Abnormal correlation factor determination method and system for battery cell

Abstract

The invention relates to the technical field of production analysis, and discloses an abnormal correlation factor determination method and system for a battery cell. According to the method, quality detection is carried out on the preset quality detection indexes, the multiple battery cell batches are classified into the target batch and the contrast batch, and data statistics is carried out on the process parameters corresponding to the battery cell batches, so that the parameter distribution data corresponding to the process parameters in the battery cell batches are obtained; the correlation between the process parameters and the preset quality inspection indexes is determined according to the parameter distribution difference between the target batch and the contrast batch, and the process parameters with the correlation are determined as the abnormal correlation factors corresponding to the target batch. The abnormal correlation factors are determined according to the process parameter distribution difference between the normal battery cell batch and the abnormal battery cell batch, and the quality inspection automation of the battery cell batch and the troubleshooting automation of the abnormal correlation factors are realized, so that the troubleshooting efficiency of the abnormal correlation factors is improved, and support is provided for process decision making in time.

Description

Technical Field

[0001] The present invention relates to the field of production analysis technology, and in particular to a method and system for determining abnormality-related factors of battery cells. Background Art

[0002] Against the backdrop of rapid development of the new energy industry, battery cells, as core components of energy storage systems, have performance stability that directly determines the market competitiveness of end products. The production of battery cells involves hundreds of interrelated process parameters in the complete process chain from electrode preparation to cell packaging. These parameters include not only quantifiable processing indicators such as coating speed, roller pressure, and injection volume, but also difficult-to-digitize raw material properties such as particle size distribution of positive and negative electrode materials, electrolyte ion mobility, and membrane porosity fluctuations. These process parameters will have a direct impact on the quality of the process output, leading to abnormal conditions in the battery cells such as increased internal resistance, increased self-discharge rate, and capacity attenuation, ultimately affecting the battery cell capacity.

[0003] Currently, factors affecting battery cell abnormalities are checked through manual experience or threshold monitoring. However, since there are more than a hundred process parameters that affect battery cell capacity, the efficiency of checking abnormality-related factors is low, making it difficult to quickly determine the factors causing battery cell abnormalities and unable to provide strong support for production decisions. Summary of the Invention

[0004] In order to provide a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. The summary is not an extensive review, nor is it intended to identify key / critical elements or delineate the scope of protection of these embodiments, but rather serves as a prelude to the detailed description that follows.

[0005] In view of the above-mentioned shortcomings of the prior art, the present application provides a method and system for determining abnormality-related factors of battery cells to improve the efficiency of troubleshooting abnormality-related factors.

[0006] The present application provides a method for determining abnormality-related factors for battery cells, comprising: obtaining process parameters corresponding to battery cells, wherein one or more battery cells are regarded as a cell batch; performing quality inspection on preset quality inspection indicators corresponding to the cell batch to obtain quality inspection labels, and classifying multiple cell batches according to the quality inspection labels to obtain target batches and control batches, wherein the quality inspection labels corresponding to the target batches are all abnormal labels, and the quality inspection labels corresponding to the control batches are all qualified labels; obtaining parameter distribution data corresponding to the process parameters in each of the cell batches by performing data statistics on the process parameters; performing calculations based on each of the parameter distribution data to obtain parameter distribution differences between the target batch and the control batch, so as to determine the correlation between the process parameters and the preset quality inspection indicators based on the parameter distribution differences, and determining the process parameters with correlation as the abnormality-related factors corresponding to the target batch.

[0007] In one embodiment of the present application, obtaining process parameters corresponding to battery cells includes: producing battery cells through a battery cell production line, wherein the battery cell production line is associated with multiple business systems; in response to any battery cell batch being in a production completion state, obtaining original data corresponding to the battery cell batch by collecting data from each of the business systems; synchronizing the original data corresponding to different business systems to obtain integrated data, and summarizing the integrated data according to different process dimensions to obtain dimensional data corresponding to each of the process dimensions, so as to generate a full life cycle table based on the integrated data and / or the dimensional data; and extracting data from the full life cycle table to obtain one or more process data corresponding to the battery cells.

[0008] In one embodiment of the present application, quality inspection is performed on the preset quality inspection indicators corresponding to the battery cell batches to obtain a quality inspection label, including: obtaining a product data stream, wherein the product data stream is used to carry the preset quality inspection indicators corresponding to each of the battery cell batches in sequence according to the production time; performing data statistics on the preset quality inspection indicators corresponding to the batch set according to the product data stream to obtain quality inspection statistical results corresponding to the sub-batch, wherein the batch set includes one or more battery cell batches with an adjacent relationship, and the sub-batch is any battery cell batch in the batch set; if the quality inspection statistical result meets the preset quality inspection standard, a qualified label is assigned to the sub-batch; if the quality inspection statistical result does not meet the quality inspection standard, an abnormal label is assigned to the sub-batch.

[0009] In one embodiment of the present application, by performing data statistics on the process parameters, parameter distribution data corresponding to the process parameters in each of the battery cell batches are obtained, including at least one of the following: obtaining the target quartiles corresponding to the process parameters in the form of a box plot, and using the target quartiles as parameter distribution data, wherein the target quartiles include the first quartile, the median or the third quartile; obtaining the target quartiles corresponding to the process parameters in the form of a box plot, and using the target quartiles as the interval benchmark to determine the quartile interval, and using the proportion of process parameters within the quartile interval as parameter distribution data; obtaining the parameter extreme values ​​in the process parameters in the form of a line graph, and using the parameter extreme values ​​as parameter distribution data, wherein the parameter extreme values ​​include the parameter maximum value and / or the parameter minimum value; obtaining the standard specification interval in the form of a scatter plot, and using the proportion of process parameters within the standard specification interval as parameter distribution data.

[0010] In one embodiment of the present application, the method further includes: obtaining parameter fluctuations corresponding to the abnormal correlation factors by performing data statistics on the abnormal correlation factors corresponding to the target batch; extracting random variation parts and / or abnormal fluctuation parts from the parameter fluctuations according to the fluctuation trend of the parameter fluctuations; and generating a root cause analysis report corresponding to the target batch based on the abnormal fluctuation parts in the abnormal correlation factors.

[0011] In one embodiment of the present application, the method further includes: taking the abnormal correlation factor corresponding to any battery cell batch and the correlation threshold corresponding to the abnormal correlation factor as historical data, wherein the correlation threshold is used to judge the correlation between the process parameters and the preset quality inspection indicators based on the parameter distribution difference; by performing data statistics on multiple historical data, in order of the frequency of occurrence of the abnormal correlation factors from high to low, determining the high-risk factor from the abnormal correlation factors of the historical data, and adjusting the correlation threshold corresponding to the high-risk factor to obtain a production warning standard; if a battery cell batch in production meets the high-risk factor, the battery cell batch is monitored according to the production warning standard corresponding to the high-risk factor; if it is monitored that the process data of the battery cell batch does not meet the production warning standard, the production of the battery cell batch is stopped.

[0012] In one embodiment of the present application, the method further includes: taking the process parameters corresponding to any battery cell batch as training samples, performing model training based on multiple training samples with quality inspection labels to obtain a classification model; if the number of abnormal correlation factors corresponding to the target batch is multiple, using the classification model to classify the quality inspection labels corresponding to the target batch according to the abnormal correlation factors corresponding to the target batch to obtain a classification result, and by analyzing the classification result, obtaining the contribution degree corresponding to each of the abnormal correlation factors; sorting the abnormal correlation factors corresponding to the target batch according to the contribution degree to obtain a correlation sequence.

[0013] In one embodiment of the present application, model training is performed based on multiple training samples with quality inspection labels to obtain a classification model, including: obtaining a training sample set, wherein the training sample set includes training samples corresponding to qualified labels and abnormal labels respectively; using a distributed queue to execute multiple model training tasks, and using a cross-validation method to improve the model evaluation index output by each of the model training tasks, wherein the model training task includes training a preset tree model based on the training sample set, and adjusting the model parameters of the tree model according to the Bayesian optimization method, and using the trained tree model as an intermediate model; determining a classification model based on the intermediate model output by each of the model training tasks.

[0014] In one embodiment of the present application, the method further includes: obtaining the process dimension corresponding to the abnormality-related factor, the benchmark threshold corresponding to the abnormality-related factor, and the process abnormality score corresponding to the process dimension; calculating according to the contribution value corresponding to the abnormality-related factor and the benchmark deviation value between the abnormality-related factor and the benchmark threshold, to obtain the parameter abnormality level corresponding to the abnormality-related factor, wherein the parameter abnormality level is positively correlated with the contribution value and the benchmark deviation value; adding the parameter abnormality level to the process abnormality score corresponding to the process dimension; if the process abnormality score is greater than or equal to the preset process abnormality threshold, generating a process rectification notice corresponding to the process dimension, and displaying the process rectification notice to the user.

[0015] The present application provides a system for determining abnormality-related factors of battery cells, comprising: an acquisition module for acquiring process parameters corresponding to battery cells, wherein one or more battery cells are regarded as a cell batch; a quality inspection module for performing quality inspection on preset quality inspection indicators corresponding to the cell batches to obtain quality inspection labels, and classifying multiple cell batches according to the quality inspection labels to obtain target batches and control batches, wherein the quality inspection labels corresponding to the target batches are all abnormal labels, and the quality inspection labels corresponding to the control batches are all qualified labels; a statistics module for performing data statistics on the process parameters to obtain parameter distribution data corresponding to the process parameters in each of the cell batches; a determination module for performing calculations based on each of the parameter distribution data to obtain a parameter distribution difference between the target batch and the control batch, so as to determine the correlation between the process parameters and the preset quality inspection indicators based on the parameter distribution difference, and determine the process parameters with correlation as the abnormality-related factors corresponding to the target batch.

[0016] Beneficial effects of this application:

[0017] By performing quality inspections on preset quality inspection indicators, multiple battery cell batches are classified into target batches and control batches, and data statistics are collected on the process parameters corresponding to the battery cell batches to obtain the parameter distribution data corresponding to the process parameters in each battery cell batch. Based on the difference in parameter distribution between the target batch and the control batch, the correlation between the process parameters and the preset quality inspection indicators is determined, and the correlated process parameters are determined as the abnormality-related factors corresponding to the target batch. In this way, compared to using manual experience or establishing threshold monitoring to troubleshoot the factors that cause battery cell abnormalities, the abnormality-related factors are determined based on the difference in process parameter distribution between normal battery cell batches and abnormal battery cell batches, achieving automated quality inspection of battery cell batches and automated troubleshooting of abnormality-related factors, thereby improving the efficiency of troubleshooting abnormality-related factors and providing timely support for process decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a flow chart of a method for determining abnormality-related factors of a battery cell in an embodiment of the present application;

[0019] Figure 2 This is a structural diagram of a system framework for implementing a method for determining abnormality-related factors for a battery cell in an embodiment of the present application;

[0020] Figure 3 This is a flow chart of a method for collecting process data of a battery cell in an embodiment of the present application;

[0021] Figure 4 This is a flow chart of a process data analysis method for a battery cell in an embodiment of the present application;

[0022] Figure 5 It is a structural diagram of a system for determining abnormality-related factors of battery cells in an embodiment of the present application. DETAILED DESCRIPTION

[0023] The following describes the embodiments of the present invention through specific examples. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through different specific embodiments. The details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and sub-samples in the embodiments can be combined with each other unless there is a conflict.

[0024] It should be noted that the illustrations provided in the following embodiments are merely schematic illustrations of the basic concept of the present invention. Therefore, the illustrations only show components related to the present invention and are not drawn according to the number, shape, and size of components in actual implementation. In actual implementation, the type, quantity, and proportion of each component may be changed arbitrarily, and the component layout may also be more complex.

[0025] In the following description, numerous details are discussed to provide a more thorough explanation of the embodiments of the present invention. However, it will be apparent to those skilled in the art that the embodiments of the present invention may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring the embodiments of the present invention.

[0026] The terms "first," "second," and the like in the specification and claims of this application and the accompanying drawings are used to distinguish similar items and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate for the embodiments of the present application described herein. In addition, the terms "including," "having," and any variations thereof are intended to cover non-exclusive inclusions.

[0027] Unless otherwise stated, the term "plurality" means two or more.

[0028] In this application, the character " / " indicates that the preceding and following objects are in an "or" relationship. For example, A / B means: A or B.

[0029] The term "and / or" describes an association between objects, indicating that three relationships can exist. For example, A and / or B means: A or B, or A and B.

[0030] Combine Figure 1As shown, the present application provides a method for determining abnormality-related factors of a battery cell, comprising:

[0031] Step S101, obtaining process parameters corresponding to the battery cell;

[0032] Wherein, one or more battery cells are regarded as a battery cell batch;

[0033] Step S102: Perform quality inspection on the preset quality inspection indicators corresponding to the battery cell batches to obtain quality inspection labels, and classify the multiple battery cell batches according to the quality inspection labels to obtain target batches and reference batches;

[0034] Among them, the quality inspection labels corresponding to the target batches are all abnormal labels, and the quality inspection labels corresponding to the control batches are all qualified labels;

[0035] Step S103, obtaining parameter distribution data corresponding to the process parameters in each battery cell batch by performing data statistics on the process parameters;

[0036] In step S104, calculations are performed based on the parameter distribution data to obtain the parameter distribution differences between the target batch and the control batch, so as to determine the correlation between the process parameters and the preset quality inspection indicators based on the parameter distribution differences, and determine the process parameters with correlation as the abnormal correlation factors corresponding to the target batch.

[0037] The method for determining abnormality-related factors for battery cells provided in this application is used to classify multiple battery cell batches into target batches and control batches by performing quality inspections on preset quality inspection indicators. The process parameters corresponding to the battery cell batches are statistically analyzed to obtain parameter distribution data corresponding to the process parameters in each battery cell batch. The correlation between the process parameters and the preset quality inspection indicators is determined based on the parameter distribution differences between the target batch and the control batch, and the process parameters with correlation are determined as the abnormality-related factors corresponding to the target batch. In this way, compared to manually identifying factors that cause battery cell abnormalities through experience or establishing threshold monitoring, the abnormality-related factors are determined based on the process parameter distribution differences between normal and abnormal battery cell batches, achieving automated quality inspection of battery cell batches and automated screening of abnormality-related factors, thereby improving the efficiency of screening abnormality-related factors and providing timely support for process decision-making.

[0038] Combine Figure 2 As shown, an embodiment of the present disclosure provides a system architecture to implement a method for determining abnormality-related factors of battery cells, wherein the system architecture includes a data source, a data acquisition layer, a data processing layer, a data analysis layer, and an application layer.

[0039] In some embodiments, the data source is composed of multiple business systems, including a manufacturing execution system (MES), a quality management system (QMS), a factory control and monitoring system (FCMS), etc.

[0040] In some embodiments, the manufacturing execution system is a production information management system for the workshop level, responsible for real-time monitoring of production processes, optimizing resource scheduling, and connecting the enterprise planning level and the equipment control level. Among them, the production-related parameters of the battery cells, including the first effect, injection volume, formation current, etc., are collected from the database of the manufacturing execution system.

[0041] In some embodiments, the quality management system is a systematic quality management framework that covers activities such as quality planning, control, assurance, and improvement to ensure that products or services comply with regulations and customer requirements. The raw material parameters of the battery cells are collected from the database of the quality management system, including the gram capacity of the positive electrode main material, graphite batch, and electrolyte expiration date.

[0042] In some embodiments, the factory control and monitoring system is an equipment-level control system for the production site, responsible for real-time collection of equipment data, execution of control instructions, and integration with the manufacturing execution system or enterprise resource planning system to achieve production automation and intelligence. The equipment operating parameters of the battery cells, including the immersion temperature, formation dew point, aging temperature, etc., are collected from the database of the factory control and monitoring system.

[0043] In some embodiments, the data collection layer adopts a distributed synchronization cluster, is deployed in the factory intranet, and is directly connected to the database of each business system. The data collection layer uses ETL (Extract-Transform-Load) tools to synchronize the original data corresponding to multiple business systems to ensure the integrity and timeliness of the original data.

[0044] Optionally, obtaining process parameters corresponding to battery cells includes: producing battery cells through a battery cell production line, wherein the battery cell production line is associated with multiple business systems; in response to any battery cell batch being in a production completion state, obtaining original data corresponding to the battery cell batch by collecting data from each business system; synchronizing the original data corresponding to different business systems to obtain integrated data, and summarizing the integrated data according to different process dimensions to obtain dimensional data corresponding to each process dimension, so as to generate a full life cycle table based on the integrated data and / or dimensional data; and extracting data from the full life cycle table to obtain one or more process data corresponding to the battery cells.

[0045] In some embodiments, the data processing layer adopts a distributed storage cluster (Doris) to process the original data to obtain a full life cycle table of the battery cell batch, wherein the distributed storage cluster includes an operational data storage layer (ODS layer, Operational Data Store), a data warehouse service layer (DWS layer, Data Warehouse Summary), an application data service layer (ADS layer, Application Data Service) and a distributed scheduling cluster (Gschedule).

[0046] In some embodiments, the operational data storage layer is used to store original data, and retain all fields in the original data according to different historical versions.

[0047] In some embodiments, the data warehouse service layer is used to aggregate data according to different process dimensions to obtain dimensional data corresponding to each process dimension to form a dimensional data table, where the process dimensions include coating, injection, formation, etc.

[0048] In some embodiments, the application data service layer is used to construct a wide table, obtain a full life cycle table, and support data query and data analysis, wherein the process parameters stored in the full life cycle table include differentiated discharge capacity, positive electrode slurry solid content, negative electrode surface density, viscosity, temperature, etc.

[0049] In some embodiments, the distributed scheduling cluster is used to manage data tasks such as data cleaning, timestamp alignment, data deduplication, missing value filling, data conversion, data management, and data association.

[0050] In this way, through the collaborative cooperation of data sources, data collection layers, and data processing layers, a multi-dimensional parameter capture framework covering the entire life cycle of battery cells is constructed, integrating multiple process parameters such as raw materials, manufacturing, environment, and equipment, and monitoring a large number of process parameters in real time to avoid parameter omissions and improve the comprehensiveness and accuracy of anomaly detection.

[0051] Combine Figure 3 As shown, the present application provides a method for collecting process data for battery cells, comprising:

[0052] Step S301, collecting data from multiple business systems associated with the battery cell production line to obtain original data corresponding to the battery cell batch;

[0053] Step S302, processing the original data;

[0054] Among them, data processing includes data cleaning, timestamp alignment, data deduplication, missing value filling, data conversion, data management, data association, etc.

[0055] Step S303: Synchronize the original data corresponding to the multiple business systems using ETL tools;

[0056] Step S304: storing the original data, retaining all fields in the original data according to different historical versions;

[0057] Step S305: Aggregate data according to different process dimensions to obtain dimensional data corresponding to each process dimension;

[0058] Step S306: Generate a full lifecycle table in a wide table format based on the integrated data and / or dimensional data;

[0059] Step S307: extract data from the full life cycle table to obtain process data corresponding to the battery cell.

[0060] In some embodiments, the data analysis layer is developed in Python and deployed based on Kubernetes containers, dynamically expanding resources. The data analysis layer includes a real-time error detection engine, a data analysis module, an offline error detection engine, and a contribution value module.

[0061] In some implementations, the real-time error detection engine is provided with customized quality inspection standards, for example, the median of a certain quality inspection indicator shows an upward trend for multiple consecutive batches, the defective rate of a battery cell batch reaches a defective rate threshold, etc., wherein the real-time error detection engine is used to obtain the product data stream from the distributed storage cluster through a Python scheduled task, and processes the product data stream at the minute level according to the quality inspection standards, thereby performing real-time quality inspection on different battery cell batches.

[0062] Optionally, quality inspection is performed on preset quality inspection indicators corresponding to the battery cell batch to obtain a quality inspection label, including: obtaining a product data stream, wherein the product data stream is used to carry the preset quality inspection indicators corresponding to each battery cell batch in sequence according to the production time; performing data statistics on the preset quality inspection indicators corresponding to the batch set according to the product data stream to obtain quality inspection statistical results corresponding to the sub-batch, wherein the batch set includes one or more battery cell batches with an adjacent relationship, and the sub-batch is any battery cell batch in the batch set; if the quality inspection statistical results meet the preset quality inspection standards, a qualified label is assigned to the sub-batch; if the quality inspection statistical results do not meet the quality inspection standards, an abnormal label is assigned to the sub-batch.

[0063] In some embodiments, the preset quality inspection standard includes battery cell capacity.

[0064] In some embodiments, the data analysis module is used to use analysis charts such as box plots, scatter plots, line charts, and normal distribution charts as parameter statistics to detect whether there are parameter distribution anomalies in the process parameters of the battery cell batch, thereby extracting abnormal related factors in the process parameters.

[0065] Optionally, by performing data statistics on the process parameters, parameter distribution data corresponding to the process parameters in each battery cell batch is obtained, including at least one of the following: obtaining the target quartiles corresponding to the process parameters in the form of a box plot, and using the target quartiles as parameter distribution data, wherein the target quartiles include the first quartile, the median or the third quartile; obtaining the target quartiles corresponding to the process parameters in the form of a box plot, and using the target quartiles as the interval benchmark to determine the quartile interval, and using the proportion of process parameters within the quartile interval as parameter distribution data; obtaining the parameter extreme values ​​in the process parameters in the form of a line graph, and using the parameter extreme values ​​as parameter distribution data, wherein the parameter extreme values ​​include the parameter maximum value and / or the parameter minimum value; obtaining the standard specification interval in the form of a scatter plot, and using the proportion of process parameters within the standard specification interval as parameter distribution data.

[0066] In some embodiments, the process parameters corresponding to the target batch are denoted as [X i ], the process parameters corresponding to the control batch are recorded as [X′ i ], where i is the number of battery cells in the battery cell batch; after arranging the process parameters in ascending order and mapping them to the box plot, the process parameter at 25% is taken as the first quartile Q1, the process parameter at 50% is taken as the median Q2, and the process parameter at 75% is taken as the third quartile Q3.

[0067] In some embodiments, the process parameters [X i ] compared with the first quartile Q1 of the process parameter [X′ i] exceeds the first preset threshold value N1, the process parameters of the target batch are determined to be abnormal related factors.

[0068] In some embodiments, the process parameters [X i ] compared with the third quartile Q3 of the process parameter [X′ i ] exceeds the second preset threshold value N2, the process parameters of the target batch are determined to be abnormal related factors.

[0069] In some embodiments, if the target quartile is the first quartile Q1, the quartile interval corresponding to the first quartile is (-∞, Q1-N3×IQR], where N3 is the third preset threshold, IQR is the interquartile range between the first quartile Q1 and the third quartile Q3; the process parameter corresponding to the target batch [X i ], process parameters corresponding to the control batch [X′ i ] respectively have quartile intervals; process parameters [X i ]The data proportion in the corresponding quartile interval is A(Q1), and the process parameter [X′ i ]The data proportion in the corresponding quartile interval is A′(Q1); if the difference between the data proportion A(Q1) and the data proportion A′(Q1) is greater than or equal to the fourth preset threshold N4, the process parameters of the target batch are determined to be abnormal related factors.

[0070] In some embodiments, if the target quartile is the third quartile Q3, the quartile interval corresponding to the first quartile is [Q3+N5×IQR,+∞), where N5 is the fifth preset threshold; similarly, the process parameter corresponding to the target batch [X i ], process parameters corresponding to the control batch [X′ i ] respectively have quartile intervals; process parameters [X i ]The proportion of data in the quartile range is A(Q3), and the process parameter [X′ i ]The data proportion in the quartile interval is A′(Q3); if the difference between the data proportion A(Q3) and the data proportion A′(Q3) is greater than or equal to the sixth preset threshold N6, the process parameters of the target batch are determined to be abnormal related factors.

[0071] In some embodiments, the process parameters corresponding to the target batch [X i ] is denoted as Min[X i ], the process parameters corresponding to the target batch [X i ] is recorded as Max[X i ], the process parameters corresponding to the control batch [X′ i] is recorded as Min[X′ i ], the process parameters corresponding to the control batch [X i The maximum value of the parameter of ] is recorded as Max[X′ i ].

[0072] In some embodiments, a standard specification interval [LSL, USL] corresponding to the process parameters is obtained, where LSL is the lower specification limit and USL is the upper specification limit.

[0073] In some embodiments, if Min[X i ] is less than the lower specification limit LSL, Min[X′ i ] is less than the lower specification limit LSL, and Min[X i ] Compared with Min[X′ i ] exceeds the seventh preset threshold value N7, the process parameters of the target batch are determined to be abnormal related factors.

[0074] In some embodiments, if Max[X i ] is greater than the upper specification limit USL, Max[X′ i ] is greater than the upper specification limit USL, and Max[X i ] Compared with Max[X′ i ] exceeds an eighth preset threshold value N8, the process parameters of the target batch are determined to be abnormal related factors.

[0075] In some embodiments, the process parameters corresponding to the target batch [X i ]The data ratio in the interval (-∞, LSL] is A(LSL), and the process parameters corresponding to the control batch [X′ i ]The data ratio in the interval (-∞,LSL] is A′(LSL). If the difference between the data ratio A(LSL) and the data ratio A′(LSL) is greater than or equal to the ninth preset threshold N9, the process parameters of the target batch are determined to be abnormal related factors.

[0076] In some embodiments, the process parameters corresponding to the target batch [X i ]The data ratio in the interval [USL, +∞) is A(USL), and the process parameters corresponding to the control batch [X′ i ] The data ratio in the interval [USL, +∞) is A′(USL). If the difference between the data ratio A(USL) and the data ratio A′(USL) is greater than or equal to the tenth preset threshold N 10 , then the process parameters of the target batch are determined to be abnormal related factors.

[0077] In some embodiments, the offline anomaly detection engine is used to combine business rules and statistical process control (SPC) to generate a root cause analysis report based on anomaly-related factors.

[0078] Optionally, the method also includes: performing data statistics on the abnormal correlation factors corresponding to the target batch to obtain parameter fluctuations corresponding to the abnormal correlation factors; extracting random variation parts and / or abnormal fluctuation parts from the parameter fluctuations according to the fluctuation trend of the parameter fluctuations; and generating a root cause analysis report corresponding to the target batch based on the abnormal fluctuation parts in the abnormal correlation factors.

[0079] In some embodiments, the contribution value module is used to obtain a classification model through model training, and to interpret a single classification result of the classification model based on the Shapley value (Shapley Value) in game theory to clarify the contribution value of each process parameter to the classification result.

[0080] Optionally, the method also includes: taking the process parameters corresponding to any battery cell batch as training samples, performing model training based on multiple training samples with quality inspection labels to obtain a classification model; if the target batch corresponds to multiple abnormal correlation factors, using the classification model to classify the quality inspection labels corresponding to the target batch according to the abnormal correlation factors corresponding to the target batch to obtain a classification result, and by analyzing the classification result, obtaining the contribution degree corresponding to each abnormal correlation factor; sorting the abnormal correlation factors corresponding to the target batch according to the contribution degree to obtain a correlation sequence.

[0081] In some embodiments, a training sample set is divided into a control group and an abnormal group, wherein the control group is training samples corresponding to qualified labels, and the abnormal group is training samples corresponding to abnormal labels; redundant parameters and necessary parameters are eliminated from the process data of the training samples, and the contribution value of the redundant parameters is set to the minimum value, while the contribution value of the necessary parameters is set to the maximum value, wherein the redundant parameters include vacuum leak rate, etc., and the necessary parameters include infiltration time, etc.; a gradient boosting tree (XGBoost, Extreme Gradient Boosting) is used as the tree model to be trained, which has strong ability to model nonlinear relationships of high-dimensional parameter data; a binary classification problem of qualified labels and abnormal labels is used as the target variable, and the tree model is trained based on the training sample set. During training, the model parameters are adjusted through Bayesian optimization to improve classification accuracy, and cross-validation is used to ensure model generalization; a tree model interpreter (e.g., TreeExplainer tool) is used to interpret the classification results output by the tree model to obtain the SHAP value of each process parameter in a single training sample; and the contribution value of each abnormality-related factor to the classification result is quantified based on the SHAP value corresponding to one or more training samples.

[0082] In this way, the SHAP model is used to analyze the contribution of parameters to the classification results, output an interpretable contribution value ranking, and dynamically adjust the weight coefficient of each abnormality-related factor to the product abnormality based on the contribution value to ensure that the abnormality-related factors with large contributions are more in line with the abnormal product itself, avoid analysis redundancy, improve analysis efficiency and analysis accuracy, and help users quickly understand the cause of the abnormality.

[0083] Optionally, model training is performed based on multiple training samples with quality inspection labels to obtain a classification model, including: obtaining a training sample set, wherein the training sample set includes training samples corresponding to qualified labels and abnormal labels respectively; using a distributed queue to execute multiple model training tasks, and using a cross-validation method to improve the model evaluation indicators output by each model training task, wherein the model training task includes training a preset tree model based on the training sample set, and adjusting the model parameters of the tree model according to the Bayesian optimization method, and using the trained tree model as an intermediate model; determining the classification model based on the intermediate model output by each model training task.

[0084] In some embodiments, data preprocessing is performed on the training sample set, wherein the data preprocessing includes one or more of data standardization, data normalization, missing value filling, outlier removal, data segmentation, etc.; data standardization adopts the Z-score method; data normalization is used to eliminate dimensional differences; missing value filling adopts the interpolation method; outlier presentation adopts the box-and-whisker plot or the 3σ principle; data segmentation divides the training sample set into a training set and a test set according to a preset ratio.

[0085] In some embodiments, the distributed queue uses Celery to asynchronously analyze and compare training tasks; model evaluation indicators include accuracy, F1 score (F1 Score, that is, the harmonic mean of precision and recall), AUC-ROC (Area Under the ROC Curve) curve, etc.

[0086] Combine Figure 4 As shown, the present application provides a process data analysis method for battery cells, comprising:

[0087] Step S401, performing quality inspection on the preset quality inspection indicators corresponding to the battery cell batch to obtain a quality inspection label;

[0088] Step S402, classifying multiple battery cell batches according to the quality inspection labels to obtain target batches and reference batches;

[0089] Step S403, obtaining parameter distribution data corresponding to the process parameters in each battery cell batch by performing data statistics on the process parameters;

[0090] Step S404, performing calculations based on the parameter distribution data to obtain the parameter distribution differences between the target batch and the control batch;

[0091] Step S405, determine whether the parameter distribution difference meets the correlation threshold, if so, jump to step S406, if not, jump to step S410;

[0092] Step S406: Determine whether there is a correlation between the process parameters and the preset quality inspection indicators, determine the process parameters as the abnormal correlation factors corresponding to the target batch, and jump to steps S407 and S408;

[0093] Step S407: Generate a root cause analysis report corresponding to the abnormality-related factors according to statistical process control.

[0094] Step S408: Using the classification model to classify the quality inspection labels corresponding to the target batch according to the abnormality-related factors corresponding to the target batch, a classification result is obtained, and the classification result is analyzed to obtain the contribution degree corresponding to each abnormality-related factor;

[0095] Step S409 : sorting the abnormal correlation factors corresponding to the target batch according to the contribution to obtain a correlation sequence.

[0096] Step S409: Determine whether there is no correlation between the process parameters and the preset quality inspection indicators.

[0097] Optionally, the method also includes: taking the abnormal correlation factor corresponding to any battery cell batch and the correlation threshold corresponding to the abnormal correlation factor as historical data, wherein the correlation threshold is used to judge the correlation between the process parameters and the preset quality inspection indicators based on the difference in parameter distribution; by performing data statistics on multiple historical data, in order of the frequency of occurrence of the abnormal correlation factors from high to low, determining the high-risk factor from the abnormal correlation factors of the historical data, and adjusting the correlation threshold corresponding to the high-risk factor to obtain the production warning standard; if the battery cell batch in production meets the high-risk factor, the battery cell batch is monitored according to the production warning standard corresponding to the high-risk factor; if the process data of the battery cell batch is monitored to not meet the production warning standard, the production of the battery cell batch is stopped.

[0098] In some embodiments, after the abnormal correlation factors of each batch of battery cells are detected, the batch identification, quality inspection label, abnormal correlation factors, correlation thresholds triggered by the abnormal correlation factors, one or more abnormal correlation factors with the largest contribution value, etc. are stored as historical data in a historical database; by performing cluster analysis on the historical data in the historical database, the statistical distribution of the correlation thresholds is calculated to form a standard library for judging abnormalities for special process parameters.

[0099] In some embodiments, a batch of battery capacitors contains 3,000 battery cells, and the abnormal correlation factor is the gram capacity of the positive electrode main material of the raw materials provided by supplier A; if the abnormal correlation factor appears continuously in multiple batches of battery cells, the abnormal correlation factor is determined to be a high-risk factor; if the battery cells in production use batteries with raw materials provided by supplier A, the box plot abnormality threshold of the gram capacity of the positive electrode main material is lowered from 0.5 to 0.3, wherein the box plot abnormality threshold includes a third preset threshold, a fourth preset threshold, a fifth preset threshold and a sixth preset threshold; if the gram capacity of the positive electrode main material of the battery cell triggers the box plot abnormality threshold, the battery cell batch is intercepted, reducing 3,000 abnormal products.

[0100] In this way, by deeply mining historical data, we can support root cause tracing and knowledge accumulation, ensure that abnormal products are discovered in the early stages of the process, reduce ineffective processing and resource waste in subsequent processes, and provide data support for process optimization.

[0101] In some embodiments, the application layer includes a user center, analysis task management, predictive analysis, and a visual dashboard; the user center is used for user management, process route configuration, and rule group configuration settings; the analysis task management is used to create troubleshooting tasks for abnormal related factors and track the task status of the troubleshooting tasks; predictive analysis is used to output capacity deviation prediction results and root cause analysis reports; the visual dashboard uses the Vue.js framework to display parameter trends, warning information, and root cause maps to users.

[0102] Optionally, the method also includes: obtaining the process dimension corresponding to the abnormality-related factor, the benchmark threshold corresponding to the abnormality-related factor, and the process anomaly score corresponding to the process dimension; calculating based on the contribution value corresponding to the abnormality-related factor and the benchmark deviation value between the abnormality-related factor and the benchmark threshold, to obtain the parameter anomaly level corresponding to the abnormality-related factor, wherein the parameter anomaly level is positively correlated with the contribution value and the benchmark deviation value; adding the parameter anomaly level to the process anomaly score corresponding to the process dimension; if the process anomaly score is greater than or equal to the preset process anomaly threshold, generating a process rectification notice corresponding to the process dimension, and displaying the process rectification notice to the user.

[0103] In some embodiments, the abnormality-related factors include the injection volume in the injection process; the contribution value of the injection volume is 0.7; the median injection volume of the battery cells in this batch is 5, and the corresponding benchmark threshold is 2, then the benchmark deviation value of the injection volume is 3; according to the contribution value and the benchmark offset, the parameter abnormality level is calculated to be 2; the parameter abnormality level 2 is added to the process abnormality score of the injection process; if the process abnormality score of the injection process is greater than or equal to the process abnormality threshold 8, the injection process needs to be rectified and a warning is given to the user.

[0104] Combine Figure 5 As shown, the present application provides a system for determining abnormality-related factors of battery cells, including an acquisition module 501 , a quality inspection module 502 , a statistics module 503 and a determination module 504 .

[0105] The acquisition module 501 is used to acquire process parameters corresponding to battery cells, wherein one or more battery cells are regarded as a battery cell batch.

[0106] The quality inspection module 502 is used to perform quality inspection on the preset quality inspection indicators corresponding to the battery cell batches, obtain quality inspection labels, and classify multiple battery cell batches according to the quality inspection labels to obtain target batches and control batches, among which the quality inspection labels corresponding to the target batches are all abnormal labels, and the quality inspection labels corresponding to the control batches are all qualified labels.

[0107] The statistical module 503 is used to obtain parameter distribution data corresponding to each battery cell batch by performing data statistics on the process parameters.

[0108] The determination module 504 is used to perform calculations based on the parameter distribution data to obtain the parameter distribution differences between the target batch and the control batch, so as to determine the correlation between the process parameters and the preset quality inspection indicators based on the parameter distribution differences, and determine the process parameters with correlation as the abnormal correlation factors corresponding to the target batch.

[0109] The abnormality-related factor determination system for battery cells provided by the present application is used to classify multiple battery cell batches into target batches and control batches by performing quality inspections on preset quality inspection indicators. The process parameters corresponding to the battery cell batches are statistically analyzed to obtain parameter distribution data corresponding to the process parameters in each battery cell batch. The correlation between the process parameters and the preset quality inspection indicators is determined based on the parameter distribution differences between the target batches and the control batches, and the process parameters with correlation are determined as the abnormality-related factors corresponding to the target batches. In this way, compared to manually identifying factors that cause battery cell abnormalities through experience or establishing threshold monitoring, the abnormality-related factors are determined based on the process parameter distribution differences between normal battery cell batches and abnormal battery cell batches, achieving automated quality inspection of battery cell batches and automated screening of abnormality-related factors, thereby improving the efficiency of screening of abnormality-related factors and providing timely support for process decision-making.

[0110] The above description and the accompanying drawings fully illustrate the embodiments of the present disclosure to enable those skilled in the art to practice them. Other embodiments may include structural, logical, electrical, process and other changes. The embodiments represent only possible variations. Unless expressly required, individual components and functions are optional, and the order of operations may vary. Parts and subsamples of some embodiments may be included in or replace parts and subsamples of other embodiments. Moreover, the terms used in this application are only used to describe the embodiments and are not used to limit the claims. As used in the description of the embodiments and claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to also include plural forms. Similarly, the term "and / or" as used in this application refers to any and all possible combinations of one or more associated listings. In addition, when used in this application, the term "comprise" and its variations "comprises" and / or comprising refer to the presence of a stated subsample, whole, step, operation, element, and / or component, but do not exclude the presence or addition of one or more other subsamples, wholes, steps, operations, elements, components and / or groups of these. In the absence of further restrictions, an element defined by the statement "comprises a..." does not exclude the presence of other identical elements in the process, method or device that includes the element. In this article, each embodiment may focus on the differences from other embodiments, and the same and similar parts between the various embodiments can be referenced to each other. For the methods, products, etc. disclosed in the embodiments, if they correspond to the method part disclosed in the embodiments, then the relevant parts can be found in the description of the method part.

[0111] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software may depend on the specific application and design constraints of the technical solution. Technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application. Technicians can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0112] In the embodiments disclosed herein, the disclosed methods and products (including but not limited to devices, equipment, etc.) can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units can be merely a logical functional division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some sub-samples can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between each other shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, and can be electrical, mechanical or other forms. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to implement this embodiment. In addition, the functional units in this application may be integrated into a processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0113] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of the systems, methods, and computer program products of the present application. In this regard, each block in a flowchart or block diagram may represent a module, program segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in an order different from that marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, or they may sometimes be executed in the opposite order, depending on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different blocks may also occur in an order different from that disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps may actually be executed substantially in parallel, or they may sometimes be executed in the opposite order, depending on the functions involved. Each block in the block diagram and / or flowchart, as well as combinations of blocks in the block diagram and / or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or action, or may be implemented using a combination of dedicated hardware and computer instructions.

Claims

1. A method for determining abnormality-related factors of a battery cell, characterized in that: include: Obtaining process parameters corresponding to battery cells, wherein one or more battery cells are considered a battery cell batch; Performing quality inspection on preset quality inspection indicators corresponding to the battery cell batches to obtain quality inspection labels, and classifying multiple battery cell batches according to the quality inspection labels to obtain target batches and control batches, wherein the quality inspection labels corresponding to the target batches are all abnormal labels, and the quality inspection labels corresponding to the control batches are all qualified labels; By performing data statistics on the process parameters, parameter distribution data corresponding to the process parameters in each of the battery cell batches are obtained; Calculations are performed based on the parameter distribution data to obtain the parameter distribution differences between the target batch and the control batch, so as to determine the correlation between the process parameters and the preset quality inspection indicators based on the parameter distribution differences, and determine the process parameters with correlation as the abnormal correlation factors corresponding to the target batch.

2. The method according to claim 1, characterized in that Obtain the process parameters corresponding to the battery cell, including: Producing battery cells through a cell production line, wherein the cell production line is associated with multiple business systems; In response to any battery cell batch being in a production completion state, acquiring raw data corresponding to the battery cell batch by collecting data from each of the business systems; Synchronize the original data corresponding to different business systems to obtain integrated data, and summarize the integrated data according to different process dimensions to obtain dimensional data corresponding to each process dimension, so as to generate a full life cycle table based on the integrated data and / or the dimensional data; Data is extracted from the full life cycle table to obtain one or more process data corresponding to the battery cell.

3. The method according to claim 1, characterized in that Perform quality inspection on the preset quality inspection indicators corresponding to the battery cell batch to obtain a quality inspection label, including: Obtaining a product data stream, wherein the product data stream is used to carry preset quality inspection indicators corresponding to each of the battery cell batches in sequence according to production time; Performing data statistics on preset quality inspection indicators corresponding to a batch set according to the product data stream to obtain quality inspection statistical results corresponding to a sub-batch, wherein the batch set includes one or more battery cell batches having an adjacent relationship, and the sub-batch is any battery cell batch in the batch set; If the quality inspection statistical results meet the preset quality inspection standards, a qualified label is assigned to the sub-batch; If the quality inspection statistical result does not meet the quality inspection standard, an abnormal label is assigned to the sub-batch.

4. The method according to claim 1, wherein By performing data statistics on the process parameters, parameter distribution data corresponding to the process parameters in each of the battery cell batches is obtained, including at least one of the following: Obtaining target quartiles corresponding to the process parameters in a box plot format, and using the target quartiles as parameter distribution data, wherein the target quartiles include the first quartile, the median, or the third quartile; Obtain the target quartiles corresponding to the process parameters in a box plot format, and use the target quartiles as interval benchmarks to determine quartile intervals, and use the proportion of process parameters within the quartile intervals as parameter distribution data; Obtaining extreme values ​​of the process parameters in a broken line graph, and using the extreme values ​​as parameter distribution data, wherein the extreme values ​​include a maximum value and / or a minimum value; The standard specification interval is obtained in the form of a scatter plot, and the proportion of process parameters within the standard specification interval is used as parameter distribution data.

5. The method according to claim 1, wherein The method further comprises: By performing data statistics on the abnormal correlation factors corresponding to the target batch, parameter fluctuations corresponding to the abnormal correlation factors are obtained; extracting a random variation portion and / or an abnormal fluctuation portion from the parameter fluctuation according to a fluctuation trend of the parameter fluctuation; A root cause analysis report corresponding to the target batch is generated based on the abnormal fluctuation part of the abnormal correlation factor.

6. The method according to claim 1, wherein The method further comprises: The abnormal correlation factor corresponding to any battery cell batch and the correlation threshold corresponding to the abnormal correlation factor are used as historical data, wherein the correlation threshold is used to determine the correlation between the process parameters and the preset quality inspection indicators based on the parameter distribution difference; By performing data statistics on a plurality of historical data, high-risk factors are determined from the abnormal correlation factors of the historical data in descending order of the frequency of occurrence of the abnormal correlation factors, and the correlation threshold corresponding to the high-risk factors is adjusted to obtain a production early warning standard; If a cell batch in production meets a high-risk factor, the cell batch is monitored according to the production warning standard corresponding to the high-risk factor; If it is monitored that the process data of the battery cell batch does not meet the production warning standard, the production of the battery cell batch is stopped.

7. The method according to any one of claims 1 to 6, characterized in that The method further comprises: The process parameters corresponding to any battery cell batch are used as training samples, and the model is trained based on multiple training samples with quality inspection labels to obtain a classification model; If the target batch has multiple abnormal correlation factors, the classification model is used to classify the quality inspection labels corresponding to the target batch according to the abnormal correlation factors corresponding to the target batch to obtain a classification result, and the classification result is analyzed to obtain the contribution degree corresponding to each abnormal correlation factor; The abnormal correlation factors corresponding to the target batch are sorted according to the contribution to obtain a correlation sequence.

8. The method according to claim 7, characterized in that The model is trained based on multiple training samples with quality inspection labels to obtain a classification model, including: Obtaining a training sample set, wherein the training sample set includes training samples corresponding to qualified labels and abnormal labels respectively; Utilizing a distributed queue to execute multiple model training tasks, and employing a cross-validation method to improve the model evaluation index output by each of the model training tasks, wherein the model training task includes training a preset tree model based on the training sample set, adjusting the model parameters of the tree model according to the Bayesian optimization method, and using the trained tree model as an intermediate model; The classification model is determined based on the intermediate model output by each of the model training tasks.

9. The method according to claim 7, characterized in that The method further comprises: Obtaining a process dimension corresponding to an abnormality-related factor, a benchmark threshold corresponding to the abnormality-related factor, and a process abnormality score corresponding to the process dimension; Calculating according to the contribution value corresponding to the abnormal correlation factor and the reference deviation value between the abnormal correlation factor and the reference threshold value, to obtain the parameter abnormality level corresponding to the abnormal correlation factor, wherein the parameter abnormality level is positively correlated with the contribution value and the reference deviation value; Adding the parameter abnormality level to the process abnormality score corresponding to the process dimension; If the process anomaly score is greater than or equal to a preset process anomaly threshold, a process rectification notice corresponding to the process dimension is generated and displayed to the user.

10. A system for determining abnormality-related factors of battery cells, characterized in that: include: an acquisition module, configured to acquire process parameters corresponding to battery cells, wherein one or more battery cells are considered a battery cell batch; a quality inspection module, configured to perform quality inspection on preset quality inspection indicators corresponding to the battery cell batches, obtain quality inspection labels, and classify the multiple battery cell batches according to the quality inspection labels to obtain target batches and control batches, wherein the quality inspection labels corresponding to the target batches are all abnormal labels, and the quality inspection labels corresponding to the control batches are all qualified labels; A statistical module, configured to obtain parameter distribution data corresponding to each of the battery cell batches by performing data statistics on the process parameters; A determination module is used to perform calculations based on the parameter distribution data to obtain the parameter distribution differences between the target batch and the control batch, so as to determine the correlation between the process parameters and the preset quality inspection indicators based on the parameter distribution differences, and determine the process parameters with correlation as the abnormal correlation factors corresponding to the target batch.

Images