefficients; Receive normalized abnormal data, perform ridge regression, and output updated thickness prediction model coefficients to reduce the influence of historical data. Use weighted regression to give higher weight to recent abnormal data based on the exponential decay function. The residuals (mean square error) of the updated model are compared with the historical model. If the mean square error of the updated model decreases (for example, by 10%), the historical model is replaced. The verified models (including the initial thickness prediction model and each updated model) are stored in the edge database according to the timestamp, production recipe, and sensor calibration status. Rollback to the historical model is supported to ensure convergence with the on-site working conditions and reduce thickness prediction errors caused by drift or recipe switching.

[0041] Example 3, see Figure 3 The structure block diagram of the lithium battery separator full process production data transmission optimization system is different from embodiments 1 and 2 in that the embodiment of the present invention provides: a lithium battery separator full process production data transmission optimization system, including: The sensor data acquisition module collects sensor data of the entire process of lithium battery separators through the sensor array and transmits it to the local edge database through the edge node; The abnormal data screening module extracts sensor data to obtain actual process parameters, compares the actual process parameters with the preset process parameters, and outputs abnormal data; The deviation source identification module receives abnormal data and, based on the spatiotemporal topology of the sensors, calculates the spatial-temporal-functional coupling weights between the abnormal data, constructs a structural coupling matrix, and identifies error cluster candidates based on the structural coupling matrix, extracts their spatiotemporal distribution characteristics and abnormal combination patterns; structurally classifies the error cluster candidates using a preset rule engine and outputs a deviation report; An instruction execution module outputs sensor calibration instructions and process fine-tuning instructions based on the deviation report, executes the sensor calibration instructions and process fine-tuning instructions, and monitors the calibration results and process adjustment results; The method of executing process fine-tuning instructions is as follows: constructing a coupling model that integrates temperature, pressure and thickness data, and realizing dynamic switching prediction under multiple production recipe scenarios through integrated learning algorithms; introducing an online residual monitoring mechanism based on a sliding window to realize real-time diagnosis and abnormality location of prediction deviations; generating minimum disturbance process fine-tuning instructions according to the importance and deviation direction of key variables.

[0042] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for optimizing data transmission in the entire production process of lithium battery separators, characterized in that: The following steps are involved: Collect sensor data of the entire process of lithium battery separator through the sensor array; Extract sensor data to obtain actual process parameters, compare the actual process parameters with the preset process parameters, and output abnormal data; Deviation source identification: Receive abnormal data, calculate the spatial-temporal-functional coupling weights between abnormal data based on the spatiotemporal topological structure of sensors, and construct a structural coupling matrix to reflect the correlation between sensor anomalies; Based on the structural coupling matrix, error cluster candidates are identified and their spatiotemporal distribution characteristics and abnormal combination patterns are extracted; Use the preset rule engine to structurally classify error cluster candidates and output deviation reports; Output sensor calibration instructions and process fine-tuning instructions, and monitor calibration results and process adjustment results; The method of executing process fine-tuning instructions is as follows: constructing a coupling model that integrates temperature, pressure and thickness data, and realizing dynamic switching prediction under multiple production recipe scenarios through integrated learning algorithms; introducing an online residual monitoring mechanism based on a sliding window to realize real-time diagnosis and abnormality location of prediction deviations; generating minimum disturbance process fine-tuning instructions according to the importance and deviation direction of key variables.

2. A lithium battery separator full-process production data transmission optimization method according to claim 1, characterized in that: When the cumulative amount or frequency of abnormal data exceeds the limit, an alarm message is issued. Based on the abnormal score of the sensor data and the current network bandwidth status, the network bandwidth is matched to the sensor data to ensure the transmission priority of the abnormal sensor data. Sensor data with high abnormal score values ​​are transmitted first.

3. A lithium battery separator full-process production data transmission optimization method according to claim 3, characterized in that: The abnormality score is quantified by comprehensively calculating the abnormality score based on the degree of deviation between the actual process parameters and the preset process parameters according to the sensor data, the duration of the deviation, and the impact of the deviation on the production quality.

4. The method for optimizing data transmission in the entire production process of a lithium battery separator according to claim 1, characterized in that: The sensor deviation source identification process includes the following steps: Step S101: Data preprocessing and feature extraction: After receiving abnormal data from each sensor data channel, the graph neural network model is input based on the spatiotemporal topological structure of the sensor. The spatial-temporal-functional coupling weights between the abnormal data are calculated based on temporal synchrony, spatial proximity, and functional category similarity to form a structural coupling matrix. Based on the structural coupling matrix, error cluster candidates are screened and obtained; Step S102: Analyze error cluster candidates, extract their spatiotemporal distribution characteristics and abnormal combination patterns, perform structural classification on the error cluster candidates based on a rule engine and a machine learning classification model, and output error labels; Step S103: Output a deviation report including the deviation type and coverage. The deviation types include process-induced errors, recipe-induced errors, and sensor-induced errors. The coverage refers to the time range in which the deviation occurs, the work sections and sensor indexes involved, and the degree of impact of the thickness deviation.

5. The method for optimizing data transmission in the entire production process of a lithium battery separator according to claim 1, characterized in that: The sensor calibration instruction refers to adjusting the gain / offset calibration coefficient of the sensor in the abnormal section; the process fine-tuning instruction refers to adjusting the process parameters in the abnormal section. The method of executing the sensor calibration instruction is: If the sensor's historical drift amplitude exceeds the threshold for a long period of stability or the residual fluctuation before calibration is abnormally severe, it is judged that the sensor performance has deteriorated or failed, and a prompt is given to replace the sensor component. If the replacement conditions are not met, continue the calibration process; Send sensor calibration instructions to the edge node where the corresponding sensor is located. The execution of the sensor calibration instructions includes the calibration target value and the calibration coefficient adjustment strategy. Depending on the sensor type, perform gain calibration or offset calibration. Gain calibration refers to correcting the slope coefficient of the sensor measurement output through linear regression, while offset calibration refers to compensating the measurement value for the baseline offset. Compare the sensor output within a certain time window before and after calibration with the preset standard parameters to calculate the calibration error residual; When the calibration error is lower than the set threshold, the calibration is considered successful and the calibration parameters and their effects are recorded.

6. A lithium battery separator full-process production data transmission optimization method according to claim 5, characterized in that: The way to execute process fine-tuning instructions is: Step S201: Based on the identified error cluster data, using the time-aligned temperature, pressure, and thickness measurement data, a temperature-pressure-thickness coupling model with an adaptive weight adjustment mechanism is constructed. An ensemble learning algorithm is used to fuse ridge regression and decision tree regression to build corresponding sub-models for different production recipe scenarios. Dynamic switching is performed under different production recipe scenarios to improve prediction accuracy under nonlinear working conditions. Step S202: Introducing an online residual monitoring mechanism based on a sliding window to evaluate the residual between the predicted thickness and the actual thickness data in real time. When the residual continuously exceeds the threshold range or shows a deviation trend, triggering abnormal location reasoning, automatically locating the variable causing the residual and its corresponding work section; Step S203: generating a process fine-tuning instruction under the minimum disturbance principle according to the importance ranking and deviation direction of the variables; Step S204: After the correction instruction is issued, the execution segment data is collected in real time and fed back to the temperature-pressure-thickness coupling model to update the model parameters, achieve dynamic convergence of the adaptive working condition evolution, and fine-tune the process parameters based on the predicted thickness; Step S205: The update log and the results of each round of process fine-tuning instructions, thickness improvement, and residual fluctuation changes are all archived in the edge database for subsequent quality backtracking and process stability assessment to improve the system's steady-state operation capability.

7. The method for optimizing data transmission in the entire production process of a lithium battery separator according to claim 1, characterized in that: The method further comprises: Step 5: Collect historical production data covering the temperature in the temperature control zone, pressure in the stretching zone, and thickness data, covering different production recipes and working conditions, analyze the historical production data, use the ridge regression equation, combine cross-validation to optimize the regression coefficient, and establish an initial thickness prediction model; select a drift-free verification data set to test the prediction performance, trigger the thickness prediction model update based on the abnormal data accumulation threshold, test the mean square error of the updated model under actual working conditions, and enable the new model if it meets the standard; otherwise, roll back to the historical version and issue an alarm.

8. A lithium battery separator full-process production data transmission optimization method according to claim 7, characterized in that: The thickness prediction model is constructed as follows: Step S301: collecting historical production data, including temperature data of the heat control zone, pressure data of the stretching zone, and actual thickness data of the downstream thickness gauge; Step S302: Correct the timing offset between the sensors in the thermal control zone and the stretching zone to ensure synchronization of temperature, pressure, and thickness data; remove random noise through sliding average or low-pass filtering; correct sensor drift based on historical calibration records; use the 3σ criterion or median absolute deviation to eliminate some abnormal data to reduce data deviation caused by drift or faults; normalize the temperature, pressure, and thickness data to the range of 0 to 1 to eliminate dimensional differences; Step S303: Using ridge regression to train the ridge regression equation, and using cross-validation to optimize the regression coefficient; training until the loss function meets the requirements, where the loss function is the mean square error between the actual thickness data and the predicted thickness; Step S304: After the training is completed, a set of known drift-free validation data sets is selected to test the prediction performance of the initial thickness prediction model; after the verification is passed, the application is deployed.

9. A lithium battery separator full-process production data transmission optimization method according to claim 8, characterized in that: Based on the accumulation of abnormal data, the thickness prediction model is updated and the updated model is output; the thickness prediction model is updated in the following manner: When the abnormal data accumulates to the threshold, the update of the thickness prediction model is started; Aggregate abnormal data by production batch and perform pre-processing operations; At least including timestamp alignment, denoising and drift correction, and normalization processing; Receive normalized abnormal data, perform ridge regression, and output updated thickness prediction model coefficients to reduce the influence of historical data. Use weighted regression to give higher weight to recent abnormal data based on the exponential decay function. The residuals of the updated model are compared with the historical model. If the mean square error of the updated model is reduced, the historical model is replaced, and the verified model is stored in the edge database according to the timestamp, production recipe and sensor calibration status.

10. A lithium battery separator full-process production data transmission optimization system, used to implement the lithium battery separator full-process production data transmission optimization method according to any one of claims 1 or 7, characterized in that: include: The sensor data acquisition module collects sensor data of the entire process of lithium battery separators through the sensor array and transmits it to the local edge database through the edge node; The abnormal data screening module extracts sensor data to obtain actual process parameters, compares the actual process parameters with the preset process parameters, and outputs abnormal data; The deviation source identification module receives abnormal data, calculates the spatial-temporal-functional coupling weights between abnormal data based on the spatiotemporal topological structure of the sensor, constructs a structural coupling matrix, identifies error cluster candidates based on the structural coupling matrix, and extracts their spatiotemporal distribution characteristics and abnormal combination patterns; Use the preset rule engine to structurally classify error cluster candidates and output deviation reports; An instruction execution module outputs sensor calibration instructions and process fine-tuning instructions based on the deviation report, executes the sensor calibration instructions and process fine-tuning instructions, and monitors the calibration results and process adjustment results; The method of executing process fine-tuning instructions is as follows: constructing a coupling model that integrates temperature, pressure and thickness data, and realizing dynamic switching prediction under multiple production recipe scenarios through integrated learning algorithms; introducing an online residual monitoring mechanism based on a sliding window to realize real-time diagnosis and abnormality location of prediction deviations; generating minimum disturbance process fine-tuning instructions according to the importance and deviation direction of key variables.

Images