Method and device for detecting battery

Abstract

A method and apparatus for detecting a battery. The method comprises: acquiring formation process data of each of N batteries to be detected (210); according to the formation process data of each to-be-detected battery and an anomaly detection model, obtaining anomaly index data of each to-be-detected battery (220), the anomaly detection model being obtained by training according to the formation process data of the M detected batteries, and according to the anomaly index data of the N to-be-detected batteries, detecting whether there is an anomaly in the N to-be-detected batteries (230).

Description

Method and device for detecting battery Technical Field

[0001] The present application relates to the field of battery testing, and in particular, to a method and apparatus for testing a battery. Background Art

[0002] If anomalies occur during the battery production process, such as incorrect electrolyte injection or impurities in the electrode, these problems may not be apparent in the test section of the production line, but may lead to lithium deposition, low capacity, and cycle drop after repeated use. The loading of these abnormal batteries will affect the rate and cycle performance of the battery system.

[0003] Therefore, it is desired to provide a method for detecting batteries so as to promptly and conveniently detect battery abnormalities and handle them.

[0004] Summary of the Invention

[0005] The present application provides a method and device for detecting a battery, which can detect battery anomalies.

[0006] In a first aspect, a method for detecting batteries is provided, comprising: obtaining formation process data of each of N batteries to be tested, where N is a positive integer greater than 1; obtaining abnormal index data of each battery to be tested based on the formation process data of each battery to be tested and an abnormality detection model, where the abnormality detection model is trained based on the formation process data of M tested batteries, where M is a positive integer greater than 1; and detecting whether the N batteries to be tested have abnormalities based on the abnormal index data of the N batteries to be tested.

[0007] In this embodiment, by obtaining the formation process data of N batteries to be tested, and obtaining the abnormal index data of the N batteries to be tested based on the formation process data of the N batteries to be tested and an abnormality detection model, it is detected whether the N batteries to be tested have abnormalities based on the abnormal index data of the N batteries to be tested, which is conducive to realizing batch detection of batteries.

[0008] In one possible implementation, based on the abnormal indicator data of the N batteries to be tested, detecting whether the N batteries to be tested have abnormalities includes: generating time series data based on the abnormal indicator data of the N batteries to be tested and the abnormal indicator data of P tested batteries, where P is a positive integer greater than 1; and detecting whether the N batteries to be tested have abnormalities based on the time series data.

[0009] In this embodiment, based on the time series data generated from the abnormal indicator data of N batteries to be tested and the abnormal indicator data of P tested batteries, it is detected whether the N batteries to be tested have abnormalities, which is conducive to batch detection of batteries.

[0010] In one possible implementation, time series data is generated based on abnormal indicator data of N batteries to be tested and abnormal indicator data of P tested batteries, including: obtaining first abnormal indicator data from the abnormal indicator data of the N batteries to be tested; obtaining at least one second abnormal indicator data from the abnormal indicator data of the P tested batteries; and generating time series data based on the first abnormal indicator data and the at least one second abnormal indicator data.

[0011] In this embodiment, time series data is generated based on the first abnormal indicator data obtained from the abnormal indicator data of N batteries to be tested and at least one second abnormal indicator data obtained from the abnormal indicator data of P tested batteries, rather than directly based on all the abnormal indicator data of the N batteries to be tested and the P tested batteries. This is beneficial to reducing the amount of data in the time series data, thereby reducing the complexity of battery detection.

[0012] In one possible implementation, abnormal indicator data of N batteries to be tested are obtained within a first time interval, and abnormal indicator data of P tested batteries are obtained within a second time interval, the second time interval includes at least one time series sequence, and obtaining first abnormal indicator data from the abnormal indicator data of the N batteries to be tested includes: obtaining first abnormal indicator data from the first time interval; obtaining at least one second abnormal indicator data from the abnormal indicator data of the P tested batteries includes: obtaining one second abnormal indicator data from each time series sequence in at least one time series sequence.

[0013] In this embodiment, obtaining first abnormal indicator data from a first time interval and obtaining second abnormal indicator data from each time series in at least one time series sequence is beneficial for making the time series data include abnormal indicator data uniformly distributed in the time dimension, thereby facilitating improving the accuracy of battery detection.

[0014] In one possible implementation, the second time interval is divided into at least one time series according to a preset time length, and the first abnormal indicator data is obtained from the first time interval, including: within the first time interval, statistics are performed on the abnormal indicator data of N batteries to be tested; according to a preset statistical quantile, the first abnormal indicator data is obtained from the first time interval; and a second abnormal indicator data is obtained from each time series in the at least one time series, including: within each time series, statistics are performed on the abnormal indicator data of the tested battery corresponding to each time series; according to the preset statistical quantile, a second abnormal indicator data is obtained from each time series.

[0015] In this embodiment, an abnormal indicator data is obtained from the first time interval and a second abnormal indicator data is obtained from each time series sequence in the second time interval according to the same preset statistical quantile, so that the time series data includes abnormal indicator data that is uniformly distributed in the time dimension, thereby improving the accuracy of battery detection.

[0016] In one possible implementation, obtaining first abnormal indicator data from a first time interval includes: when N is greater than a first preset threshold, obtaining the first abnormal indicator data from the first time interval; obtaining a second abnormal indicator data from each time series in at least one time series, including: when the number of measured batteries corresponding to each time series is greater than the first preset threshold, obtaining a second abnormal indicator data from each time series.

[0017] In this embodiment, when N is greater than the first preset threshold, the first abnormal indicator data is obtained within the first time interval, and when the number of measured batteries corresponding to each time series is greater than the first preset threshold, the second abnormal indicator data is obtained within each time series. This can reduce the probability of obtaining abnormal indicator data with large deviations, thereby improving the accuracy of battery detection.

[0018] In one possible implementation, based on abnormal indicator data of N batteries to be tested, detecting whether N batteries to be tested have abnormalities includes: determining labels of the N batteries to be tested based on the abnormal indicator data of the N batteries to be tested, where the labels are abnormal labels for indicating that the batteries have abnormalities or normal labels for indicating that the batteries do not have abnormalities; generating time series data based on the labels of the N batteries to be tested and the labels of P tested batteries, where P is a positive integer greater than 1; and detecting whether the N batteries to be tested have abnormalities based on the time series data.

[0019] In this embodiment, based on the time series data generated from the tags of N batteries to be tested and the tags of P tested batteries, it is detected whether the N batteries to be tested have abnormalities, which is conducive to batch detection of batteries.

[0020] In one possible implementation, the labels of N batteries to be tested are obtained within a first time interval, and the labels of P tested batteries are obtained within a second time interval. The second time interval is divided into at least one time series according to a preset time length. Time series data is generated based on the labels of the N batteries to be tested and the labels of the P tested batteries, including: within the first time interval, counting the number of abnormal labels in the labels of the N batteries to be tested; within each time series in at least one time series, counting the number of abnormal labels in the labels of the tested batteries corresponding to each time series; and generating time series data based on the number of abnormal labels counted within the first time interval and the number of abnormal labels counted within each time series.

[0021] In one possible implementation, the labels of N batteries to be tested are obtained within a first time interval, and the labels of P tested batteries are obtained within a second time interval. The second time interval is divided into at least one time series according to a preset time length. Time series data is generated based on the labels of the N batteries to be tested and the labels of the P tested batteries, including: calculating the proportion of abnormal labels in the labels of the N batteries to be tested within the first time interval; calculating the proportion of abnormal labels in the labels of the tested batteries in each time series in at least one time series; and generating time series data based on the proportion of abnormal labels in the first time interval and the proportion of abnormal labels in each time series.

[0022] In one possible implementation, abnormal index data of each battery to be tested is obtained based on the formation process data and an anomaly detection model of each battery to be tested, including: respectively obtaining the film formation key segment data of each battery to be tested from the formation process data of each battery to be tested, the film formation key segment data refers to all formation process data of the battery between zero and a second preset threshold value during the formation process, and the second preset threshold value is related to the reaction potential of the additive expected to be added to the electrolyte of the N batteries to be tested; obtaining the abnormal index data of each battery to be tested based on the film formation key segment data and the anomaly detection model of each battery to be tested, and the anomaly detection model is obtained by training based on the film formation key segment data of M tested batteries.

[0023] In this embodiment, by acquiring data of key film-forming sections, the anomaly detection model can be focused on these areas, preventing the effect of anomaly detection from being affected by factors such as battery polarization and capacity differences in the later stage. In addition, the amount of data storage can be effectively reduced.

[0024] In one possible implementation, abnormal index data of each battery to be tested is obtained based on the key film-forming segment data and the abnormality detection model of each battery to be tested, including: using a sliding window method to obtain a set of formation process data with the highest similarity to the standard film-forming segment curve from the key film-forming segment data of each battery to be tested, the standard film-forming segment curve is formed by fitting based on the key film-forming segment data of L tested batteries, where L is a positive integer greater than 1; based on a set of formation process data and the abnormality detection model of each battery to be tested, the abnormal index data of each battery to be tested is obtained.

[0025] In this embodiment, a sliding window method is used to obtain a set of formation process data with the highest similarity to the standard film formation curve from the key film formation section data of the battery to be tested, which can further reduce the data storage amount.

[0026] In one possible implementation, abnormal index data of each battery to be tested is obtained based on a set of formation process data of each battery to be tested and an abnormality detection model, including: downsampling a set of formation process data of each battery to be tested and generating downsampled data of each battery to be tested; inputting the downsampled data of each battery to be tested into the abnormality detection model, and outputting the abnormal index data of each battery to be tested, the abnormality detection model is trained based on the downsampled data of M tested batteries, and the standard film formation section curve is formed by fitting based on the downsampled data of L tested batteries.

[0027] In this embodiment, downsampling is performed on a set of formation data input to the anomaly detection model, which can continue to reduce the amount of data storage and the consumption of computing resources while maintaining the ability to detect batteries.

[0028] In one possible implementation, the downsampled data of the i-th battery to be tested among N batteries to be tested and the downsampled data of each tested battery among M tested batteries both include Q data, the Q data of the i-th battery to be tested corresponding to the Q data of each tested battery one-to-one, i is any positive integer from 1 to N, and Q is a positive integer greater than 1. The downsampled data of each battery to be tested is input into an anomaly detection model, and the anomaly indicator data of each battery to be tested is output, including: calculating M distances between the j-th data among the Q data of the i-th battery to be tested and the corresponding data among the Q data of each tested battery, where j is any positive integer from 1 to Q; determining the K data with the closest distance from the M data corresponding to the j-th data, where K is a positive integer; determining the average distance of the j-th data based on the distances from the K data to the j-th data; and determining the anomaly indicator data of the i-th battery to be tested based on the average distance between each data among the Q data of the i-th battery to be tested and each data.

[0029] In one possible implementation, detecting whether N batteries to be tested have abnormalities based on time series data includes: performing change point detection on the time series data to obtain a change point detection result; and determining whether the N batteries to be tested have abnormalities based on the change point detection result.

[0030] In this embodiment, by performing change point detection on the time series data, it is possible to effectively detect whether an abnormality occurs in the battery to be tested.

[0031] In one possible implementation, change point detection is performed on the time series data to obtain a change point detection result, including: obtaining a prior probability distribution based on the data of the measured batteries in the time series data; obtaining a new probability distribution based on the data of the batteries to be tested in the time series data and the data of the measured batteries in the time series data; and obtaining a change point detection result based on the prior probability distribution and the new probability distribution.

[0032] In one possible implementation, a change point detection result is obtained based on the prior probability distribution and the new probability distribution, including: when the prior probability distribution and the new probability distribution are different, determining that the change point detection result is that the data of the battery to be tested in the time series data has a mutation compared to the data of the tested battery.

[0033] In one possible implementation, change point detection is performed on the time series data to obtain a change point detection result, including: respectively calculating the difference between two adjacent data in the time series data; and obtaining the change point detection result based on the relationship between the difference and a third preset threshold.

[0034] In one possible implementation, the method further includes: when an abnormality is detected in N batteries to be tested, obtaining data related to the electrolyte injection process when producing the N batteries to be tested; determining whether there is a mismatch between the data related to the electrolyte injection process and a preset control range; and when there is a mismatch between the data related to the electrolyte injection process and the preset control range, determining that an electrolyte injection abnormality has occurred in the electrolyte of the N batteries to be tested.

[0035] In a possible implementation, the method further includes: generating electrolyte abnormality cause information of the N batteries to be tested when it is determined that electrolyte injection abnormality occurs in the N batteries to be tested.

[0036] In a possible implementation, the method further includes: generating alarm information according to the electrolyte abnormality cause information to intercept the N batteries to be tested.

[0037] In a second aspect, a device for detecting batteries is provided, comprising: an acquisition module for acquiring formation process data of each of N batteries to be tested, where N is a positive integer greater than 1; the acquisition module is further configured to acquire abnormal index data of each battery to be tested based on the formation process data of each battery to be tested and an abnormality detection model, wherein the abnormality detection model is trained based on the formation process data of M tested batteries, where M is a positive integer greater than 1; and a detection module for detecting whether the N batteries to be tested have abnormalities based on the abnormal index data of the N batteries to be tested.

[0038] In one possible implementation, the detection module includes: a generation submodule for generating time series data based on abnormal indicator data of N batteries to be tested and abnormal indicator data of P tested batteries, where P is a positive integer greater than 1; and a detection submodule for detecting whether there are any abnormalities in the N batteries to be tested based on the time series data.

[0039] In one possible implementation, the acquisition module is further used to: obtain first abnormal indicator data from the abnormal indicator data of N batteries to be tested; obtain at least one second abnormal indicator data from the abnormal indicator data of P tested batteries; and the generation submodule is specifically used to: generate time series data based on the first abnormal indicator data and at least one second abnormal indicator data.

[0040] In one possible implementation, abnormal indicator data of N batteries to be tested are obtained within a first time interval, and abnormal indicator data of P tested batteries are obtained within a second time interval. The second time interval includes at least one time series. The acquisition module is specifically used to: obtain first abnormal indicator data from the first time interval; and obtain a second abnormal indicator data from each time series in at least one time series.

[0041] In one possible implementation, the second time interval is divided into at least one time series according to a preset time length, and the acquisition module is specifically used to: within the first time interval, count the abnormal indicator data of N batteries to be tested; according to the preset statistical quantile, obtain the first abnormal indicator data from the first time interval; within each time series, count the abnormal indicator data of the tested battery corresponding to each time series; according to the preset statistical quantile, obtain a second abnormal indicator data from each time series.

[0042] In one possible implementation, the acquisition module is specifically used to: acquire first abnormal indicator data from a first time interval when N is greater than a first preset threshold; and acquire a second abnormal indicator data from each time series when the number of measured batteries corresponding to each time series is greater than the first preset threshold.

[0043] In one possible implementation, the detection module includes: a determination submodule for determining labels of N batteries to be tested based on abnormal indicator data of the N batteries to be tested, where the labels are abnormal labels indicating that the batteries are abnormal or normal labels indicating that the batteries are not abnormal; a generation submodule for generating time series data based on the labels of the N batteries to be tested and the labels of P tested batteries, where P is a positive integer greater than 1; and a detection submodule for detecting whether the N batteries to be tested are abnormal based on the time series data.

[0044] In one possible implementation, the labels of N batteries to be tested are obtained within a first time interval, and the labels of P tested batteries are obtained within a second time interval. The second time interval is divided into at least one time series according to a preset time length. The generation submodule is specifically used to: count the number of abnormal labels in the labels of the N batteries to be tested within the first time interval; count the number of abnormal labels in the labels of the tested batteries corresponding to each time series within each time series; and generate time series data based on the number of abnormal labels counted within the first time interval and the number of abnormal labels counted within each time series.

[0045] In one possible implementation, the labels of N batteries to be tested are obtained within a first time interval, and the labels of P tested batteries are obtained within a second time interval. The second time interval is divided into at least one time series according to a preset time length. The generation submodule is specifically used to: calculate the proportion of abnormal labels in the labels of the N batteries to be tested within the first time interval; calculate the proportion of abnormal labels in the labels of the tested batteries in each time series in at least one time series; and generate time series data based on the proportion of abnormal labels in the first time interval and the proportion of abnormal labels in each time series.

[0046] In one possible implementation, the acquisition module is specifically used to: obtain the key film-forming segment data of each battery to be tested from the formation process data of each battery to be tested, where the key film-forming segment data refers to all formation process data of the battery whose voltage ranges from zero to a second preset threshold value during the formation process, and the second preset threshold value is related to the reaction potential of the additive expected to be added to the electrolyte of the N batteries to be tested; obtain the abnormal index data of each battery to be tested based on the key film-forming segment data of each battery to be tested and an abnormality detection model, where the abnormality detection model is trained based on the key film-forming segment data of M tested batteries.

[0047] In one possible implementation, the acquisition module is specifically used to: use a sliding window method to obtain a set of formation process data with the highest similarity to the standard film formation section curve from the key film formation section data of each battery to be tested, where the standard film formation section curve is formed by fitting the key film formation section data of L tested batteries, where L is a positive integer greater than 1; and obtain abnormal index data of each battery to be tested based on a set of formation process data and an anomaly detection model of each battery to be tested.

[0048] In one possible implementation, the acquisition module is specifically used to: downsample a set of formation process data of each battery to be tested and generate downsampled data for each battery to be tested; input the downsampled data of each battery to be tested into an anomaly detection model, and output the abnormal indicator data of each battery to be tested. The anomaly detection model is trained based on the downsampled data of M tested batteries, and the standard film formation section curve is formed by fitting based on the downsampled data of L tested batteries.

[0049] In one possible implementation, the downsampled data of the i-th battery to be tested among N batteries to be tested and the downsampled data of each tested battery among M tested batteries both include Q data, the Q data of the i-th battery to be tested correspond one-to-one to the Q data of each tested battery, i is any positive integer from 1 to N, and Q is a positive integer greater than 1, and the acquisition module is specifically used to: calculate M distances between the j-th data among the Q data of the i-th battery to be tested and the corresponding data among the Q data of each tested battery, j is any positive integer from 1 to Q; determine the K data with the closest distance from the M data corresponding to the j-th data, where K is a positive integer; determine the average distance of the j-th data based on the distances from the K data to the j-th data; and determine the abnormal indicator data of the i-th battery to be tested based on the average distance between each data among the Q data of the i-th battery to be tested and each data.

[0050] In a possible implementation, the detection submodule is specifically configured to: perform change point detection on the time series data to obtain a change point detection result; and determine whether N batteries to be tested have an abnormality based on the change point detection result.

[0051] In one possible implementation, the detection submodule is specifically used to: obtain a prior probability distribution based on the data of the measured batteries in the time series data; obtain a new probability distribution based on the data of the batteries to be tested in the time series data and the data of the measured batteries in the time series data; and obtain a change point detection result based on the prior probability distribution and the new probability distribution.

[0052] In a possible implementation, the detection submodule is specifically configured to: when the prior probability distribution and the new probability distribution are different, determine that the change point detection result is a mutation in the data of the battery to be tested compared to the data of the tested batteries in the time series data.

[0053] In a possible implementation, the detection submodule is specifically configured to: respectively calculate the difference between two adjacent data in the time series data; and obtain a change point detection result based on a relationship between the difference and a third preset threshold.

[0054] In one possible implementation, the acquisition module is further used to: when an abnormality is detected in N batteries to be tested, obtain data related to the electrolyte injection process when producing N batteries to be tested; the device also includes: a determination module, used to determine whether there is a mismatch between the data related to the electrolyte injection process and a preset control range; the determination module is further used to: when there is a mismatch between the data related to the electrolyte injection process and the preset control range, determine that an electrolyte injection abnormality has occurred in the electrolyte of the N batteries to be tested.

[0055] In a possible implementation, the apparatus further includes: a generating module configured to generate electrolyte abnormality cause information of the N batteries to be tested when it is determined that electrolyte injection abnormality occurs in the N batteries to be tested.

[0056] In a possible implementation, the generating module is further configured to generate alarm information according to the electrolyte abnormality cause information, so as to intercept N batteries to be tested.

[0057] In a third aspect, an electronic device is provided, comprising a memory and a processor, wherein the memory is used to store instructions, and the processor is used to read the instructions and execute the method in the first aspect and any possible implementation thereof according to the instructions.

[0058] In a fourth aspect, a storage medium is provided for storing a computer program, wherein the computer program enables a computer to execute the method in the first aspect and any possible implementation manner of the first aspect.

[0059] In a fifth aspect, a chip is provided, comprising a processor for calling and running a computer program from a memory, so that a device equipped with the chip executes the method in the first aspect and any possible implementation thereof.

[0060] In a sixth aspect, a computer program is provided, which enables a computer to execute the method in the first aspect and any possible implementation manner of the first aspect.

[0061] In the seventh aspect, a program product is provided, comprising a computer-readable code, or a non-volatile computer-readable storage medium carrying the computer-readable code. When the computer-readable code runs in an electronic device, the processor in the electronic device executes the method as in the first aspect and any possible implementation of the first aspect.

[0062] In the present application, an anomaly detection model can be generated by training based on the down-sampling data of historically produced batteries within the anomaly recognition threshold range, and used to calculate the anomaly index data of newly produced batteries, thereby achieving self-adjustment of the model while ensuring the stability of the data and model; and by proposing the key film formation segment data of the formation process data and adopting the down-sampling method, while maintaining the anomaly detection capability, the consumption of storage and computing resources can also be reduced; in addition, by extracting the anomaly index data at time intervals for statistical value calculation, the influence of individual anomaly data can be eliminated, making the data more stable, and the use of a time series data mutation detection algorithm has obvious effects on batch anomaly recognition; finally, if the alarm of the sudden detection is triggered, the failure analysis is initiated, and the extraction of relevant data for root cause analysis can lock the cause of the anomaly. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the drawings without creative work.

[0064] FIG1 is a schematic diagram of the structure of the system architecture provided by this application.

[0065] FIG2 is a first schematic block diagram of a method for detecting a battery according to an embodiment of the present application.

[0066] FIG3 is a second schematic block diagram of a method for detecting a battery according to an embodiment of the present application.

[0067] FIG4 is a third schematic block diagram of a method for detecting a battery according to an embodiment of the present application.

[0068] FIG5 is a fourth schematic block diagram of a method for detecting a battery according to an embodiment of the present application.

[0069] FIG6 is a fifth schematic block diagram of a method for detecting a battery according to an embodiment of the present application.

[0070] FIG7 is a sixth schematic block diagram of a method for detecting a battery according to an embodiment of the present application.

[0071] FIG8 is a seventh schematic block diagram of a method for detecting a battery according to an embodiment of the present application.

[0072] FIG9 is an eighth schematic block diagram of a method for detecting a battery according to an embodiment of the present application.

[0073] FIG10 is a ninth schematic block diagram of a method for detecting a battery according to an embodiment of the present application.

[0074] FIG11 is a tenth schematic block diagram of a method for detecting a battery according to an embodiment of the present application.

[0075] FIG12 is an eleventh schematic block diagram of a method for detecting a battery according to an embodiment of the present application.

[0076] FIG13 is a schematic flowchart of a method for detecting a battery according to an embodiment of the present application.

[0077] FIG14 is a schematic block diagram of a device for detecting a battery according to an embodiment of the present application.

[0078] FIG15 is another schematic block diagram of a device for detecting a battery according to an embodiment of the present application.

[0079] FIG16 is a schematic block diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0080] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0081] Unless otherwise defined, all technical and scientific terms used in this application have the same meanings as commonly understood by those skilled in the art to which this application belongs. The terms used in the specification of this application are for the purpose of describing specific embodiments only and are not intended to limit this application. The terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first" and "second" in the specification and claims of this application or the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order or a primary-secondary relationship.

[0082] References to "embodiments" in this application mean that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described in this application may be combined with other embodiments.

[0083] The term "and / or" in this application simply describes an association between related objects, indicating that three possible relationships exist. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this application generally indicates that the related objects are in an "or" relationship.

[0084] If anomalies occur during the battery production process, such as incorrect electrolyte injection or impurities in the electrode, these problems may not be apparent during the testing phase of the production line. However, after repeated use, they can lead to problems such as lithium deposition, low capacity, and cycle failure. The loading of these abnormal batteries will affect the rate and cycle performance of the battery system. Current methods for identifying abnormal batteries have limited capacity for batch testing.

[0085] Therefore, an embodiment of the present application provides a method for detecting batteries, which obtains abnormal index data of N batteries to be tested based on the formation process data of N batteries to be tested and an abnormality detection model obtained by training the formation process data of M tested batteries, and performs detection based on the abnormal index data of the N batteries to be tested, which is conducive to realizing batch detection of batteries and can improve the accuracy of battery detection.

[0086] In order to better understand the solution of the embodiment of the present application, the possible application scenarios of the embodiment of the present application are briefly introduced below with reference to Figure 1.

[0087] As shown in Figure 1, an embodiment of the present application provides a system architecture 100. In Figure 1, a data acquisition device 160 is used to collect formation process data of a tested battery. For example, the data acquisition device 160 can be a formation device.

[0088] After collecting the battery formation process data, the data acquisition device 160 stores the formation process data of the tested battery in the database 130. The training device 120 obtains the target model / rule 101 based on the formation process data of multiple tested batteries in the database 130. For example, the target model / rule can be an anomaly detection model.

[0089] The above-mentioned target model / rule 101 can be used to realize the abnormality detection of the battery of the embodiment of the present application. The target model / rule 101 in the embodiment of the present application can specifically be a neural network. It should be noted that, in actual applications, the formation process data of the measured battery maintained in the database 130 does not necessarily all come from the collection of the data acquisition device 160, and may also be received from other devices. It should also be noted that the training device 120 does not necessarily train the target model / rule 101 completely based on the formation process data of the measured battery maintained by the database 130, and it is also possible to obtain formation process data from the cloud or other places for model training. The above description should not be used as a limitation to the embodiment of the present application.

[0090] The target model / rule 101 obtained through training with the training device 120 can be applied to different systems or devices, such as the execution device 110 shown in FIG1 . The execution device 110 can be a terminal, such as a mobile phone terminal, a tablet computer, a laptop computer, etc., or a server or a cloud. In FIG1 , the execution device 110 is configured with an input / output (I / O) interface 112 for data interaction with external devices. The user can input data into the I / O interface 112 through the client device 140. The input data can include, in the embodiment of the present application, formation process data of the battery to be tested.

[0091] In some implementations, the client device 140 and the execution device 110 may be the same device. For example, the client device 140 and the execution device 110 may both be terminal devices.

[0092] In other embodiments, the client device 140 may be a different device from the above-mentioned execution device 110. For example, the client device 140 is a terminal device, and the execution device 110 is a cloud, server, or other device. The client device 140 may interact with the execution device 310 through a communication network with any communication mechanism / communication standard. The communication network may be a wide area network, a local area network, a point-to-point connection, or any combination thereof.

[0093] The calculation module 111 of the execution device 110 is used to process input data (such as formation process data of the battery to be tested) received by the I / O interface 112. During the process of performing calculations and other related processing by the calculation module 111 of the execution device 110, the execution device 110 can call data, code, etc. in the data storage system 150 for corresponding processing, and can also store the data, instructions, etc. obtained from the corresponding processing in the data storage system 150.

[0094] Finally, the I / O interface 112 returns the processing result, such as the abnormal indicator data obtained above, to the client device 140 to provide it to the user.

[0095] It is worth noting that the training device 120 can generate corresponding target models / rules 101 based on different training data for different goals or different tasks. The corresponding target models / rules 101 can be used to achieve the above goals or complete the above tasks, thereby providing users with the desired results.

[0096] In the scenario shown in FIG1 , the user can manually input data, which can be performed through the interface provided by I / O interface 112. Alternatively, client device 140 can automatically send input data to I / O interface 112. If user authorization is required for client device 140 to automatically send input data, the user can set the corresponding permissions in client device 140. The user can view the output of execution device 110 on client device 140, which can be presented in a display, sound, action, or other specific form. Client device 140 can also serve as a data acquisition terminal, collecting input data and output results from I / O interface 112 as new sample data and storing them in database 130. Of course, collection can also be performed without client device 140, with I / O interface 112 directly storing the input data and output results from I / O interface 112 as new sample data in database 130.

[0097] It is worth noting that Figure 1 is only a schematic diagram of a system architecture provided by an embodiment of the present application. The positional relationship between the devices, components, modules, etc. shown in the figure does not constitute any limitation. For example, in Figure 1, the data storage system 150 is an external memory relative to the execution device 110. In other cases, the data storage system 150 can also be placed in the execution device 110.

[0098] As shown in Figure 1, the target model / rule 101 is obtained through training with the training device 120. The target model / rule 101 can be a neural network in the embodiment of the present application. Specifically, the neural network in the embodiment of the present application can be a convolutional neural network (CNN), a regional convolutional neural network (RCNN) or other types of neural networks, etc., and the present application does not make specific limitations on this.

[0099] Figure 2 shows a schematic block diagram of a method 200 for detecting a battery according to an embodiment of the present application. The method 200 may be executed by a detection apparatus, for example, the execution device 110 shown in Figure 1. Optionally, as shown in Figure 2, the method 200 may include some or all of the following contents.

[0100] Step 210 , obtaining formation process data of each of N batteries to be tested, where N is a positive integer greater than 1.

[0101] Step 220 , obtaining abnormal index data of each battery to be tested based on the formation process data of each battery to be tested in the N batteries to be tested and an abnormality detection model, wherein the abnormality detection model is trained based on the formation process data of M tested batteries, where M is a positive integer greater than 1.

[0102] Step 230 : detecting whether the N batteries to be tested have any abnormalities based on the abnormality indicator data of the N batteries to be tested.

[0103] The formation of a battery is a key step in battery production, and the result of the formation directly affects the cycle life, rate performance, and high and low temperature performance of the battery. The process data of the battery during the formation process are data such as the steps, time, current, voltage, temperature, and air pressure of the battery formation stored in seconds and milliseconds, which record the parameter changes of the battery during the formation process in detail. Among them, current, voltage, temperature, and air pressure are important parameters that characterize the battery formation process. In the embodiment of the present application, by analyzing the process data recorded by the battery during the formation process, it is possible to realize abnormal detection of the battery.

[0104] It should be noted that, in the embodiments of the present application, the process data of the battery during the formation process can be referred to as the formation process data of the battery. In addition, the battery to be tested and the tested battery in the embodiments of the present application can refer to batteries produced on the same production line. Since the production conditions of the batteries produced on the same production line are similar, that is, when there are no abnormalities in the production conditions, the batteries produced on the same production line should not differ much. Therefore, when testing the battery to be tested, the relevant data of the tested batteries produced on the same production line that have been tested and have no abnormalities can be referred to, thereby realizing batch testing of the batteries.

[0105] Specifically, the detection device can sequentially obtain the formation process data of a plurality of batteries, wherein the plurality of batteries may include the formation process data of a plurality of batteries to be tested and a plurality of tested batteries, wherein the tested batteries refer to batteries that have been tested and have no abnormalities. Optionally, the tested batteries may be detected by the method provided in the embodiments of the present application, or may be detected by other methods. For example, the tested batteries may be detected by the differential capacity curve of the battery. Optionally, the detection device can train the formation process data of the plurality of tested batteries to generate an anomaly detection model, and generate anomaly index data of the plurality of batteries to be tested based on the formation process data of the plurality of batteries to be tested and the anomaly detection model, and then detect whether the plurality of batteries to be tested have anomalies based on the anomaly index data of the plurality of batteries to be tested. For example, the detection device can detect whether each battery to be tested has an anomaly based on the relationship between the anomaly index data of each battery to be tested and a preset threshold value.

[0106] Optionally, the detection device may only obtain the formation process data of multiple batteries to be tested, and the abnormality detection model may be generated by other devices based on the training of the formation process data of multiple tested batteries. The detection device can obtain the abnormality detection model from other devices when it needs to detect multiple batteries to be tested.

[0107] Optionally, the anomaly detection model can also be adaptively adjusted. For example, after multiple batteries to be tested are detected without abnormalities at one time, the anomaly detection model can be further updated. That is, each time multiple batteries to be tested are detected, there is an anomaly detection model corresponding to each. It should be noted that for batteries produced on the same production line, since there are no tested batteries before the first battery, it is impossible to train and obtain an anomaly detection model. The first battery can be judged by other methods to determine whether it has an abnormality. In the case where the first battery is judged to be normal, since there are tested batteries before the subsequent batteries, the subsequent batteries can all be detected using the method of the embodiment of the present application. In other words, the technical solution provided by the embodiment of the present application is applicable to any battery produced on the same production line after the first battery that is detected to be normal.

[0108] In this embodiment, by obtaining the formation process data of N batteries to be tested, and obtaining the abnormal index data of the N batteries to be tested based on the formation process data of the N batteries to be tested and the abnormality detection model, it is detected whether the N batteries to be tested have abnormalities based on the abnormal index data of the N batteries to be tested, which is conducive to realizing batch detection of batteries and can realize adaptive adjustment of the abnormality detection model.

[0109] Alternatively, in an embodiment of the present application, the abnormal index data can be used to characterize the similarity between the formation process data of the battery to be tested and the formation process data of the tested battery. The type of the abnormal index data can be related to the type of the abnormality detection model. For example, if the abnormality detection model is a KNN model, the abnormal index data can be a distance value. For another example, if the abnormality detection model is a LOF model, the abnormality detection model can be a density value.

[0110] Optionally, in one embodiment, as shown in FIG3 , step 230, i.e., detecting whether the N batteries to be tested have abnormalities based on the abnormal indicator data of the N batteries to be tested, may include: step 310, generating time series data based on the abnormal indicator data of the N batteries to be tested and the abnormal indicator data of P tested batteries, where P is a positive integer greater than 1; and step 320, detecting whether the N batteries to be tested have abnormalities based on the time series data.

[0111] The detection device can sequentially obtain abnormal index data of multiple batteries produced on the same production line, that is, the abnormal index data of multiple batteries are time-series. The detection device can store the abnormal index data of the battery in the database every time the abnormal index data of a battery is obtained. After the database stores the abnormal index data of N batteries to be tested, the detection device can obtain the abnormal index data of the N batteries to be tested and the abnormal index data of P tested batteries from the database, and generate time series data according to the order of the acquisition time of the abnormal index data of the N batteries to be tested and the abnormal index data of the P tested batteries. Furthermore, the detection device can detect whether the N batteries to be tested have abnormalities based on the time series data. Wherein, N and P can both be positive integers greater than 1, and the P tested batteries in this embodiment can have no correlation with the M tested batteries in the above embodiment.

[0112] It should be noted that the abnormal indicator data of any battery in the embodiments of the present application can be obtained using the technical solution provided in the present application. However, those skilled in the art will appreciate that the embodiments of the present application are not limited thereto.

[0113] In this embodiment, based on the time series data generated from the abnormal indicator data of N batteries to be tested and the abnormal indicator data of P tested batteries, it is detected whether the N batteries to be tested have abnormalities, which is conducive to batch detection of batteries.

[0114] Optionally, in an embodiment of the present application, as shown in Figure 4, step 310, i.e., generating time series data based on the abnormal indicator data of the N batteries to be tested and the abnormal indicator data of the P tested batteries, may include: step 410, obtaining first abnormal indicator data from the abnormal indicator data of the N batteries to be tested; step 420, obtaining at least one second abnormal indicator data from the abnormal indicator data of the P tested batteries; step 430, generating the time series data based on the first abnormal indicator data and the at least one second abnormal indicator data.

[0115] Optionally, if the abnormal indicator data of N batteries to be tested and the abnormal indicator data of P tested batteries are relatively large, the abnormal indicator data of the N batteries to be tested and the abnormal indicator data of the P tested batteries can be compressed first. For example, at least one first abnormal indicator data can be obtained from the abnormal indicator data of the N batteries to be tested, the number of the at least one first abnormal indicator data is less than N, and at least one second abnormal indicator data can be obtained from the abnormal indicator data of the P tested batteries, the number of the at least one second abnormal indicator data is less than P. Then, time series data is generated based on the at least one first abnormal indicator data and the at least one second abnormal indicator data. For another example, the abnormal indicator data of the N batteries to be tested and the abnormal indicator data of the P tested batteries can be averaged, so that a first abnormal indicator data and a second abnormal indicator data can be obtained.

[0116] Optionally, in other embodiments, if the abnormal indicator data of the N batteries to be tested and the abnormal indicator data of the P tested batteries are not large, the abnormal indicator data of the N batteries to be tested and the abnormal indicator data of the P tested batteries may not be compressed, and the time series data may be directly combined in the order of timestamps to form time series data.

[0117] In this embodiment, time series data is generated based on the first abnormal indicator data obtained from the abnormal indicator data of N batteries to be tested and at least one second abnormal indicator data obtained from the abnormal indicator data of P tested batteries, rather than directly based on all the abnormal indicator data of the N batteries to be tested and the P tested batteries. This is beneficial to reducing the amount of data in the time series data, thereby reducing the complexity of battery detection.

[0118] Optionally, in an embodiment of the present application, the abnormal indicator data of the N batteries to be tested is obtained within a first time interval, and the abnormal indicator data of the P tested batteries is obtained within a second time interval, and the second time interval may include at least one time series. Further, as shown in FIG5 , step 410, i.e., obtaining first abnormal indicator data from the abnormal indicator data of the N batteries to be tested, may include: step 411, obtaining first abnormal indicator data from the first time interval; step 420, i.e., obtaining at least one second abnormal indicator data from the abnormal indicator data of the P tested batteries, may include: step 422, obtaining one second abnormal indicator data from each time series in the at least one time series.

[0119] As mentioned above, after obtaining the abnormal index data of the battery, it can be stored in the database in the order of the time of acquisition. Then, the abnormal index data of N batteries to be tested and P tested batteries can be obtained from the database according to the time interval. For example, the detection device can obtain the abnormal index data of the battery to be tested within a first time interval (for example, the latest 10 minutes) from the database, and the abnormal index data before the latest 10 minutes are all abnormal index data of the tested battery. The detection device can also obtain the abnormal index data within a second time interval (for example, 50 minutes before the latest 10 minutes) from the database. Optionally, the two time intervals can be continuous with the first time interval, for example, the abnormal index data within the 50 minutes and the abnormal index data within the current 10 minutes are continuously stored in the database. Furthermore, the detection device can also divide the 50 minutes into 5 time series, each time series has a duration of 10 minutes. That is, the detection device can obtain the abnormal index data within the current hour, divide the hour into 6 time series, each time series has a duration of 10 minutes. The detection device can obtain an abnormal indicator data from each time series, and thus form time series data by arranging the abnormal indicator data in the order of the acquisition time.

[0120] Optionally, the duration of each time series in the at least one time series divided by the second time interval may be the same or different. Furthermore, the duration of each time series may be the same as or different from the duration of the first time interval, which is not limited in this embodiment of the present application.

[0121] Optionally, the first time interval may be continuous with the second time interval, or may be discontinuous with the second time interval, which is not limited in the present embodiment.

[0122] Optionally, when the detection device obtains the first abnormal indicator data from the first time interval, it can refer to the acquisition time of the abnormal indicator data of the N batteries to be tested within the first time interval. For example, the abnormal indicator data with the acquisition time closest to the middle position of the first time interval can be selected as the first abnormal indicator data. Similarly, when the detection device obtains a second abnormal indicator data from each time sequence in the second time interval, it can refer to the acquisition time of the abnormal indicator data of the tested batteries in each time sequence. For example, the abnormal indicator data close to the middle position of each time sequence can be selected as the second abnormal indicator data. It should be understood that the embodiments of the present application do not limit how to obtain the first abnormal indicator data based on the acquisition time of the abnormal indicator data of the N batteries to be tested and how to obtain the second abnormal indicator data based on the acquisition time of the abnormal indicator data of the tested batteries in each time sequence.

[0123] In this embodiment, obtaining first abnormal indicator data from a first time interval and obtaining second abnormal indicator data from each time series in at least one time series sequence is beneficial for making the time series data include abnormal indicator data uniformly distributed in the time dimension, thereby facilitating improving the accuracy of battery detection.

[0124] Optionally, in an embodiment of the present application, the second time interval is divided into at least one time series according to a preset time length. Further, as shown in Figure 6, step 411, i.e., obtaining the first abnormal indicator data from the first time interval, may include: step 510, within the first time interval, performing statistics on the abnormal indicator data of the N batteries to be tested; step 520, according to a preset statistical quantile, obtaining the first abnormal indicator data from the first time interval; step 412, i.e., obtaining a second abnormal indicator data from each time series in at least one time series, may include: step 530, within each time series, performing statistics on the abnormal indicator data of the tested battery corresponding to each time series; step 540, according to the preset statistical quantile, obtaining a second abnormal indicator data from each time series.

[0125] For example, the acquisition time of the abnormal indicator data of N batteries to be tested is between 9:00 and 9:10, and the acquisition time of the abnormal indicator data of P tested batteries is between 8:10 and 9:00, that is, the length of the first time interval is 10 minutes, and the length of the second time interval is 50 minutes. The second time interval is divided into 5 continuous and equal-length time series, that is, the 8:10-9:00 is divided into five time series: 8:10-8:20, 8:20-8:30, 8:30-8:40, 8:40-8:50 and 8:50-9:00. Assuming that there are abnormal indicator data of 3 batteries to be tested in the first time interval and abnormal indicator data of 3 tested batteries in each time series, statistics can be performed on the abnormal indicator data of the 3 batteries to be tested in the first time interval, and statistics can be performed on the abnormal indicator data of the 3 tested batteries in each time series in the second time interval, and the first abnormal indicator data can be obtained from the abnormal indicator data of the 3 batteries to be tested in the first time interval and the second abnormal indicator data can be obtained from the abnormal indicator data of the 3 tested batteries in each time series according to preset statistical quantiles, such as the median, the 25% quantile or the 75% quantile. For example, the abnormal indicator data of three batteries to be tested within the first time interval can be sorted according to the acquisition time of the abnormal indicator data of the batteries, and the second abnormal indicator data within the first time interval can be used as the first abnormal indicator data; the abnormal indicator data of three tested batteries in each time series can be sorted according to the acquisition time of the abnormal indicator data of the batteries, and the second abnormal indicator data in each time series of five time series can be used as the second abnormal indicator data; finally, the first abnormal indicator data and the five second abnormal indicator data can be sorted in chronological order according to the acquisition time of the abnormal indicator data of the batteries to form time series data.

[0126] Optionally, the preset duration and the preset statistical quantile are both preset values. The preset duration may, for example, be equal to the duration of the first time interval. The preset statistical quantile may, for example, be the median, the 25th percentile, or the 75th percentile. The specific values ​​of the preset duration and the preset statistical quantile are not limited in this embodiment of the application.

[0127] In this embodiment, an abnormal indicator data is obtained from the first time interval and a second abnormal indicator data is obtained from each time series sequence in the second time interval according to the same preset statistical quantile, so that the time series data includes abnormal indicator data that is uniformly distributed in the time dimension, thereby improving the accuracy of battery detection.

[0128] Optionally, in an embodiment of the present application, step 411, i.e., obtaining first abnormal indicator data from a first time interval, may include: when N is greater than a first preset threshold, obtaining first abnormal indicator data from the first time interval; step 412, i.e., obtaining a second abnormal indicator data from each time series in at least one time series, may include: when the number of measured batteries corresponding to each time series is greater than the first preset threshold, obtaining a second abnormal indicator data from each time series.

[0129] In other words, before obtaining the abnormal indicator data for generating the time series data from each time series sequence or time interval, the relationship between the number of batteries corresponding to the time series sequence or time interval and the first preset threshold value can be determined one by one. If the number of batteries corresponding to a time series sequence or time interval is small, the time series sequence or time interval can be eliminated, that is, the abnormal indicator data for generating the time series data is not obtained in the time series sequence or time interval. If the number of batteries corresponding to a time series sequence or time interval is large, the abnormal indicator data for generating the time series data can be directly obtained in the time series sequence or time interval. For example, if a time series sequence includes less than 3 abnormal indicator data, the second abnormal indicator data is not obtained in the time series sequence; if a time series sequence includes no less than 3 reaction peak data, the second abnormal indicator data is obtained in the time series sequence. It should be understood that if only one abnormal indicator data in a first time interval is the abnormal indicator data of the battery to be tested, it is also possible to obtain the first abnormal indicator data directly from the first time interval without determining the number of batteries to be tested corresponding to the first time interval. However, if there are multiple abnormal indicator data within the first time interval that are abnormal indicator data of the battery to be tested, before obtaining the first abnormal indicator data, it is still necessary to judge the number of batteries to be tested corresponding to each first time interval, but it must be ensured that the finally generated time series data contains at least one abnormal indicator data of the battery to be tested.

[0130] In this embodiment, when N is greater than the first preset threshold, the first abnormal indicator data is obtained within the first time interval, and when the number of measured batteries corresponding to each time series is greater than the first preset threshold, the second abnormal indicator data is obtained within each time series. This can reduce the probability of obtaining abnormal indicator data with large deviations, thereby improving the accuracy of battery detection.

[0131] It should be understood that the first preset threshold may be a value pre-set based on experience, and the embodiment of the present application does not limit the value of the first preset threshold.

[0132] Optionally, as shown in FIG7 , step 230, i.e., detecting whether the N batteries to be tested have abnormalities based on the abnormal indicator data of the N batteries to be tested, may include: step 610, determining the labels of the N batteries to be tested based on the abnormal indicator data of the N batteries to be tested, where the label may be an abnormal label for indicating that the battery has an abnormality or a normal label for indicating that the battery does not have an abnormality; step 620, generating timing data based on the labels of the N batteries to be tested and the labels of P tested batteries, where P is a positive integer greater than 1; and step 630, detecting whether the N batteries to be tested have abnormalities based on the timing data.

[0133] Optionally, the detection device may determine the battery label based on the relationship between the abnormal indicator data and a preset threshold value. For example, when the abnormal indicator data is greater than the preset threshold value, the battery label is determined to be an abnormal label. For another example, when the abnormal indicator data is less than or equal to the preset threshold value, the battery label is determined to be a normal label.

[0134] After the detection device obtains the label of each battery, the label can also be stored in the above-mentioned database in the order of the acquisition time. That is to say, the database can include information of multiple batteries, and the information of each battery includes both the abnormal indicator data of the battery and the label of the battery. After the database stores the labels of N batteries to be tested, the detection device can obtain the labels of the N batteries to be tested and the labels of P tested batteries from the database, and generate time series data according to the order of the acquisition time of the labels of the N batteries to be tested and the labels of the P tested batteries. Furthermore, the detection device can detect whether the N batteries to be tested have abnormalities based on the time series data. Wherein, N and P can both be positive integers greater than 1, and there can be no correlation between the P tested batteries in this embodiment and the M tested batteries in the above-mentioned embodiment.

[0135] It should be noted that the label of any battery in the embodiments of the present application can be obtained using the technical solution provided in the present application. However, those skilled in the art will appreciate that the embodiments of the present application are not limited thereto.

[0136] In this embodiment, based on the time series data generated from the tags of N batteries to be tested and the tags of P tested batteries, it is detected whether the N batteries to be tested have abnormalities, which is conducive to batch detection of batteries.

[0137] Optionally, in an embodiment of the present application, the tags of the N batteries to be tested are obtained in a first time interval, and the tags of the P tested batteries are obtained in a second time interval, and the second time interval is divided into at least one time sequence according to a preset duration.

[0138] As mentioned above, after obtaining the battery labels, they can be stored in the database in the order of their acquisition time. Then, the labels of N batteries to be tested and P tested batteries can be obtained from the database according to the time interval. For example, the detection device can obtain the labels of the batteries to be tested within the first time interval (for example, the latest 10 minutes) from the database, and the labels before the latest 10 minutes are all labels of the tested batteries. The detection device can also obtain the labels within the second time interval (for example, 50 minutes before the current 10 minutes) from the database. Optionally, the second time interval can be continuous with the first time interval, for example, the labels within the 50 minutes and the labels within the current 10 minutes are continuously stored in the database. Furthermore, the detection device can also divide the second time interval into at least one time series. For example, the 50 minutes are divided into 5 time series, and the duration of each time series is 10 minutes. That is to say, the detection device can obtain the tags within the current hour, divide the hour into 6 time sequences, each time sequence is 10 minutes long, the tags in the latest time sequence are the tags of the battery to be tested, and the tags in other time sequences are the tags of the tested batteries.

[0139] Optionally, the duration of each time series in the at least one time series divided by the second time interval may be the same or different. Furthermore, the duration of each time series may be the same as or different from the duration of the first time interval, which is not limited in this embodiment of the present application.

[0140] Optionally, in other embodiments, the first time interval may be discontinuous with the second time interval.

[0141] Optionally, in one embodiment, as shown in Figure 8, step 620, i.e., generating time series data based on the labels of the N batteries to be tested and the labels of the P tested batteries, may include: step 710, counting the number of abnormal labels in the labels of the N batteries to be tested within a first time interval; step 720, counting the number of abnormal labels in the labels of the tested batteries corresponding to each time series sequence in each time series sequence; step 730, generating time series data based on the number of abnormal labels counted within the first time interval and the number of abnormal labels counted in each time series sequence.

[0142] For example, the acquisition time of the tags of N batteries to be tested is between 9:00 and 9:10, and the acquisition time of the tags of P tested batteries is between 8:10 and 9:00. That is, the duration of the first time interval is 10 minutes, and the duration of the second time interval is 50 minutes. The second time interval is divided into 5 consecutive time series of equal length, that is, the 8:10-9:00 is divided into 8:10-8:20, 8:20-8:30, 8:30-8:40, 8:40-8:50, and 8:50-9:00. Assume that there are 5 tags of batteries to be tested in the first time interval, and each time series also has 5 tags of tested batteries. Among them, there is 1 abnormal tag in the first time interval, and the number of abnormal tags in the 5 time series is 0, 0, 0, 1, 1 respectively. In chronological order, the generated time series data are [0, 0, 0, 1, 1, 1].

[0143] Optionally, in another embodiment, as shown in Figure 9, step 620, i.e., generating time series data based on the labels of the N batteries to be tested and the labels of the P tested batteries, may include: step 810, calculating the proportion of abnormal labels in the labels of the N batteries to be tested in the first time interval; step 820, respectively calculating the proportion of abnormal labels in the labels of the tested batteries in each time series in the at least one time series; step 830, generating time series data based on the proportion of abnormal labels in the first time interval and the proportion of abnormal labels in each time series.

[0144] For example, the acquisition time of the tags of N batteries to be tested is between 9:00 and 9:10, and the acquisition time of the tags of P tested batteries is between 8:10 and 9:00, that is, the length of the first time interval is 10 minutes, and the length of the second time interval is 50 minutes. The second time interval is divided into 5 consecutive and equal-length time sequences, that is, the 8:10-9:00 is divided into five time sequences of 8:10-8:20, 8:20-8:30, 8:30-8:40, 8:40-8:50 and 8:50-9:00. Assume that there are 5 labels of batteries to be tested in the first time interval, and there are 5 labels of tested batteries in each time series. Among them, there is 1 abnormal label in the first time interval, accounting for 20%, and the number of abnormal labels in the 5 time series are 0, 0, 0, 1, 1, respectively, accounting for 0, 0, 0, 20%, 20% respectively. In chronological order, the generated time series data are [0, 0, 0, 20%, 20%, 20%].

[0145] Optionally, as shown in Figure 10, step 220, i.e., obtaining abnormal index data of each battery to be tested based on the formation process data and the abnormality detection model of each battery to be tested, may include: step 910, respectively obtaining the film formation key segment data of each battery to be tested from the formation process data of each battery to be tested, the film formation key segment data refers to all formation process data of the battery between zero and a second preset threshold value during the formation process, and the second preset threshold value is related to the reaction potential of the additive expected to be added to the electrolyte of the N batteries to be tested; step 920, obtaining the abnormal index data of each battery to be tested based on the film formation key segment data and the abnormality detection model of each battery to be tested, the abnormality detection model is trained based on the film formation key segment data of M tested batteries.

[0146] Typically, the formation process data of a battery is the data such as the steps, time, current, voltage, temperature and air pressure of the battery formation stored in seconds and milliseconds, but the main purpose of the formation of a battery is to form a solid electrolyte interface (SEI) film, that is, the performance of the battery tends to be stable after the SEI film is formed. Therefore, the formation process data from the beginning to the formation of the SEI film is the key to detecting whether the battery is abnormal. In an embodiment of the present application, after obtaining the formation process data of the battery, the detection device can first extract this section of data of SEI film formation from the formation process data of the battery, which can be referred to as film-forming key section data. Specifically, all formation process data of the battery voltage arriving at the second preset threshold from zero during the formation process can be obtained from the formation process data of the battery. Optionally, the second preset threshold can be related to the reaction potential of the additive in the electrolyte of the N batteries to be tested. If the technical solution of the present application is applicable to the detection of batteries on the same production line, then the second preset threshold is related to the additive in the electrolyte of the batteries produced on the production line. The second preset threshold may be 2.5V or 3V, for example.

[0147] Should be understood that, although the formation process data comprises various types of data, for example, data such as voltage, current, temperature, air pressure, the data processing involved in the embodiment of the present application is directed to a type of data. For example, all data processing in the embodiment of the present application is directed to the voltage of the battery in the formation process.

[0148] In addition, the key film formation segment data of any battery in the embodiments of the present application can be obtained using the technical solution provided in the present application, but those skilled in the art understand that the embodiments of the present application are not limited thereto.

[0149] It should be understood that the second preset threshold value may be a value preset based on experience, and the embodiment of the present application does not limit the value of the second preset threshold value.

[0150] In this embodiment, by acquiring data of key film-forming sections, the anomaly detection model can be focused on these areas, preventing the effect of anomaly detection from being affected by factors such as battery polarization and capacity differences in the later stage. In addition, the amount of data storage can be effectively reduced.

[0151] Optionally, as shown in Figure 11, step 920, that is, obtaining the abnormal index data of each battery to be tested based on the film formation key segment data and the abnormality detection model of each battery to be tested, includes: step 1010, using a sliding window method to obtain a group of formation process data with the highest similarity to the standard film formation segment curve from the film formation key segment data of each battery to be tested, and the standard film formation segment curve is formed by fitting based on the film formation key segment data of L tested batteries, where L is a positive integer greater than 1; step 1020, obtaining the abnormal index data of each battery to be tested based on a group of formation process data and the abnormality detection model of each battery to be tested.

[0152] For example, the formation process data is voltage data, and the key film formation segment data of the battery to be tested includes 500 voltage data. Similarly, 500 voltage data corresponding to the 500 voltage data included in the key film formation segment data of the battery to be tested are obtained from the standard film formation segment curve. The length of the sliding window is 100, and the step of the sliding window is 1. Take 100 voltage data as a group of voltage data, and calculate the similarity between each group of voltage data and the corresponding voltage data in the standard film formation segment curve. For example, calculate the average distance between each group of voltage data and the corresponding voltage data in the standard film formation segment curve. The smaller the average distance, the higher the similarity. It should be understood that the length and step of the sliding window can be designed according to actual conditions.

[0153] In addition, the standard film-forming section curve can be formed by fitting the key film-forming section data of the L measured batteries. Optionally, in one embodiment, the standard film-forming section curve can be generated by fitting the detection device, that is, the detection device can first obtain the key film-forming section data of the L measured batteries, and fit the key film-forming section data of the L measured batteries to generate a standard film-forming section curve. Optionally, in other embodiments, the standard film-forming section curve can also be generated by fitting other devices, and obtained from other devices when the detection device needs it. It should be noted that the measured batteries here also refer to batteries that have been tested to have no abnormalities. The L measured batteries in the embodiment of the present application may have no correlation with the M measured batteries and the P measured batteries in the above embodiment.

[0154] For any battery in the embodiments of the present application, a set of formation process data with the highest similarity to the standard film formation curve can be obtained using the technical solution provided in the present application, but those skilled in the art understand that the embodiments of the present application are not limited thereto.

[0155] Optionally, the embodiment of the present application does not limit the method for calculating similarity, for example, a method for calculating similarity by distance, a cosine similarity algorithm, and a method for calculating a correlation coefficient.

[0156] In this embodiment, a sliding window method is used to obtain a set of formation process data with the highest similarity to the standard film formation curve from the key film formation section data of the battery to be tested, which can further reduce the data storage amount.

[0157] Optionally, as shown in Figure 12, step 1020, that is, obtaining abnormal index data of each battery to be tested based on a set of formation process data of each battery to be tested and an abnormality detection model, includes: step 2010, downsampling a set of formation process data of each battery to be tested obtained by step 1010 and generating downsampled data of each battery to be tested; step 2020, inputting the downsampled data of each battery to be tested into the abnormality detection model, and outputting the abnormal index data of each battery to be tested, the abnormality detection model is obtained by training based on the downsampled data of M tested batteries, and the standard film formation section curve is formed by fitting based on the downsampled data of L tested batteries.

[0158] Specifically, after adopting the above-described embodiment to obtain one group of formation process data that is the highest with the standard film forming section curve similarity, can further this group of formation process data be carried out down sampling.This down sampling mode can comprise the mode that time interval extracts or modes such as LTTB down sampling.With the mode that time interval extracts is example, suppose that one group of formation process data that is the highest with the standard film forming section curve similarity comprises 100 voltage data, because these 100 voltage data are recorded in chronological order, then can every interval 5 data extraction, for example, these 100 voltage data are respectively V1, V2, ..., V100 in chronological order, then down sampling data is V5, V10, V15, V20, V25, V30, V35, V40, V45, V50, V55, V60, V65, V70, V75, V80, V85, V90, V95, these 20 voltage data of V100.

[0159] The down-sampling data of any battery in the embodiments of the present application can be obtained using the technical solution provided in the present application, but those skilled in the art understand that the embodiments of the present application are not limited thereto.

[0160] In this embodiment, downsampling is performed on a set of formation data input to the anomaly detection model, which can continue to reduce the amount of data storage and the consumption of computing resources while maintaining the ability to detect batteries.

[0161] Optionally, in an embodiment of the present application, the anomaly detection model may be a clustering method based on distance, density, or angle metrics, such as a KNN model, a LOF model, or an ABOD model.

[0162] Taking the KNN model as an example of an anomaly detection model, the downsampled data of the i-th battery to be tested among the N batteries to be tested and the downsampled data of each tested battery among the M tested batteries both include Q data, and the Q data of the i-th battery to be tested corresponds one-to-one with the Q data of each tested battery, wherein i is any positive integer from 1 to N, and Q is a positive integer greater than 1. In step 2020, the downsampled data of each battery to be tested is input into the anomaly detection model, and the abnormal index data of each battery to be tested is output, which may include : Calculate M distances between the jth data in the Q data of the i-th battery to be tested and the corresponding data in the Q data of each tested battery, where j is any positive integer from 1 to Q; determine the K data with the closest distance from the M data corresponding to the j-th data, where K is a positive integer; determine the average distance of the j-th data based on the distances from the K data to the j-th data; determine the abnormal indicator data of the i-th battery to be tested based on each data in the Q data of the i-th battery to be tested and the average distance of each data.

[0163] The following describes a technical solution for obtaining abnormal indicator data based on a KNN model, using the i-th battery under test among N batteries under test as an example. Assume that the downsampled data for the battery under test includes five voltage data points, and the KNN model is fitted using the downsampled data from two tested batteries. Each tested battery also has five voltage data points, and there is a one-to-one correspondence between the five voltage data points of the battery under test and the ten voltage data points of the tested batteries. See Table 1 for details.

[0164] Table 1

[0165] First, the distances between each data of the i-th battery to be tested and the data corresponding to the first tested battery and the data corresponding to the second tested battery are calculated respectively, as shown in Table 2 for details.

[0166] Table 2

[0167] For each downsampled data of the i-th battery to be tested, the downsampled data corresponding to the K nearest tested batteries are selected. For example, if K=2, for 0.50, the closest data are the downsampled data 0.55 of the second tested battery and the downsampled data 0.52 of the third tested battery, etc.

[0168] Based on Calculate the average distance of each downsampled data of the i-th battery to be tested. 1 Where x1, x2, ..., xN are the downsampled data of the K tested batteries corresponding to each downsampled data of the i-th battery to be tested, see Table 3 for details.

[0169] Table 3

[0170] Finally, according to Calculate the abnormal index data of the i-th battery to be tested. For example, according to the average distance in Table 3, the abnormal index data of the i-th battery to be tested can be calculated to be 0.08.

[0171] Optionally, in an embodiment of the present application, step 220 or step 530, i.e., detecting whether N batteries to be tested have abnormalities based on the time series data, may include: performing change point detection on the time series data to obtain a change point detection result; and determining whether the N batteries to be tested have abnormalities based on the change point detection result.

[0172] In this embodiment, by performing change point detection on the time series data, it is possible to effectively detect whether an abnormality occurs in the battery to be tested.

[0173] Optionally, the time series data may be detected using a Bayesian change point detection algorithm, a threshold control algorithm, a Markov chain algorithm, and an AriMA time series prediction algorithm.

[0174] Taking the Bayesian change point detection algorithm as an example, optionally, in an embodiment of the present application, change point detection is performed on the time series data to obtain a change point detection result, which may include: obtaining a priori probability distribution based on the data of the measured batteries in the time series data; obtaining a new probability distribution based on the data of the batteries to be tested in the time series data and the data of the measured batteries in the time series data; and obtaining a change point detection result based on the priori probability distribution and the new probability distribution.

[0175] Further optionally, obtaining a change point detection result based on the prior probability distribution and the new probability distribution may include: when the prior probability distribution and the new probability distribution are different, determining that the change point detection result indicates that the data of the battery to be tested in the time series data has a sudden change compared to the data of the tested batteries. Alternatively, when the prior probability distribution and the new probability distribution are the same, determining that the change point detection result indicates that the data of the battery to be tested in the time series data has no sudden change compared to the data of the tested batteries.

[0176] Bayesian change point detection focuses on causal predictive filtering, which only considers the observed data and generates an accurate distribution of the next unseen data in the sequence.

[0177] Time series data is represented by [x1, x2, ..., xt], which is divided into different distribution partitions. The division between partitions is called a change point. Further assume that for a partition ρ, ρ = 1, 2, 3 ...., where the data comes from a probability distribution p(x t |θ ρ ), each θ ρ The independence condition is met, so the prior probability distribution of the interval between the change points is expressed as p gap (g). Therefore, the posterior probability density is:

[0178] For p(r t |x 1:t ) to perform conditional decomposition:

[0179] Assuming that the probability value before each sequence is known, then r t The probability depends on two possibilities, namely, the change point occurs or the change point does not occur, when the run length increases by 1. In order to judge these two possibilities and calculate the probability in these two cases, it is necessary to calculate the risk function:

[0180] Notice: If p gap It is a discrete exponential (geometric) distribution with a time scale of λ. The process is memoryless and the hazard function H(τ) = 1 / λ is a constant.

[0181] Taking the threshold control algorithm as an example, optionally, in an embodiment of the present application, the time series data is subjected to change point detection to obtain a change point detection result, including: respectively calculating the difference between two adjacent data in the time series data; and obtaining the change point detection result based on the relationship between the difference and a third preset threshold.

[0182] For example, the time series data includes 6 data, including data of 1 battery to be tested and data of 5 tested batteries. The time series data can be represented by [x1, x2, x3, x4, x5, x6]. The 5 values ​​x1-x2, x2-x3, x3-x4, x4-x5, and x5-x6 are calculated respectively, and the 5 values ​​are compared with the third preset threshold value respectively. If one of them is greater than the third preset threshold value, it is considered that the change point detection result is that the reaction peak data of the battery to be tested in the time series data has a sudden change compared with the reaction peak data of the tested battery; if none of them is greater than the second preset threshold value, it is considered that the change point detection result is that the reaction peak data of the battery to be tested in the time series data has no sudden change compared with the reaction peak data of the tested battery.

[0183] It should be understood that the third preset threshold may be a value pre-set based on experience, and the embodiment of the present application does not limit the value of the third preset threshold.

[0184] Optionally, in an embodiment of the present application, the method 200 further includes: when an abnormality is detected in N batteries to be tested, obtaining data related to the electrolyte injection process when producing the N batteries to be tested; determining whether there is a mismatch between the data related to the electrolyte injection process and a preset control range; when there is a mismatch between the data related to the electrolyte injection process and the preset control range, determining that an electrolyte injection abnormality has occurred in the electrolyte of the N batteries to be tested.

[0185] In the case of detecting abnormalities in N batteries to be tested, failure analysis can be performed on the N batteries to be tested. For example, the relevant data of the failure mode can be matched, and the relevant data can be checked to see if it exceeds a set threshold. If it exceeds the set threshold, the problem attribution can be returned. For example, the failure mode is electrolyte injection abnormality, and the relevant data of the failure mode may include data related to the electrolyte injection process, such as electrolyte injection quality data, electrolyte replacement time, etc. The preset control range may, for example, refer to a control range that the pre-set injection quality cannot exceed or a control range that the pre-set injection pressure cannot exceed (usually, the density of different electrolyte products will be different, resulting in differences in injection quality).

[0186] Optionally, in the embodiment of the present application, the method 200 may further include: generating electrolyte abnormality cause information of the N batteries to be tested when it is determined that electrolyte injection abnormality occurs in the N batteries to be tested.

[0187] Optionally, in the embodiment of the present application, the method 200 may further include: generating alarm information according to the electrolyte abnormality cause information to intercept N batteries to be tested.

[0188] Optionally, if it is determined that an electrolyte injection anomaly has occurred in N batteries under test, electrolyte anomaly cause information can be generated and output. This electrolyte anomaly cause information can indicate an electrolyte injection anomaly in the batteries under test. Furthermore, based on this electrolyte anomaly cause information, an alarm can be generated to mark and intercept the N batteries under test. Furthermore, the battery injection process on the production line can be stopped, and the correct electrolyte can be replaced to avoid the recurrence of electrolyte injection anomalies during subsequent battery production.

[0189] The method for detecting a battery according to an embodiment of the present application will be described in detail below with reference to FIG13. As shown in FIG13, the method may include some or all of the following contents.

[0190] Step 2001, acquisition of formation process data. Specifically, it can include: (1) formation process data input. The process data of battery formation, such as steps, time, current, voltage, temperature, and air pressure, stored at the second and millisecond levels, record in detail the parameter changes of the battery during the formation process, among which current, voltage, temperature, and air pressure are important parameters for characterizing the battery formation process; (2) abnormal formation process data removal. Due to the interruption of formation, abnormal sampling of formation equipment, no participation in formation, interruption of film forming section, re-testing of formation, etc., the abnormal data is significantly different from the normal data, so it is necessary to delete the abnormal formation process data to prevent it from affecting the calculation results of abnormal indicators; (3) formation process data interpolation compensation. Since there will be sampling omissions at some time during the formation process, in order to prevent fluctuations caused by missing formation process data in the later abnormal indicator calculation process, which affects the calculation results of abnormal indicators, it is necessary to interpolate and fill in the gaps.

[0191] Step 2002, downsampling of the formation process data. Specifically, it can include: (1) Extraction of the key film-forming segment data of the formation process data. Select all current, voltage, temperature or air pressure data from the initial charging of the battery until the voltage reaches 2.5V or 3.0V, because the reaction potential of the additive is usually in the range of 0 to 2.0V (SEI film formation segment). By extracting the key film-forming segment data, the anomaly detection algorithm can be focused on these areas to prevent the effect of anomaly detection from being affected by factors such as battery polarization and capacity differences in the later stage. (2) By using a fixed sliding window to calculate the similarity with the standard film-forming segment curve, the formation process data with the same phase as the standard film-forming segment curve is intercepted, and then the intercepted data is further reduced by the downsampling method to reduce the storage requirements and computing resources.

[0192] Step 2003: Abnormal indicator data is obtained by calculating the abnormality detection model. Specifically, the following steps may be performed: (1) The abnormality detection model may be an algorithm such as KNN or LOF, and the model is trained by calling the downsampled data of batteries that do not have abnormalities in the historical data. (2) The downsampled data of the most recently produced batteries is input into the abnormality detection model to obtain abnormal indicator data.

[0193] Step 2004: Determine the battery label. Specifically, the battery label can be determined based on the relationship between the abnormal indicator data obtained in step 2003 and the threshold. If the abnormal indicator data exceeds the threshold, label = 1; if the abnormal indicator data does not exceed the threshold, label = 0. Optionally, in step 2003, the anomaly detection model is trained using downsampled data from batteries with label = 0 in the historical data.

[0194] Step 2005: Save the database. Specifically, the transformation process data, down-sampling data, abnormal indicator data and labels can be saved in the database.

[0195] Step 2006: Extract data for historically produced and recently produced batteries from the database. For example, extract abnormal indicator data for historically produced batteries and abnormal indicator data for recently produced batteries at time intervals. For another example, extract labels for historically produced batteries and labels for recently produced batteries at time intervals.

[0196] Step 2007: Calculate statistics based on time to generate time series data. Specifically, (1) divide the hour into six time intervals, each 10 minutes long, and select the abnormal indicator data at the 25%, 50%, or 75% level within each time interval. Time intervals where the number of batteries is less than a preset value can be deleted, thereby forming a combined time series data. (2) Count the number and proportion of abnormal labels within different time intervals to form time series data.

[0197] Step 2008: Time series data mutation detection. Specifically, (1) Bayesian change point detection: Calculate the statistics of the abnormal indicator data belonging to the first 50 minutes in the time series data to obtain the distribution prior, and then import the Bayesian change point detection algorithm to detect whether there is a data mutation in the time series data. (2) Threshold control algorithm: Calculate the quality deviation by sliding the time window to detect whether there is a data mutation in the abnormal number or proportion of labels in the time series data.

[0198] Step 2009: If a data mutation is detected in the time series data in step 2008, a failure analysis of the battery may be performed. Specifically, step 2009 may include the following:

[0199] In step 3001, if the time series mutation detection algorithm identifies abnormal fluctuations in the time series data, relevant data for the failure mode, including process parameters and equipment data, is obtained. For example, if the failure mode is abnormal electrolyte injection, the relevant data for this failure mode includes process parameters related to the injection process, such as injection quality data and electrolyte replacement time.

[0200] Optionally, if the time series mutation detection algorithm does not identify abnormal fluctuations in the time series data, return to step 2006.

[0201] Step 3002 is to check whether the relevant data of the failure mode exceeds a set threshold, for example, to determine whether the electrolyte injection quality exceeds the specification (usually different electrolyte products have different densities, resulting in different injection quality).

[0202] Step 3003: If the relevant data of the failure mode exceeds the set threshold, the problem attribution is output and the abnormal battery is marked and intercepted. Optionally, in step 3003, the injection of electrolyte on the production line can be stopped and replaced with the correct electrolyte.

[0203] Optionally, if it is determined that the relevant data of the failure mode does not exceed the set threshold, the process returns to step 2006 .

[0204] In this embodiment, an anomaly detection model can be generated by training based on the down-sampling data of historically produced batteries that are within the anomaly recognition threshold range, and used to calculate the anomaly index data of newly produced batteries. While ensuring the stability of the data and the model, the self-adjustment of the model is achieved; and by proposing the key film formation segment data of the formation process data and adopting the down-sampling method, while maintaining the anomaly detection capability, the consumption of storage and computing resources can also be reduced; in addition, by extracting the anomaly index data at time intervals for statistical value calculation, the influence of individual anomaly data can be eliminated, making the data more stable, and the use of a time series data mutation detection algorithm has obvious effects on batch anomaly recognition; finally, if the alarm of the sudden detection is triggered, the failure analysis is started, and the extraction of relevant data for root cause analysis can lock the cause of the anomaly.

[0205] The above describes in detail the method for detecting batteries according to an embodiment of the present application. The following will describe the device for detecting batteries according to an embodiment of the present application in combination with Figures 14 and 15. The technical features described in the method embodiment are applicable to the following device embodiments.

[0206] Figure 14 shows a schematic block diagram of a battery detection device 4000 according to an embodiment of the present application. As shown in Figure 14, the device 4000 may include some or all of the following contents.

[0207] Acquisition module 4100 is used to obtain formation process data of each of N batteries to be tested, where N is a positive integer greater than 1. Acquisition module 4100 is also used to obtain abnormal indicator data of each battery to be tested based on the formation process data of each battery to be tested and an abnormality detection model, where the abnormality detection model is trained based on the formation process data of M tested batteries, where M is a positive integer greater than 1.

[0208] The detection module 4200 is used to detect whether N batteries to be tested have abnormalities based on the abnormal indicator data of the N batteries to be tested.

[0209] Optionally, as shown in Figure 15, the detection module 4200 includes: a generation submodule 4210, which is used to generate time series data based on the abnormal indicator data of N batteries to be tested and the abnormal indicator data of P tested batteries, where P is a positive integer greater than 1; and a detection submodule 4220, which is used to detect whether there are abnormalities in the N batteries to be tested based on the time series data.

[0210] Optionally, in an embodiment of the present application, the acquisition module 4100 is further used to: obtain first abnormal indicator data from the abnormal indicator data of N batteries to be tested; obtain at least one second abnormal indicator data from the abnormal indicator data of P tested batteries; the generation submodule 4210 is specifically used to: generate time series data based on the first abnormal indicator data and at least one second abnormal indicator data.

[0211] Optionally, in an embodiment of the present application, abnormal indicator data of N batteries to be tested are obtained within a first time interval, and abnormal indicator data of P tested batteries are obtained within a second time interval, the second time interval includes at least one time series sequence, and the acquisition module 4100 is specifically used to: obtain first abnormal indicator data from the first time interval; obtain a second abnormal indicator data from each time series sequence in at least one time series sequence.

[0212] Optionally, in an embodiment of the present application, the second time interval is divided into at least one time series according to a preset time length, and the acquisition module 4100 is specifically used to: within the first time interval, count the abnormal indicator data of N batteries to be tested; according to a preset statistical quantile, obtain the first abnormal indicator data from the first time interval; within each time series, count the abnormal indicator data of the tested battery corresponding to each time series; according to the preset statistical quantile, obtain a second abnormal indicator data from each time series.

[0213] Optionally, in an embodiment of the present application, the acquisition module 4100 is specifically used to: when N is greater than a first preset threshold, obtain first abnormal indicator data from a first time interval; when the number of measured batteries corresponding to each time series is greater than the first preset threshold, obtain a second abnormal indicator data from each time series.

[0214] Optionally, as shown in Figure 15, the detection module 4200 also includes: a determination submodule 4030, which is used to determine the labels of N batteries to be tested based on the abnormal indicator data of the N batteries to be tested, where the label is an abnormal label used to indicate that the battery has an abnormality or a normal label used to indicate that the battery does not have an abnormality; the generation submodule 4210 is also used to: generate timing data based on the labels of the N batteries to be tested and the labels of P tested batteries, where P is a positive integer greater than 1; the detection submodule 4220 is also used to: detect whether the N batteries to be tested have an abnormality based on the timing data.

[0215] Optionally, in an embodiment of the present application, the labels of N batteries to be tested are obtained within a first time interval, and the labels of P tested batteries are obtained within a second time interval. The second time interval is divided into at least one time series sequence according to a preset time length. The generation submodule 4210 is specifically used to: within the first time interval, count the number of abnormal labels in the labels of the N batteries to be tested; within each time series sequence in at least one time series sequence, count the number of abnormal labels in the labels of the tested batteries corresponding to each time series sequence; generate time series data based on the number of abnormal labels counted within the first time interval and the number of abnormal labels counted within each time series sequence.

[0216] Optionally, in an embodiment of the present application, the labels of N batteries to be tested are obtained within a first time interval, and the labels of P tested batteries are obtained within a second time interval. The second time interval is divided into at least one time series according to a preset time length. The generation submodule 4210 is specifically used to: calculate the proportion of abnormal labels in the labels of the N batteries to be tested within the first time interval; calculate the proportion of abnormal labels in the labels of the tested batteries in each time series in at least one time series; and generate time series data based on the proportion of abnormal labels in the first time interval and the proportion of abnormal labels in each time series.

[0217] Optionally, in an embodiment of the present application, the acquisition module 4100 is specifically used to: obtain the key film formation segment data of each battery to be tested from the formation process data of each battery to be tested, the key film formation segment data refers to all formation process data of the battery between zero and a second preset threshold value during the formation process, and the second preset threshold value is related to the reaction potential of the additive expected to be added to the electrolyte of the N batteries to be tested; according to the key film formation segment data of each battery to be tested and the anomaly detection model, obtain the abnormal index data of each battery to be tested, and the anomaly detection model is obtained by training based on the key film formation segment data of M tested batteries.

[0218] Optionally, in an embodiment of the present application, the acquisition module 4100 is specifically used to: use a sliding window method to obtain a set of formation process data with the highest similarity to the standard film formation section curve from the film formation key section data of each battery to be tested, the standard film formation section curve is formed by fitting based on the film formation key section data of L tested batteries, and L is a positive integer greater than 1; based on a set of formation process data of each battery to be tested and an anomaly detection model, obtain the abnormal index data of each battery to be tested.

[0219] Optionally, in an embodiment of the present application, the acquisition module 4100 is specifically used to: downsample a set of formation process data of each battery to be tested and generate downsampled data for each battery to be tested; input the downsampled data of each battery to be tested into an anomaly detection model, and output the abnormal index data of each battery to be tested. The anomaly detection model is obtained by training based on the downsampled data of M tested batteries, and the standard film formation section curve is formed by fitting based on the downsampled data of L tested batteries.

[0220] Optionally, in an embodiment of the present application, the down-sampled data of the i-th battery to be tested among the N batteries to be tested and the down-sampled data of each tested battery among the M tested batteries both include Q data, and the Q data of the i-th battery to be tested correspond one-to-one to the Q data of each tested battery, i is any positive integer from 1 to N, and Q is a positive integer greater than 1. The acquisition module 4100 is specifically used to: calculate M distances between the j-th data in the Q data of the i-th battery to be tested and the corresponding data in the Q data of each tested battery, j is any positive integer from 1 to Q; determine the K data with the closest distance from the M data corresponding to the j-th data, where K is a positive integer; determine the average distance of the j-th data based on the distances from the K data to the j-th data respectively; and determine the abnormal indicator data of the i-th battery to be tested based on each data in the Q data of the i-th battery to be tested and the average distance of each data.

[0221] Optionally, in the embodiment of the present application, the detection submodule 4220 is specifically used to: perform change point detection on the time series data to obtain a change point detection result; and determine whether N batteries to be tested have abnormalities based on the change point detection result.

[0222] Optionally, in an embodiment of the present application, the detection submodule 4220 is specifically used to: obtain a prior probability distribution based on the data of the measured batteries in the time series data; obtain a new probability distribution based on the data of the batteries to be tested in the time series data and the data of the measured batteries in the time series data; and obtain a change point detection result based on the prior probability distribution and the new probability distribution.

[0223] Optionally, in an embodiment of the present application, the detection submodule 4220 is specifically used to: when the prior probability distribution and the new probability distribution are different, determine that the change point detection result is that the data of the battery to be tested in the time series data has a mutation compared with the data of the tested battery.

[0224] Optionally, in an embodiment of the present application, the detection submodule 4220 is specifically used to: respectively calculate the difference between two adjacent data in the time series data; and obtain a change point detection result based on the relationship between the difference and a third preset threshold.

[0225] Optionally, in an embodiment of the present application, the acquisition module 4100 is further used to: when an abnormality is detected in N batteries to be tested, obtain data related to the electrolyte injection process when producing N batteries to be tested; the device 4000 also includes: a determination module for determining whether there is a mismatch between the data related to the electrolyte injection process and a preset control range; the determination module is further used to: when there is a mismatch between the data related to the electrolyte injection process and the preset control range, determine that an electrolyte injection abnormality has occurred in the electrolyte of the N batteries to be tested.

[0226] Optionally, in the embodiment of the present application, the apparatus 4000 further includes: a generating module configured to generate electrolyte abnormality cause information of the N batteries to be tested when it is determined that electrolyte injection abnormality occurs in the N batteries to be tested.

[0227] Optionally, in an embodiment of the present application, the generating module is further configured to generate alarm information based on the electrolyte abnormality cause information to intercept N batteries to be tested.

[0228] It should be understood that the device 4000 according to the embodiment of the present application may correspond to the execution entity in the embodiment of the method 100 of the present application, and the above-mentioned and other operations and / or functions of each unit in the device 4000 are respectively for realizing the corresponding processes in each method of Figures 1 to 12. For the sake of brevity, they will not be repeated here.

[0229] As shown in Figure 16, an embodiment of the present application further provides an electronic device 5000, which may be the apparatus 4000 in Figures 14 to 15, and which can be used to execute the contents of each method in Figures 2 to 13. The electronic device 5000 shown in Figure 16 includes a processor 5010, which can call and execute a computer program from a memory to implement the method in the embodiment of the present application.

[0230] Optionally, as shown in FIG16 , the electronic device 5000 may further include a memory 5020. The processor 5010 may call and execute a computer program from the memory 5020 to implement the method in the embodiment of the present application.

[0231] The memory 5020 may be a separate device independent of the processor 5010 , or may be integrated into the processor 2010 .

[0232] Optionally, as shown in FIG16 , the electronic device 5000 may further include a transceiver 5030 , and the processor 5010 may control the transceiver 5030 to communicate with other devices, specifically, to send information or data to other devices, or to receive information or data sent by other devices.

[0233] The transceiver 5030 may include a transmitter and a receiver. The transceiver 2030 may further include an antenna, and the number of antennas may be one or more.

[0234] In a specific implementation, the acquisition module and the detection module in the device 4000 may be implemented by the processor 5010 in FIG. 16 .

[0235] Optionally, an embodiment of the present application further provides a chip, including a processor, which can call and run a computer program from a memory to implement the method in the embodiment of the present application.

[0236] Optionally, an embodiment of the present application further provides a computer-readable storage medium for storing a computer program, which enables a computer to execute the method in the embodiment of the present application.

[0237] Optionally, an embodiment of the present application further provides a computer program product, including computer program instructions, which enable a computer to execute the method in the embodiment of the present application.

[0238] Optionally, the embodiments of the present application further provide a computer program that enables a computer to execute the method in the embodiments of the present application.

[0239] It should be understood that the processor of the embodiments of the present application may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method embodiment can be completed by hardware integrated logic circuits in the processor or software instructions. The above processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The various methods, steps, and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in the embodiments of the present application can be directly embodied as being executed by a hardware decoding processor, or can be executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium mature in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0240] It is understood that the memory in the embodiments of the present application may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct RAM bus random access memory (DR RAM). It should be noted that the memory of the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0241] Although the present application has been described with reference to preferred embodiments, various modifications may be made thereto and components may be substituted with equivalents without departing from the scope of the present application. In particular, the various technical features described in the various embodiments may be combined in any manner as long as there are no structural conflicts. The present application is not limited to the specific embodiments disclosed herein, but encompasses all technical solutions within the scope of the claims.

Claims

1. A method for detecting a battery, characterized in that: include: Obtaining formation process data of each of N batteries to be tested, where N is a positive integer greater than 1; Obtaining abnormal index data of each battery to be tested according to the formation process data of each battery to be tested and an abnormality detection model, wherein the abnormality detection model is trained based on the formation process data of M tested batteries, where M is a positive integer greater than 1; According to the abnormal indicator data of the N batteries to be tested, it is detected whether the N batteries to be tested have abnormalities.

2. The method according to claim 1, characterized in that The detecting whether the N batteries to be tested have abnormalities according to the abnormal indicator data of the N batteries to be tested includes: Generate time series data based on the abnormal indicator data of the N batteries to be tested and the abnormal indicator data of the P tested batteries, where P is a positive integer greater than 1; According to the time series data, it is detected whether the N batteries to be tested have any abnormality.

3. The method according to claim 2, characterized in that The generating of time series data according to the abnormal indicator data of the N batteries to be tested and the abnormal indicator data of the P tested batteries includes: Acquire first abnormal indicator data from the abnormal indicator data of the N batteries to be tested; Obtain at least one second abnormal indicator data from the abnormal indicator data of the P tested batteries; The time series data is generated according to the first abnormality indicator data and the at least one second abnormality indicator data.

4. The method according to claim 3, characterized in that The abnormal indicator data of the N batteries to be tested is obtained in a first time interval, and the abnormal indicator data of the P tested batteries is obtained in a second time interval, the second time interval includes at least one time series, and the obtaining of first abnormal indicator data from the abnormal indicator data of the N batteries to be tested includes: Acquire the first abnormality indicator data from the first time interval; The obtaining at least one second abnormal indicator data from the abnormal indicator data of the P tested batteries includes: A second abnormality indicator data is obtained from each time series in the at least one time series.

5. The method according to claim 4, characterized in that The second time interval is divided into the at least one time series according to a preset time length, and the obtaining of the first abnormality indicator data from the first time interval includes: In the first time interval, collecting statistics on abnormal indicator data of the N batteries to be tested; Obtaining the first abnormal indicator data from the first time interval according to a preset statistical quantile; The acquiring of a second abnormality indicator data from each time series in the at least one time series respectively includes: In each of the time series, statistics are collected on abnormal indicator data of the tested batteries corresponding to each of the time series; According to the preset statistical quantile, a second abnormal indicator data is obtained from each time series.

6. The method according to claim 4 or 5, characterized in that The acquiring the first abnormality indicator data from the first time interval includes: When N is greater than a first preset threshold, obtaining the first abnormality indicator data from the first time interval; The acquiring of a second abnormality indicator data from each time series in the at least one time series respectively includes: When the number of tested batteries corresponding to each time series is greater than the first preset threshold, a second abnormality indicator data is obtained from each time series.

7. The method according to claim 1, characterized in that The detecting whether the N batteries to be tested have abnormalities according to the abnormal indicator data of the N batteries to be tested includes: Determining labels of the N batteries to be tested based on the abnormality indicator data of the N batteries to be tested, where the labels are abnormal labels indicating that the batteries are abnormal or normal labels indicating that the batteries are not abnormal; Generate time series data according to the labels of the N batteries to be tested and the labels of the P tested batteries, where P is a positive integer greater than 1; According to the time series data, it is detected whether the N batteries to be tested have any abnormality.

8. The method according to claim 7, characterized in that The tags of the N batteries to be tested are acquired within a first time interval, and the tags of the P tested batteries are acquired within a second time interval, the second time interval being divided into at least one time series according to a preset duration, and generating time series data based on the tags of the N batteries to be tested and the tags of the P tested batteries, including: Counting the number of abnormal tags among the tags of the N batteries to be tested within the first time interval; In each of the at least one time series, respectively counting the number of abnormal tags among the tags of the tested batteries corresponding to each time series; The time series data is generated according to the number of abnormal labels counted in the first time interval and the number of abnormal labels counted in each time series sequence.

9. The method according to claim 7, characterized in that The tags of the N batteries to be tested are acquired within a first time interval, and the tags of the P tested batteries are acquired within a second time interval, the second time interval being divided into at least one time series according to a preset duration, and generating time series data based on the tags of the N batteries to be tested and the tags of the P tested batteries, including: Calculating the proportion of abnormal tags among the tags of the N batteries to be tested within the first time interval; respectively calculating the proportion of abnormal labels among the labels of the tested batteries in each time series of the at least one time series; The time series data is generated according to the proportion of abnormal labels in the first time interval and the proportion of abnormal labels in each time series.

10. The method according to any one of claims 1 to 9, characterized in that The step of obtaining abnormality indicator data of each battery to be tested based on the formation process data of each battery to be tested and the abnormality detection model includes: Obtaining, from the formation process data of each battery to be tested, film formation key segment data of each battery to be tested, respectively, wherein the film formation key segment data refers to all formation process data of the battery between a voltage of zero and a second preset threshold value during the formation process, wherein the second preset threshold value is related to a reaction potential of an additive expected to be added to the electrolyte of the N batteries to be tested; Abnormal index data of each battery to be tested is obtained based on the film formation key segment data of each battery to be tested and the abnormality detection model, and the abnormality detection model is obtained by training based on the film formation key segment data of the M tested batteries.

11. The method according to claim 10, characterized in that The step of obtaining abnormality index data of each battery to be tested based on the film formation key segment data of each battery to be tested and the abnormality detection model includes: Using a sliding window method, a set of formation process data with the highest similarity to a standard film formation curve is obtained from the key film formation data of each battery to be tested, wherein the standard film formation curve is formed by fitting the key film formation data of L tested batteries, where L is a positive integer greater than 1; Abnormal indicator data of each battery to be tested is obtained according to the set of formation process data of each battery to be tested and the abnormality detection model.

12. The method according to claim 11, characterized in that The step of obtaining abnormality indicator data of each battery to be tested based on the set of formation process data of each battery to be tested and the abnormality detection model includes: Downsampling the set of formation process data of each battery to be tested and generating downsampled data of each battery to be tested; The down-sampling data of each battery to be tested is input into the anomaly detection model, and the abnormal index data of each battery to be tested is output. The anomaly detection model is trained based on the down-sampling data of the M tested batteries, and the standard film-forming section curve is formed by fitting based on the down-sampling data of the L tested batteries.

13. The method according to claim 12, characterized in that The downsampled data of the i-th battery to be tested among the N batteries to be tested and the downsampled data of each tested battery among the M tested batteries each include Q data, the Q data of the i-th battery to be tested corresponding to the Q data of each tested battery in a one-to-one manner, i is any positive integer from 1 to N, and Q is a positive integer greater than 1, and the downsampled data of each battery to be tested is input into the abnormality detection model, and the abnormality indicator data of each battery to be tested is output, including: Calculate M distances between the jth data in the Q data of the i-th battery to be tested and the corresponding data in the Q data of each tested battery, where j is any positive integer from 1 to Q; Determine the K data closest to the j-th data from the M data corresponding to the j-th data, where K is a positive integer; Determine the average distance of the j-th data according to the distances from the K data to the j-th data respectively; Abnormal indicator data of the i-th battery to be tested is determined according to each data in the Q data of the i-th battery to be tested and the average distance between each data.

14. The method according to any one of claims 2 to 13, characterized in that Detecting whether the N batteries to be tested have abnormalities according to the time series data includes: Performing change point detection on the time series data to obtain a change point detection result; According to the change point detection result, it is determined whether the N batteries to be tested have abnormalities.

15. The method according to claim 14, characterized in that The performing change point detection on the time series data to obtain a change point detection result includes: Obtaining a priori probability distribution based on data of the measured batteries in the time series data; Obtaining a new probability distribution according to data of the battery to be tested in the time series data and data of the tested batteries in the time series data; The change point detection result is obtained according to the prior probability distribution and the new probability distribution.

16. The method according to claim 15, characterized in that The obtaining the change point detection result according to the prior probability distribution and the new probability distribution includes: When the prior probability distribution and the new probability distribution are different, the change point detection result is determined to be that the data of the battery to be tested in the time series data has a sudden change compared with the data of the tested batteries.

17. The method according to claim 14, characterized in that The performing change point detection on the time series data to obtain a change point detection result includes: respectively calculating the difference between two adjacent data in the time series data; The change point detection result is obtained according to the relationship between the difference and a third preset threshold.

18. The method according to any one of claims 1 to 17, characterized in that The method further comprises: When an abnormality is detected in the N batteries to be tested, obtaining data related to an electrolyte injection process during the production of the N batteries to be tested; determining whether data related to the electrolyte injection process does not match a preset control range; When the data related to the electrolyte injection process does not match the preset control range, it is determined that electrolyte injection abnormality occurs in the electrolytes of the N batteries to be tested.

19. The method according to claim 18, characterized in that The method further comprises: When it is determined that electrolyte injection abnormality occurs in the N batteries to be tested, electrolyte abnormality cause information of the N batteries to be tested is generated.

20. The method according to claim 19, characterized in that The method further comprises: According to the electrolyte abnormality cause information, alarm information is generated to intercept the N batteries to be tested.

21. A device for detecting a battery, characterized in that: include: An acquisition module, configured to acquire formation process data of each of N batteries to be tested, where N is a positive integer greater than 1; The acquisition module is further configured to respectively acquire abnormality indicator data of each battery to be tested based on the formation process data of each battery to be tested and an abnormality detection model, wherein the abnormality detection model is trained based on the formation process data of M tested batteries, where M is a positive integer greater than 1; The detection module is used to detect whether the N batteries to be tested have abnormalities according to the abnormal indicator data of the N batteries to be tested.

22. The device according to claim 21, characterized in that The detection module includes: A generating submodule, configured to generate time series data based on the abnormal indicator data of the N batteries to be tested and the abnormal indicator data of the P tested batteries, where P is a positive integer greater than 1; The detection submodule is used to detect whether the N batteries to be tested have any abnormalities according to the time series data.

23. The device according to claim 22, characterized in that The acquisition module is further used for: Acquire first abnormal indicator data from the abnormal indicator data of the N batteries to be tested; Obtain at least one second abnormal indicator data from the abnormal indicator data of the P tested batteries; The generation submodule is specifically used for: The time series data is generated according to the first abnormality indicator data and the at least one second abnormality indicator data.

24. The device according to claim 23, characterized in that The abnormal indicator data of the N batteries to be tested are acquired in a first time interval, and the abnormal indicator data of the P tested batteries are acquired in a second time interval, wherein the second time interval includes at least one time series. The acquisition module is specifically configured to: Acquire the first abnormality indicator data from the first time interval; A second abnormality indicator data is obtained from each time series in the at least one time series.

25. The device according to claim 24, characterized in that The second time interval is divided into the at least one time sequence according to a preset duration, and the acquisition module is specifically configured to: In the first time interval, collecting statistics on abnormal indicator data of the N batteries to be tested; Obtaining the first abnormal indicator data from the first time interval according to a preset statistical quantile; In each of the time series, statistics are collected on abnormal indicator data of the tested batteries corresponding to each of the time series; According to the preset statistical quantile, a second abnormal indicator data is obtained from each time series.

26. The device according to claim 24 or 25, characterized in that The acquisition module is specifically used for: When N is greater than a first preset threshold, obtaining the first abnormality indicator data from the first time interval; When the number of tested batteries corresponding to each time series is greater than the first preset threshold, a second abnormality indicator data is obtained from each time series.

27. The device according to claim 21, characterized in that The detection module includes: a determination submodule, configured to determine labels of the N batteries to be tested based on the abnormality indicator data of the N batteries to be tested, wherein the labels are abnormal labels indicating that the batteries are abnormal or normal labels indicating that the batteries are not abnormal; A generating submodule, configured to generate time series data according to the labels of the N batteries to be tested and the labels of the P tested batteries, where P is a positive integer greater than 1; The detection submodule is used to detect whether the N batteries to be tested have any abnormalities according to the time series data.

28. The device according to claim 27, characterized in that The labels of the N batteries to be tested are obtained within a first time interval, and the labels of the P tested batteries are obtained within a second time interval. The second time interval is divided into at least one time sequence according to a preset duration. The generating submodule is specifically configured to: Counting the number of abnormal tags among the tags of the N batteries to be tested within the first time interval; In each of the at least one time series, respectively counting the number of abnormal tags among the tags of the tested batteries corresponding to each time series; The time series data is generated according to the number of abnormal labels counted in the first time interval and the number of abnormal labels counted in each time series sequence.

29. The device according to claim 27, characterized in that The labels of the N batteries to be tested are obtained within a first time interval, and the labels of the P tested batteries are obtained within a second time interval. The second time interval is divided into at least one time sequence according to a preset duration. The generating submodule is specifically configured to: Calculating the proportion of abnormal tags among the tags of the N batteries to be tested within the first time interval; respectively calculating the proportion of abnormal labels among the labels of the tested batteries in each time series of the at least one time series; The time series data is generated according to the proportion of abnormal labels in the first time interval and the proportion of abnormal labels in each time series.

30. The device according to any one of claims 21 to 29, characterized in that The acquisition module is specifically used for: Obtaining, from the formation process data of each battery to be tested, film formation key segment data of each battery to be tested, respectively, wherein the film formation key segment data refers to all formation process data of the battery between a voltage of zero and a second preset threshold value during the formation process, wherein the second preset threshold value is related to a reaction potential of an additive expected to be added to the electrolyte of the N batteries to be tested; Abnormal index data of each battery to be tested is obtained based on the film formation key segment data of each battery to be tested and the abnormality detection model, and the abnormality detection model is obtained by training based on the film formation key segment data of the M tested batteries.

31. The device according to claim 30, characterized in that The acquisition module is specifically used for: Using a sliding window method, a set of formation process data with the highest similarity to a standard film formation curve is obtained from the key film formation data of each battery to be tested, wherein the standard film formation curve is formed by fitting the key film formation data of L tested batteries, where L is a positive integer greater than 1; Abnormal indicator data of each battery to be tested is obtained according to the set of formation process data of each battery to be tested and the abnormality detection model.

32. The device according to claim 31, characterized in that The acquisition module is specifically used for: Downsampling the set of formation process data of each battery to be tested and generating downsampled data of each battery to be tested; The down-sampling data of each battery to be tested is input into the anomaly detection model, and the abnormal index data of each battery to be tested is output. The anomaly detection model is trained based on the down-sampling data of the M tested batteries, and the standard film-forming section curve is formed by fitting based on the down-sampling data of the L tested batteries.

33. The device according to claim 32, characterized in that The downsampled data of the i-th battery to be tested among the N batteries to be tested and the downsampled data of each tested battery among the M tested batteries each include Q data, and the Q data of the i-th battery to be tested correspond one-to-one to the Q data of each tested battery, where i is any positive integer from 1 to N, and Q is a positive integer greater than 1. The acquisition module is specifically configured to: Calculate M distances between the jth data in the Q data of the i-th battery to be tested and the corresponding data in the Q data of each tested battery, where j is any positive integer from 1 to Q; Determine the K data closest to the j-th data from the M data corresponding to the j-th data, where K is a positive integer; Determine the average distance of the j-th data according to the distances from the K data to the j-th data respectively; Abnormal indicator data of the i-th battery to be tested is determined according to each data in the Q data of the i-th battery to be tested and the average distance between each data.

34. The device according to any one of claims 22 to 33, characterized in that The detection submodule is specifically used for: Performing change point detection on the time series data to obtain a change point detection result; According to the change point detection result, it is determined whether the N batteries to be tested have abnormalities.

35. The device according to claim 34, characterized in that The detection submodule is specifically used for: Obtaining a priori probability distribution based on data of the measured batteries in the time series data; Obtaining a new probability distribution according to data of the battery to be tested in the time series data and data of the tested batteries in the time series data; The change point detection result is obtained according to the prior probability distribution and the new probability distribution.

36. The device according to claim 35, characterized in that The detection submodule is specifically used for: When the prior probability distribution and the new probability distribution are different, the change point detection result is determined to be that the data of the battery to be tested in the time series data has a sudden change compared with the data of the tested batteries.

37. The device according to claim 34, characterized in that The detection submodule is specifically used for: respectively calculating the difference between two adjacent data in the time series data; The change point detection result is obtained according to the relationship between the difference and a third preset threshold.

38. The device according to any one of claims 21 to 37, characterized in that The acquisition module is further used for: When an abnormality is detected in the N batteries to be tested, obtaining data related to an electrolyte injection process during the production of the N batteries to be tested; The device further comprises: a determination module, configured to determine whether data related to the electrolyte injection process does not match a preset control range; The determining module is further configured to: When the data related to the electrolyte injection process does not match the preset control range, it is determined that electrolyte injection abnormality occurs in the electrolytes of the N batteries to be tested.

39. The device according to claim 38, characterized in that The device further comprises: The generating module is configured to generate electrolyte abnormality cause information of the N batteries to be tested when it is determined that electrolyte injection abnormality occurs in the N batteries to be tested.

40. The device according to claim 39, characterized in that The generation module is further configured to: According to the electrolyte abnormality cause information, alarm information is generated to intercept the N batteries to be tested.

41. An electronic device, characterized in that: The electronic device includes a memory and a processor, wherein the memory is used to store instructions, and the processor is used to read the instructions and execute the method according to any one of claims 1 to 20 according to the instructions.

42. A storage medium, characterized in that Used to store a computer program, wherein the computer program causes a computer to execute the method according to any one of claims 1 to 20.

43. A program product, characterized in that The invention comprises a computer-readable code or a non-volatile computer-readable storage medium carrying the computer-readable code, wherein when the computer-readable code is executed in an electronic device, a processor in the electronic device executes the method according to any one of claims 1 to 20.

Images