Battery pack fault diagnosis method and device, electronic equipment, readable storage medium and computer program product

Abstract

The application discloses a battery pack fault diagnosis method. The method comprises the following steps: obtaining original voltage sequences of each battery cell of an electric vehicle battery pack in a driving state and a charging state; preprocessing the original voltage sequences to obtain preprocessed voltage sequences; calculating a normalized two-dimensional voltage feature vector corresponding to each sliding window according to a preset sliding window and the preprocessed voltage sequences; performing dynamic identification on each battery cell of the electric vehicle battery pack according to preset algorithm parameters, a density clustering algorithm and the normalized two-dimensional voltage feature vector to obtain an abnormal cluster identification result; and performing corresponding abnormal battery cell fault early warning operation on the abnormal cluster identification result according to a preset multi-level risk threshold. The method can detect the battery pack fault of the electric vehicle in the driving process online, accurately locate the fault battery cell, improve the response rate, avoid the dependence on offline data, thereby avoiding the occurrence of thermal runaway, and helping subsequent fault maintenance.

Description

Technical Field

[0001] This application relates to the field of vehicle fault diagnosis technology, specifically to a battery pack fault diagnosis method, device, electronic equipment, readable storage medium, and computer program product. Background Technology

[0002] Currently, the electric vehicle industry is booming at an unprecedented pace. As a core component of electric vehicles, the safety of battery packs is becoming increasingly prominent, posing a critical challenge that the entire industry urgently needs to overcome. Existing battery pack fault detection solutions typically involve pre-setting thresholds for key parameters such as voltage and temperature. Once the monitored data exceeds these thresholds, the system triggers a fault alarm. However, in practical applications, this detection method has significant shortcomings. It can only assess the overall abnormality of the battery pack and cannot accurately pinpoint specific anomalies. Moreover, the setting of thresholds heavily relies on engineers' experience and lacks scientific and precise basis, which significantly reduces the sensitivity of anomaly identification. Therefore, existing methods suffer from insufficient fault location accuracy and a high false alarm rate, seriously affecting the accuracy and timeliness of battery pack fault detection and posing potential risks to the safe and stable operation of electric vehicles. Summary of the Invention

[0003] In view of the above problems, this application provides a battery pack fault diagnosis method, device, electronic device, readable storage medium and computer program product, which can solve the problems of insufficient fault location accuracy, high false alarm rate and insufficient accuracy and timeliness of battery pack fault detection.

[0004] In a first aspect, this application provides a battery pack fault diagnosis method, including: Obtain the original voltage sequence of each cell in the electric vehicle battery pack under driving and charging conditions; The original voltage sequence is preprocessed to obtain a preprocessed voltage sequence; Based on the preset sliding window and the preprocessed voltage sequence, calculate the normalized two-dimensional voltage feature vector corresponding to each sliding window; Based on preset algorithm parameters, density clustering algorithm and the normalized two-dimensional voltage feature vector, the cells of the electric vehicle battery pack are dynamically identified to obtain the abnormal cluster identification results. Based on preset multi-level risk thresholds, the abnormal cluster identification results are used to perform corresponding abnormal cell fault early warning operations.

[0005] In the above technical solution, the method can detect battery pack faults in electric vehicles online in real time, accurately locate faulty cells, and realize multi-level risk warnings, thereby improving the accuracy and timeliness of fault diagnosis and helping to avoid safety accidents such as thermal runaway.

[0006] In some embodiments, the preprocessing of the original voltage sequence to obtain a preprocessed voltage sequence includes: The original voltage sequence is reordered in descending order of timestamps to obtain the first processed sequence; The first processing sequence is processed for missing data and outliers to obtain the second processing sequence; The second processed sequence is logarithmically transformed to obtain the preprocessed voltage sequence.

[0007] In the above technical solution, the method can effectively handle the problems of timestamp sorting errors, missing data and outliers in the original voltage sequence, thereby obtaining a high-quality preprocessed voltage sequence and providing a reliable data foundation for subsequent fault diagnosis.

[0008] In some implementations, calculating the normalized two-dimensional voltage feature vector corresponding to each sliding window based on a preset sliding window and the preprocessed voltage sequence includes: The preprocessed voltage sequence is divided by a preset sliding window to obtain multiple voltage sub-sequences; Calculate the voltage consistency feature and modified Shannon entropy feature corresponding to each sliding window based on the multiple voltage subsequences; Based on the voltage consistency feature and the modified Shannon entropy feature, construct a two-dimensional feature matrix corresponding to each sliding window; The two-dimensional feature matrix is ​​normalized to obtain the two-dimensional normalized feature matrix corresponding to each sliding window.

[0009] In the above technical solution, the method can effectively extract key features from the preprocessed voltage sequence, and provide representative feature data for subsequent density clustering analysis and fault diagnosis by constructing and normalizing a two-dimensional feature matrix.

[0010] In some implementations, the step of performing corresponding abnormal cell fault early warning operations based on the abnormal cluster identification results according to preset multi-level risk thresholds includes: Based on the abnormal cluster identification results, the frequency of each battery cell being marked as an abnormal cluster is statistically analyzed to obtain the abnormal marking frequency corresponding to each battery cell. The abnormal cell identification result is determined based on the abnormal marking frequency and the preset multi-level risk threshold; wherein, the abnormal cell identification result includes low-risk abnormal cells, medium-risk abnormal cells, and high-risk abnormal cells; wherein, the preset multi-level risk threshold includes a low-risk threshold, a medium-risk threshold, and a high-risk threshold; The output includes abnormal cell alert information that includes the abnormal cell identification results.

[0011] In the above technical solution, the method can realize multi-level early warning and refined management of battery pack failures.

[0012] In some embodiments, the method further includes: The high-risk abnormal battery cells were continuously monitored, and the monitoring results were obtained. When thermal runaway is detected in the high-risk abnormal battery cell based on the monitoring results, a thermal runaway alarm message is output.

[0013] In the above technical solution, the method can monitor high-risk abnormal cells in real time and continuously, and output alarm information in a timely manner when serious faults such as thermal runaway are detected, thereby effectively preventing battery pack safety accidents and ensuring the safe operation of electric vehicles.

[0014] In some implementations, the preset algorithm parameters include a preset neighborhood radius and a preset minimum number of samples; The preset algorithm parameters are obtained by dynamically adjusting the original algorithm parameters based on historical vehicle data, battery type, vehicle driving conditions, and vehicle working environment. The original algorithm parameters include the pre-calibrated original neighborhood radius and the pre-calibrated original minimum number of samples.

[0015] In the above technical solution, the method can dynamically adjust the original algorithm parameters to optimize the preset algorithm parameters, thereby improving the accuracy of density clustering algorithm in battery pack fault diagnosis.

[0016] Secondly, this application provides a battery pack fault diagnosis device, comprising: The acquisition unit is used to acquire the original voltage sequence of each cell in the electric vehicle battery pack under driving and charging conditions; A preprocessing unit is used to preprocess the original voltage sequence to obtain a preprocessed voltage sequence; The feature calculation unit is used to calculate the normalized two-dimensional voltage feature vector corresponding to each sliding window based on the preset sliding window and the preprocessed voltage sequence. An anomaly identification unit is used to dynamically identify each cell of the electric vehicle battery pack according to preset algorithm parameters, density clustering algorithm and the normalized two-dimensional voltage feature vector, and obtain anomaly cluster identification results. The early warning unit is used to perform corresponding abnormal cell fault early warning operations based on the abnormal cluster identification results according to preset multi-level risk thresholds.

[0017] In the above technical solution, the device can detect battery pack faults in electric vehicles online in real time, accurately locate faulty cells, and realize multi-level risk warnings, thereby improving the accuracy and timeliness of fault diagnosis and helping to avoid safety accidents such as thermal runaway.

[0018] Thirdly, this application provides an electronic device including a memory and a processor, the memory storing a computer program, and the processor running the computer program to cause the electronic device to perform the battery pack fault diagnosis method described in any one of the first aspects.

[0019] Fourthly, this application provides a readable storage medium storing a computer program, which, when executed by a processor, performs the battery pack fault diagnosis method described in any one of the first aspects.

[0020] Fifthly, this application provides a computer program product, which includes a computer program that, when executed by a processor, performs the battery pack fault diagnosis method described in any one of the first aspects.

[0021] The beneficial effects of this application are: it can detect battery pack faults in electric vehicles during operation online and accurately locate the faulty cells, while improving the response rate and avoiding dependence on offline data, thereby preventing thermal runaway and facilitating subsequent fault repair. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating a battery pack fault diagnosis method in some embodiments of this application; Figure 2 This is a schematic diagram of density-based voltage fault clustering results in some embodiments of this application; Figure 3 This is a flowchart illustrating a battery pack fault diagnosis method in some embodiments of this application; Figure 4 This is a schematic diagram of the structure of a battery pack fault diagnosis device in some embodiments of this application; Figure 5 This is a schematic diagram of the structure of an electronic device in some embodiments of this application. Detailed Implementation

[0024] The embodiments of the technical solution of this application will now be described in detail with reference to the accompanying drawings. These embodiments are only used to more clearly illustrate the technical solution of this application and are therefore merely examples, and should not be used to limit the scope of protection of this application.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms “comprising” and “having”, and any variations thereof, in the specification, claims, and foregoing description of the drawings are intended to cover non-exclusive inclusion.

[0026] In the description of the embodiments of this application, technical terms such as "first" and "second" are used only to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly specifying the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, "multiple" means two or more (including two), similarly, "multiple sets" refers to two or more sets (including two sets), and "multiple pieces" refers to two or more pieces (including two pieces) unless otherwise explicitly defined.

[0027] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0028] In the description of the embodiments in this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0029] With the increasing popularity of electric vehicles, battery pack safety has become a core challenge for the industry. Currently, widely adopted battery pack fault detection solutions mainly include rule-based judgment methods based on fixed thresholds and offline detection methods based on traditional clustering algorithms. The former triggers fault alarms by setting preset voltage / temperature thresholds (e.g., an alarm is triggered when the cell voltage exceeds the average voltage by ±50mV). Its core logic relies on engineers' experience-based settings, and while technically simple to implement, its sensitivity is low. The latter uses K-means clustering and other methods to perform offline analysis of historical charge-discharge data, identifying abnormal segments through voltage curve morphology matching. However, it requires complete charge-discharge cycle data and cannot respond to sudden faults during driving in real time. Both methods focus on overall battery pack status assessment, lacking a cell-level fault location mechanism, and their feature extraction dimensions are limited, typically using only basic statistics such as voltage mean and range.

[0030] Through benchmarking analysis, this application proposes an online fault diagnosis method based on outlier clustering to address the shortcomings of existing technologies, such as strong threshold dependence, offline detection lag, insufficient fault location accuracy, and high false alarm rate.

[0031] By implementing this method, rapid and accurate fault detection and cell-level fault location can be achieved through multidimensional feature extraction and density clustering technology, thereby systematically solving the problems of lag, missed detection rate and false alarm in existing solutions.

[0032] like Figure 1 As shown, some embodiments of this application provide a battery pack fault diagnosis method, which includes: S101. Obtain the original voltage sequence of each cell in the electric vehicle battery pack under driving and charging conditions; S102. Preprocess the original voltage sequence to obtain a preprocessed voltage sequence; S103. Based on the preset sliding window and the preprocessed voltage sequence, calculate the normalized two-dimensional voltage feature vector corresponding to each sliding window. S104. Dynamically identify each cell in the electric vehicle battery pack according to the preset algorithm parameters, density clustering algorithm and normalized two-dimensional voltage feature vector to obtain the abnormal cluster identification result. S105. Perform corresponding abnormal cell fault warning operations based on the abnormal cluster identification results according to the preset multi-level risk thresholds.

[0033] In some embodiments, the method can use sensors to collect data to obtain the voltage curves (i.e., the original voltage sequence) of each cell in the electric vehicle battery pack under driving and charging conditions. The sensor collection frequency is 0.1 Hz, meaning data is collected once every 10 seconds.

[0034] In some embodiments, the method employs a density-based clustering (DBSCAN) algorithm to dynamically identify fault segments, with the input being a normalized feature matrix of a single window. The neighborhood radius ε and the minimum number of samples are set in advance. The parameters are as follows, and the algorithm flow is as follows: (1) Randomly select an unvisited battery cell , i represents the i-th cell (at the beginning of the algorithm, none of the cells have been visited); (2) Inspection The ε-neighborhood, if If it is a core point, it is marked as a core point and a new cluster is created; otherwise, it is temporarily marked as a noise point. (3) To Density expansion: If If the core point is used, then recursively... All reachable points within the ε-neighborhood are added to the current cluster. During this process, newly discovered core points continue to expand, and this step is repeated. (4) Repeat the above steps until all points have been visited. Points that are still marked as noise points are considered outliers.

[0035] The algorithm output includes several clusters of density-connected points and outliers not included in any cluster. This is achieved by adjusting ε and... This allows for the differentiation between normal and faulty battery cells.

[0036] In some embodiments, the method sets the frequency at which cells are marked as anomalous clusters every k sliding windows. Set three risk levels (t1, t2, t3) for low, medium, and high risk, and optimize the threshold settings based on historical data.

[0037] In some embodiments, for exceeding a threshold The high-frequency abnormal cell starts a continuous monitoring and thermal runaway early warning process to reduce the impact of occasional interference.

[0038] In some embodiments, using multiple features for fault diagnosis and dynamically adjusting density-based clustering parameters according to the real-time operating conditions of the vehicle can improve accuracy by 30% compared to methods using a single feature and fixed clustering parameters.

[0039] In these embodiments, the method can detect battery pack faults in electric vehicles online in real time, accurately locate faulty cells, and achieve multi-level risk warnings, thereby improving the accuracy and timeliness of fault diagnosis and helping to avoid safety accidents such as thermal runaway.

[0040] In some embodiments, the original voltage sequence is preprocessed to obtain a preprocessed voltage sequence, including: The original voltage sequence is reordered in descending order of timestamps to obtain the first processed sequence; The first processing sequence is processed for missing data and outliers to obtain the second processing sequence. The second processed sequence is logarithmically transformed to obtain the preprocessed voltage sequence.

[0041] In some embodiments, due to issues such as sensor acquisition and data uploading, there may be timestamp sorting errors in the original voltage sequence, requiring the voltage sequence to be reordered in descending order of timestamps.

[0042] In some embodiments, the voltage sequence may also contain issues such as missing (null) or outlier (0 or 255) cell data. For missing or outlier cell voltage data, the voltage value of a cell that does not have missing data at the same time is used instead.

[0043] In some embodiments, for missing or outlier values ​​in all cell voltage data, a logarithmic transformation is performed using normal data from before and after the missing data to reduce data skewness. The logarithmic transformation formula is as follows: ; Where V is the original voltage sequence and V' is the voltage sequence after logarithmic transformation.

[0044] For example, Figure 2 A schematic diagram of density-based voltage fault clustering results is shown. Based on the results illustrated in the diagram, it can be seen that these embodiments can optimize the voltage data processing, reducing the false alarm rate to 5%–10% compared to previous implementations (where similar online detection methods had a false alarm rate of 30%–40%). Figure 2 Monomer 8, monomer 18, and monomer 81 correspond to false alarm points.

[0045] In these embodiments, the method can effectively handle timestamp sorting errors, missing data, and outliers in the original voltage sequence, thereby obtaining a high-quality preprocessed voltage sequence and providing a reliable data foundation for subsequent fault diagnosis.

[0046] In some embodiments, based on a preset sliding window and a preprocessed voltage sequence, a normalized two-dimensional voltage feature vector corresponding to each sliding window is calculated, including: The preprocessed voltage sequence is divided by a preset sliding window to obtain multiple voltage subsequences; Calculate the voltage consistency feature and modified Shannon entropy feature corresponding to each sliding window based on multiple voltage subsequences; Based on voltage consistency characteristics and modified Shannon entropy characteristics, a two-dimensional feature matrix corresponding to each sliding window is constructed. The two-dimensional feature matrix is ​​normalized to obtain the two-dimensional normalized feature matrix corresponding to each sliding window.

[0047] In some embodiments, the method can utilize voltage data collected by sensors, extract voltage difference features and corrected Shannon entropy features, and combine them with DBSCAN clustering analysis to achieve accurate identification of faulty battery cells.

[0048] In some embodiments, the voltage consistency feature quantifies the abnormal deviation trend of an individual cell by calculating the dynamic offset of the voltage of each cell within a window from the cluster mean.

[0049] In some embodiments, Shannon entropy is widely used in thermodynamics, information science, and other fields to determine the degree of disorder in a system, and in the field of battery fault diagnosis, it can be used to determine the degree of fluctuation in voltage sequences. In this method, calculating the modified Shannon entropy as a feature can reduce the requirements for online computation.

[0050] For example, the feature extraction method includes: First, a sliding window is used to divide the charging data segments; Next, the mean voltage difference is calculated for each cell voltage sequence, the Shannon entropy is corrected, and the mean of the mean voltage difference within the window is extracted as the voltage consistency feature. Finally, the fused feature of the voltage consistency feature after maximum-minimum normalization and the modified Shannon entropy feature is defined as a two-dimensional feature vector.

[0051] It is evident that this feature extraction method takes into account the voltage deviation and voltage fluctuation differences of each cell when a fault occurs, including at least voltage consistency features and modified Shannon entropy features.

[0052] In some embodiments, the method divides the voltage sequence by a sliding window (window size N), calculates two types of core features, and constructs a two-dimensional feature vector.

[0053] In some embodiments, the average deviation of the voltage of each cell within the window from the average value of all cells is used to characterize the voltage deviation state of a single cell, resulting in a voltage consistency feature matrix: ; ; In the formula, i is the cell number, T is the sliding window time length, n is the total number of cells, and V i，t Let t be the voltage value of cell i at time t.

[0054] In some embodiments, modified Shannon entropy can effectively identify voltage fluctuation anomalies by quantifying the randomness differences in voltage distribution.

[0055] In some embodiments, the method discretizes the voltage matrix for any window. Where k is the window sequence length and n is the number of battery cells, calculate the maximum value x of the matrix. max With minimum value x min Now, define the matrix: ; ; in, , l For custom hyperparameters.

[0056] At this point, calculate the frequency matrix: ; ; Then, the modified Shannon entropy feature matrix within the window is calculated based on the frequency matrix. ,in, ; n is the number of battery cells.

[0057] Among them, Shannon entropy is corrected for random anomalies in voltage distribution during battery fault diagnosis.

[0058] Finally, construct the two-dimensional feature matrix: .

[0059] In some embodiments, this method employs min-max normalization, which maps features of different dimensions to a unified interval ([0, 1]), thereby addressing the feature weight imbalance caused by differences in numerical ranges. It also enhances data comparability, ensures the model treats all features fairly, and facilitates multi-feature fusion. The specific normalization operation is as follows: ; in, for or , This is the normalized one-dimensional feature vector.

[0060] It can be seen that the features within the sliding window Performing the above operations on the row vectors yields a two-dimensional normalized feature matrix. .

[0061] In these embodiments, the method can effectively extract key features from the preprocessed voltage sequence and provide representative feature data for subsequent density clustering analysis and fault diagnosis by constructing and normalizing a two-dimensional feature matrix.

[0062] In some embodiments, based on preset multi-level risk thresholds, corresponding abnormal cell fault warning operations are performed on the abnormal cluster identification results, including: Based on the abnormal cluster identification results, the frequency of each battery cell being marked as an abnormal cluster is statistically analyzed to obtain the abnormal marking frequency corresponding to each battery cell. The abnormal cell identification result is determined based on the frequency of abnormal markers and preset multi-level risk thresholds; the abnormal cell identification result includes low-risk abnormal cells, medium-risk abnormal cells, and high-risk abnormal cells; the preset multi-level risk thresholds include low-risk threshold, medium-risk threshold, and high-risk threshold. The output includes abnormal cell identification results and abnormal cell prompt information.

[0063] In these embodiments, the method enables multi-level early warning and refined management of battery pack failures.

[0064] In some embodiments, the method further includes: Continuous monitoring of high-risk abnormal battery cells yields monitoring results; When a high-risk abnormal battery cell is detected to have thermal runaway based on monitoring results, a thermal runaway alarm message is output.

[0065] In these embodiments, the method can monitor high-risk abnormal cells in real time and continuously, and output alarm information in a timely manner when serious faults such as thermal runaway are detected, thereby effectively preventing battery pack safety accidents and ensuring the safe operation of electric vehicles.

[0066] In some embodiments, the preset algorithm parameters include a preset neighborhood radius and a preset minimum number of samples; The preset algorithm parameters are obtained by dynamically adjusting the original algorithm parameters based on historical vehicle data, battery type, vehicle driving conditions, and vehicle working environment. The original algorithm parameters include the pre-calibrated original neighborhood radius and the pre-calibrated original minimum number of samples.

[0067] In some embodiments, due to ε and The parameters need to be preset. Therefore, after combining historical data research, this method can dynamically adjust ε and φ according to battery type and vehicle driving conditions, such as charging, discharging, high temperature, low temperature, and other environmental factors. Parameters are adjusted to achieve better fault diagnosis results.

[0068] In these embodiments, the method can dynamically adjust the original algorithm parameters to optimize the preset algorithm parameters, thereby improving the accuracy of density clustering algorithm in battery pack fault diagnosis.

[0069] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in this application will be clearly and completely described below. In some embodiments, such as Figure 3 As shown, the battery pack fault diagnosis method includes: S201. Obtain the voltage sequence of each cell in the battery pack from the sensor; S202, Data Preprocessing; S203. Divide the voltage sequence into several sliding windows and calculate the normalized two-dimensional voltage feature vector in each window. S204. The obtained normalized two-dimensional voltage feature vector is input based on a density clustering algorithm to distinguish between normal clusters and abnormal clusters; S205. Based on the frequency of each cell being recorded as an abnormal cluster within a certain window, a multi-level alarm mechanism is activated to obtain the final fault diagnosis result.

[0070] In some embodiments, the application scenarios of this method can include: online fault monitoring of electric vehicle battery packs, fault cell location, thermal runaway early warning, and time-series data anomaly detection.

[0071] In these embodiments, the method takes into account the difficulty in identifying different minor faults, and incorporates logarithmic transformation to amplify the characteristic differences between faulty cells and normal cells, thereby reducing data skewness.

[0072] In these embodiments, the method can apply features with two different dimensions to the fault diagnosis algorithm, which improves the accuracy of the algorithm and reduces the false alarm rate compared to using a single feature.

[0073] In these embodiments, the method can dynamically adjust the density-based clustering parameters in conjunction with the real-time operating conditions of the vehicle, thereby improving the robustness of the algorithm.

[0074] In these embodiments, the method is designed to ignore occasional fault alarms in order to avoid false alarms.

[0075] Figure 4 A schematic diagram of a battery pack fault diagnosis device is shown. It should be understood that this device is related to... Figure 1 The method executed in the middle corresponds to the steps involved in the aforementioned method. The specific functions and effects of the device can be found in the description above. To avoid repetition, detailed descriptions are omitted here.

[0076] The battery pack fault diagnosis device includes: The acquisition unit 310 is used to acquire the original voltage sequence of each cell in the electric vehicle battery pack under driving and charging conditions; The preprocessing unit 320 is used to preprocess the original voltage sequence to obtain a preprocessed voltage sequence; The feature calculation unit 330 is used to calculate the normalized two-dimensional voltage feature vector corresponding to each sliding window based on the preset sliding window and the preprocessed voltage sequence. Anomaly identification unit 340 is used to dynamically identify each cell of the electric vehicle battery pack according to preset algorithm parameters, density clustering algorithm and normalized two-dimensional voltage feature vector, and obtain anomaly cluster identification results. The early warning unit 350 is used to perform corresponding abnormal cell fault early warning operations based on the abnormal cluster identification results according to preset multi-level risk thresholds.

[0077] In some embodiments, the preprocessing unit 320 includes: Sorting subunit 321 is used to reorder the original voltage sequence in descending order of timestamps to obtain the first processed sequence; Processing subunit 322 is used to process the first processing sequence for missing data and outliers to obtain the second processing sequence; Transformation subunit 323 is used to perform a logarithmic transformation on the second processing sequence to obtain a preprocessed voltage sequence.

[0078] In some embodiments, the feature calculation unit 330 includes: Sub-unit 331 is used to divide the preprocessed voltage sequence through a preset sliding window to obtain multiple voltage sub-sequences; Computational subunit 332 is used to calculate the voltage consistency feature and modified Shannon entropy feature corresponding to each sliding window based on multiple voltage subsequences; Subunit 333 is constructed to build a two-dimensional feature matrix corresponding to each sliding window based on voltage consistency characteristics and modified Shannon entropy characteristics; Normalization subunit 334 is used to normalize the two-dimensional feature matrix to obtain the two-dimensional normalized feature matrix corresponding to each sliding window.

[0079] In some embodiments, the warning unit 350 includes: The statistical subunit 351 is used to statistically analyze the frequency of each battery cell being marked as an abnormal cluster based on the abnormal cluster identification result, and obtain the abnormal marking frequency corresponding to each battery cell. The determination subunit 352 is used to determine the abnormal cell identification result based on the abnormal marking frequency and the preset multi-level risk threshold; wherein, the abnormal cell identification result includes low-risk abnormal cells, medium-risk abnormal cells and high-risk abnormal cells; wherein, the preset multi-level risk threshold includes low-risk threshold, medium-risk threshold and high-risk threshold. Output subunit 353 is used to output abnormal cell prompt information, including the abnormal cell identification result.

[0080] In some embodiments, the battery pack fault diagnosis device further includes: The monitoring unit 360 is used to continuously monitor high-risk abnormal cells and obtain monitoring results; The early warning unit 350 is used to output thermal runaway alarm information when a high-risk abnormal battery cell is detected to have thermal runaway based on monitoring results.

[0081] In some embodiments, the preset algorithm parameters include a preset neighborhood radius and a preset minimum number of samples; The preset algorithm parameters are obtained by dynamically adjusting the original algorithm parameters based on historical vehicle data, battery type, vehicle driving conditions, and vehicle working environment. The original algorithm parameters include the pre-calibrated original neighborhood radius and the pre-calibrated original minimum number of samples.

[0082] like Figure 5 As shown, this application provides an electronic device 400, which includes a processor 401 and a memory 402. The processor 401 and the memory 402 are interconnected and communicate with each other through a communication bus 403 and / or other forms of connection mechanism (not shown). The memory 402 stores a computer program that can be executed by the processor 401. When the computing device is running, the processor 401 executes the computer program to perform the method in any of the aforementioned optional implementations.

[0083] This application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the method in any of the aforementioned optional implementations.

[0084] The computer-readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0085] This application provides a computer program product, which includes a computer program that, when run by a processor, executes the method in any of the aforementioned optional implementations.

[0086] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and not to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. These modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and they should all be covered within the scope of the claims and specification of this application. In particular, as long as there is no conflict, the various technical features mentioned in the embodiments can be combined in any way. This application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for diagnosing battery pack faults, characterized in that, include: Obtain the original voltage sequence of each cell in the electric vehicle battery pack under driving and charging conditions; The original voltage sequence is preprocessed to obtain a preprocessed voltage sequence; Based on the preset sliding window and the preprocessed voltage sequence, calculate the normalized two-dimensional voltage feature vector corresponding to each sliding window; Based on preset algorithm parameters, density clustering algorithm and the normalized two-dimensional voltage feature vector, the cells of the electric vehicle battery pack are dynamically identified to obtain abnormal cluster identification results. Based on the preset multi-level risk thresholds, the abnormal cluster identification results are used to perform corresponding abnormal cell fault early warning operations. The step of calculating the normalized two-dimensional voltage feature vector corresponding to each sliding window based on the preset sliding window and the preprocessed voltage sequence includes: The preprocessed voltage sequence is divided by a preset sliding window to obtain multiple voltage sub-sequences; Calculate the voltage consistency feature and modified Shannon entropy feature corresponding to each sliding window based on the multiple voltage subsequences; Based on the voltage consistency feature and the modified Shannon entropy feature, construct a two-dimensional feature matrix corresponding to each sliding window; The two-dimensional feature matrix is ​​normalized to obtain the two-dimensional normalized feature matrix corresponding to each sliding window; The preset algorithm parameters include a preset neighborhood radius and a preset minimum number of samples; The preset algorithm parameters are obtained by dynamically adjusting the original algorithm parameters based on historical vehicle data, battery type, vehicle driving conditions, and vehicle working environment. The original algorithm parameters include the pre-calibrated original neighborhood radius and the pre-calibrated original minimum number of samples.

2. The battery pack fault diagnosis method according to claim 1, characterized in that, The preprocessing of the original voltage sequence to obtain a preprocessed voltage sequence includes: The original voltage sequence is reordered in descending order of timestamps to obtain the first processed sequence; The first processing sequence is processed for missing data and outliers to obtain the second processing sequence; The second processed sequence is logarithmically transformed to obtain the preprocessed voltage sequence.

3. The battery pack fault diagnosis method according to claim 1, characterized in that, The step of performing corresponding abnormal cell fault early warning operations based on the abnormal cluster identification results according to preset multi-level risk thresholds includes: Based on the abnormal cluster identification results, the frequency of each battery cell being marked as an abnormal cluster is statistically analyzed to obtain the abnormal marking frequency corresponding to each battery cell. The abnormal cell identification result is determined based on the abnormal marking frequency and the preset multi-level risk threshold; wherein, the abnormal cell identification result includes low-risk abnormal cells, medium-risk abnormal cells, and high-risk abnormal cells; wherein, the preset multi-level risk threshold includes a low-risk threshold, a medium-risk threshold, and a high-risk threshold; The output includes abnormal cell alert information that includes the abnormal cell identification results.

4. The battery pack fault diagnosis method according to claim 3, characterized in that, The method further includes: The high-risk abnormal battery cells were continuously monitored, and the monitoring results were obtained. When thermal runaway is detected in the high-risk abnormal battery cell based on the monitoring results, a thermal runaway alarm message is output.

5. A battery pack fault diagnosis device, characterized in that, The battery pack fault diagnosis device includes: The acquisition unit is used to acquire the original voltage sequence of each cell in the electric vehicle battery pack under driving and charging conditions; A preprocessing unit is used to preprocess the original voltage sequence to obtain a preprocessed voltage sequence; The feature calculation unit is used to calculate the normalized two-dimensional voltage feature vector corresponding to each sliding window based on the preset sliding window and the preprocessed voltage sequence. An anomaly identification unit is used to dynamically identify each cell of the electric vehicle battery pack according to preset algorithm parameters, density clustering algorithm and the normalized two-dimensional voltage feature vector, and obtain anomaly cluster identification results. The early warning unit is used to perform corresponding abnormal cell fault early warning operations based on the abnormal cluster identification results according to preset multi-level risk thresholds. The feature calculation unit includes: The sub-unit is used to divide the preprocessed voltage sequence through a preset sliding window to obtain multiple voltage sub-sequences; The computational subunit is used to calculate the voltage consistency feature and the modified Shannon entropy feature corresponding to each sliding window based on the multiple voltage subsequences; Construct sub-units to build a two-dimensional feature matrix corresponding to each sliding window based on the voltage consistency feature and the modified Shannon entropy feature; The normalization subunit is used to normalize the two-dimensional feature matrix to obtain the two-dimensional normalized feature matrix corresponding to each sliding window. The preset algorithm parameters include a preset neighborhood radius and a preset minimum number of samples; The preset algorithm parameters are obtained by dynamically adjusting the original algorithm parameters based on historical vehicle data, battery type, vehicle driving conditions, and vehicle working environment. The original algorithm parameters include the pre-calibrated original neighborhood radius and the pre-calibrated original minimum number of samples.

6. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory being used to store a computer program, and the processor running the computer program to cause the electronic device to perform the battery pack fault diagnosis method according to any one of claims 1 to 4.

7. A readable storage medium, characterized in that, The readable storage medium stores a computer program, which, when executed by a processor, performs the battery pack fault diagnosis method according to any one of claims 1 to 4.

8. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, performs the battery pack fault diagnosis method according to any one of claims 1 to 4.

Images