Multi-fault Diagnosis Method for Power Batteries of New Energy Vehicles

Abstract

The present invention relates to the technical field of lithium-ion battery fault diagnosis, and discloses a multi-fault diagnosis method for power batteries of new energy vehicles. The method adopts a two-layer diagnosis strategy. The first layer performs multi-level feature extraction on battery voltage data, and uses the ALOF algorithm to accurately identify various abnormal conditions of abnormal cells in the battery pack. The second layer uses the Mahalanobis distance to further perform internal short circuit fault diagnosis on abnormal single cells. In order to extract fault features in a finer granularity, the kurtosis and change rate are extracted as indicators for different time windows of the voltage data after eliminating the influence of the environment and sensor errors. Then, the ALOF algorithm is used to perform preliminary abnormal screening on the voltage changes of each single cell at different time scales. The average voltage data of normal batteries screened out after the first layer diagnosis that can reflect the current battery working condition and aging condition is used as a comparison object. The Mahalanobis distance is used to further diagnose internal short circuit faults for the single cells diagnosed as abnormal in the first layer.

Description

Technical Field

[0001] The present invention relates to the technical field of lithium-ion battery fault diagnosis, and in particular to a multi-fault diagnosis method for power batteries of new energy vehicles. Background Art

[0002] The increase in the number of new energy vehicles has led to serious power battery safety issues. The safety of power batteries is related to the safety of people's lives and property. Different from the spontaneous combustion of traditional fuel vehicles, the active lithium element in power batteries means that once thermal runaway occurs, the safe escape window is only a few minutes. There is currently no effective way to solve this problem. Therefore, it is necessary to study how to locate and diagnose potential faults in the early stages of a fault.

[0003] There are three main types of fault diagnosis methods in existing research: knowledge-based methods, model-based methods, and data-driven methods. Knowledge-based methods rely heavily on expert knowledge and have poor applicability. Model-based methods use experimental data for modeling, so it is difficult to achieve accurate diagnosis under real working conditions. Data-driven methods mainly use real vehicle data, so they have more practical application value. In recent years, with the rise of big data platforms, methods based on the combination of platform vehicle operation data and data-driven methods have good real-time performance and strong historical data utilization capabilities. However, such methods still have the following problems: (1) Current diagnostic methods are difficult to diagnose multiple faults and anomalies efficiently. (2) Commonly used algorithms are difficult to accurately judge multiple anomalies and multiple adjacent anomalies, making it easy for the algorithms to miss diagnoses; (3) Existing algorithms are basically based on fixed data or thresholds for comparative diagnosis, and do not consider the serious impact of battery operating conditions and aging, making it easy to miss diagnoses. Summary of the Invention

[0004] (1) Technical problems solved

[0005] In response to the shortcomings of the existing technology, the present invention provides a multi-fault diagnosis method for new energy vehicle power batteries, which has the advantages of identifying and locating multiple abnormalities and multiple adjacent abnormal batteries, and solves the above technical problems.

[0006] (2) Technical solution

[0007] To achieve the above objectives, the present invention provides the following technical solution: a multi-fault diagnosis method for a power battery of a new energy vehicle, comprising the following steps:

[0008] S1: Obtain the voltage data of all cells in the real vehicle battery pack during operation before the sampling time, and construct the voltage matrix of the battery pack , Represents the battery pack Single battery, represent Subsampling, As the voltage sequence at the detection time, the voltage data at the detection time is preliminarily detected for abnormalities according to the nominal upper and lower voltage limits of the battery when it leaves the factory;

[0009] S2: Using voltage matrix The voltage data of the 60 moments before the current moment are included in the calculation of the kurtosis of the voltage data of each single cell in the time window, and then the kurtosis values of all single cells are combined to generate the kurtosis feature vector ;

[0010] S3: Use the voltage data of the six moments before the current moment in the voltage matrix to calculate the voltage change rate of each single cell in the time window, and then combine the voltage change rates of all single cells to generate the voltage change rate feature vector ;

[0011] S4: The kurtosis eigenvector , voltage change rate characteristic vector and the voltage sequence at the detection moment Bring it into the ALOF algorithm for calculation and get the corresponding ALOF values, namely 、 、 ,judge 、 、 Whether any of the items exceeds the corresponding threshold 、 、 , if it exceeds the corresponding threshold, execute step S5;

[0012] S5: Calculate the Mahalanobis distance M between the historical voltage sequence of the abnormal single cell and the mean voltage of the normal cells in the same period. Greater than threshold When , it is determined that this single battery has an internal short circuit fault. The specific expression is as follows:

[0013]

[0014] in, The Mahalanobis distance is calculated between the historical voltage sequence of abnormal single cells and the mean voltage of normal cells in the same period. Indicates the historical voltage data of abnormal single cells, with superscript represents transpose, represents the historical voltage mean vector of a normal battery, represents the historical voltage covariance matrix of a normal battery, and its expression is as follows:

[0015]

[0016]

[0017] in, Indicates the voltage value of the b-th sampling point, Indicates the total number of voltage data.

[0018] As a preferred technical solution of the present invention, in step S1, the acquired data is preprocessed before constructing the voltage matrix. The preprocessing algorithm adopts a wavelet threshold noise reduction algorithm to perform wavelet decomposition on the original voltage data to obtain the wavelet coefficients of each layer, wherein the wavelet coefficient formula is:

[0019]

[0020] Where z represents the number of sampling points, Indicates taking the logarithm of the number of sampling points;

[0021] Then, the abnormal data points are judged according to the threshold and the abnormal data is removed. The soft threshold function is selected As a threshold function, its expression is:

[0022]

[0023] in, is the wavelet coefficient, is a symbolic function, is the critical threshold.

[0024] As a preferred technical solution of the present invention, the kurtosis eigenvector Middle The kurtosis of a single cell The specific expression is as follows:

[0025]

[0026]

[0027]

[0028] in, Indicates the The kurtosis of a single cell, Indicates the window size selected for the long time scale, Indicates the time window starting from the current moment The voltage value of the subsample, represents the mean voltage in the selected window, Indicates the The average value of the voltage data of each single cell in the time window, Indicates the The standard deviation of the voltage data of each single cell in the time window.

[0029] As a preferred technical solution of the present invention, the voltage change rate characteristic vector Middle Voltage change rate of a single cell The calculation formula is:

[0030]

[0031] in, represents the window size selected for the short time scale, Indicates the voltage value at the leftmost end of the time window, Indicates the voltage value at the current moment.

[0032] As a preferred technical solution of the present invention, the specific steps of the ALOF algorithm in step S4 are as follows:

[0033] S4.1: Calculate the reachable distance between objects y and x :

[0034] in, and Objects and No. eigenvalues, is the total number of eigenvalues, Express Sum the data;

[0035] S4.2: Set an initial domain size ,object Within the area size The reachable distance is:

[0036]

[0037] At this time, the object The initial neighborhood of Expressed as:

[0038]

[0039] in, represents the entire dataset, Representation object Within the area size The reachable distance, represents the distance between any object in domain k and object x;

[0040] S4.3: Computational Objects The local density is:

[0041]

[0042] in, Represents the average distance between object x and the objects in its initial neighborhood, which is used to reflect the local density. represents the number of objects contained in the initial neighborhood of object 𝑥, represents the sum of the distances between object 𝑥 and all objects y in its initial neighborhood.

[0043] S4.4: Compute local distances for all objects in the entire dataset Average value :

[0044]

[0045] Refers to the total number of objects in the dataset.

[0046] Determine the dynamically changing neighborhood size corresponding to each object based on the local density and average local density of the data for:

[0047]

[0048] in, is the proportional coefficient, which is used to control the change of the neighborhood size and is usually set to 3; the symbol Indicates rounding up to ensure is an integer; reflects the ratio of the local density of object 𝑥 to the global average density, if is smaller (i.e., the neighborhood density of 𝑥 is higher), then will be larger, and vice versa.

[0049] S4.5: For new field values , object x within the domain size The reachable distance is:

[0050]

[0051] New neighborhood value The neighborhood including the object x itself is represented as:

[0052]

[0053] New local density for:

[0054]

[0055] in represents the number of objects contained in the corresponding dynamic neighborhood of object 𝑥, Represents the reachable distance in a dynamically adjusted neighborhood:

[0056]

[0057] For objects , repetition and object The same process can be obtained ;

[0058] S4.6 The revised ALOF value is updated as follows:

[0059]

[0060] in, express ALOF value;

[0061] The threshold 、 、 The specific expression is as follows

[0062]

[0063] in, Representative eigenvalue 、 、 One of them, for The corresponding ALOF value, is the eigenvalue All single cells The mean of the values, For all single cells under this characteristic value The standard deviation of the values.

[0064] Compared with the prior art, the present invention provides a multi-fault diagnosis method for power batteries of new energy vehicles, which has the following beneficial effects:

[0065] 1. By adopting voltage characteristics at different time scales, the present invention can capture battery abnormalities in a more comprehensive and fine-grained manner, and accurately diagnose multiple faults by combining a double-layer diagnosis strategy with the ALOF algorithm and Mahalanobis distance.

[0066] 2. The present invention uses an improved LOF algorithm ALOF in the method of locating abnormal single cells. The algorithm can better identify and distinguish multiple adjacent abnormal objects by dynamically adjusting the field size and k value, and can efficiently and accurately identify and locate abnormal single cells in the battery pack.

[0067] 3. The present invention uses the fault diagnosis method to use the normal battery voltage mean data diagnosed in the first layer of the diagnostic strategy to represent the general voltage state of the normal battery under the working condition, that is, the aging degree. The Mahalanobis distance between it and the abnormal battery historical voltage data is calculated, and then the threshold is used to judge the abnormal single cell fault condition. The algorithm takes into account the impact of battery aging and working conditions on performance, which helps to reduce missed diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 It is a schematic diagram of the process of the present invention;

[0069] Figure 2 This is a schematic diagram of voltage-time data of a normal vehicle of the present invention;

[0070] Figure 3 This is a schematic diagram of voltage-time data of a faulty vehicle according to the present invention;

[0071] Figure 4 This is a schematic diagram of voltage-time data of a normal vehicle after noise reduction according to the present invention;

[0072] Figure 5 A schematic diagram of voltage-time data of a faulty vehicle after noise reduction according to the present invention;

[0073] Figure 6 This is a schematic diagram of the ALOF result of a faulty vehicle according to the present invention. DETAILED DESCRIPTION

[0074] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0075] See also Figure 1-6 , a multi-fault diagnosis method for a new energy vehicle power battery, comprising the following steps:

[0076] S1: Obtain the voltage data of all cells in the real vehicle battery pack during operation before the sampling time, and construct the voltage matrix of the battery pack , Represents the battery pack Battery cells, represent Subsampling, As the voltage sequence at the time of detection, the voltage data at the time of detection is subjected to preliminary abnormality detection according to the nominal upper and lower limit voltage values of the battery when it leaves the factory. The nominal upper limit voltage of the single battery used in the battery pack of this vehicle is 4.15V, and the nominal lower limit voltage is 2.75V;

[0077] In step S1, before constructing the voltage matrix, the acquired data is preprocessed. The preprocessing algorithm uses a wavelet threshold noise reduction algorithm to perform wavelet decomposition on the original voltage data to obtain the wavelet coefficients of each layer, where the wavelet coefficient formula is:

[0078]

[0079] Where z represents the number of sampling points, Indicates taking the logarithm of the number of sampling points;

[0080] Then, the abnormal data points are judged according to the threshold and the abnormal data is removed. The soft threshold function is selected As a threshold function, its expression is:

[0081]

[0082] in, is the wavelet coefficient, is a symbolic function, is the critical threshold.

[0083] S2: Using voltage matrix The voltage data of the 60 moments before the current moment are included in the calculation of the kurtosis of the voltage data of each single cell in the time window, and then the kurtosis values of all single cells are combined to generate the kurtosis feature vector , the kurtosis eigenvector Middle The kurtosis of a single cell The specific expression is as follows:

[0084]

[0085]

[0086]

[0087] in, Indicates the The kurtosis of a single cell, Indicates the window size selected for the long time scale, Indicates the time window starting from the current moment The voltage value of the subsample, represents the mean voltage in the selected window, Indicates the The average value of the voltage data of each single cell in the time window, Indicates the The standard deviation of the voltage data of each single cell in the time window;

[0088] S3: Use the voltage data of the six moments before the current moment in the voltage matrix to calculate the voltage change rate of each single cell in the time window, and then combine the voltage change rates of all single cells to generate the voltage change rate feature vector , the voltage change rate characteristic vector Middle Voltage change rate of a single cell The calculation formula is:

[0089]

[0090] in, represents the window size selected for the short time scale, Indicates the voltage value at the leftmost end of the time window, Indicates the voltage value at the current moment;

[0091] S4: The kurtosis eigenvector , voltage change rate characteristic vector and the voltage sequence at the detection moment Bring it into the ALOF algorithm for calculation and get the corresponding ALOF values, namely 、 、 ,judge 、 、 Whether any of the items exceeds the corresponding threshold 、 、 If the value exceeds the corresponding threshold, step S5 is executed. The specific steps of the ALOF algorithm are as follows:

[0092] S4.1: Calculate the reachable distance between objects y and x :

[0093] in, and Objects and No. eigenvalues, is the total number of eigenvalues, Express Sum the data;

[0094] S4.2: Set an initial domain size ,object Within the area size The reachable distance is:

[0095]

[0096] At this time, the object The initial neighborhood of Expressed as:

[0097]

[0098] in, represents the entire dataset, Representation object Within the area size The reachable distance, represents the distance between any object in domain k and object x;

[0099] S4.3: Computational Objects The local density is:

[0100]

[0101] in, Represents the average distance between object x and the objects in its initial neighborhood, which is used to reflect the local density. represents the number of objects contained in the initial neighborhood of object 𝑥, represents the sum of the distances between object 𝑥 and all objects y in its initial neighborhood.

[0102] S4.4: Compute local distances for all objects in the entire dataset Average value :

[0103]

[0104] Refers to the total number of objects in the dataset.

[0105] Determine the dynamically changing neighborhood size corresponding to each object based on the local density and average local density of the data for:

[0106]

[0107] in, is the proportional coefficient, which is used to control the change of the neighborhood size and is usually set to 3; the symbol Indicates rounding up to ensure is an integer; reflects the ratio of the local density of object 𝑥 to the global average density, if is smaller (i.e., the neighborhood density of 𝑥 is higher), then will be larger, and vice versa.

[0108] S4.5: For new field values , object x within the domain size The reachable distance is:

[0109]

[0110] New neighborhood value The neighborhood including the object x itself is represented as:

[0111]

[0112] New local density for:

[0113]

[0114] in represents the number of objects contained in the corresponding dynamic neighborhood of object 𝑥, Represents the reachable distance in a dynamically adjusted neighborhood:

[0115]

[0116] For objects , repetition and object The same process can be obtained ;

[0117] S4.6 The revised ALOF value is updated as follows:

[0118]

[0119] in, express ALOF value;

[0120] The threshold 、 、 The specific expression is as follows:

[0121]

[0122] in, Representative eigenvalue 、 、 One of them, for The corresponding ALOF value, is the eigenvalue All single cells The mean of the values, For all single cells under this characteristic value Standard deviation of the values;

[0123] S5: Calculate the Mahalanobis distance M between the historical voltage sequence of the abnormal single cell and the voltage mean of the normal cells in the same period. Greater than threshold When the battery is detected as short-circuit, it is determined that the battery has an internal short-circuit fault. The value is 1.12, and the specific expression is as follows:

[0124]

[0125] in, Indicates the The Mahalanobis distance is calculated by comparing the historical voltage sequence of each abnormal single cell with the mean voltage of normal cells in the same period. Indicates the historical voltage data of abnormal single cells, with superscript represents transpose, represents the historical voltage mean vector of a normal battery, represents the historical voltage covariance matrix of a normal battery, and its expression is as follows:

[0126]

[0127]

[0128] in, Indicates the voltage value of the b-th sampling point, Indicates the total number of voltage data;

[0129] Kurtosis is used in statistics to describe the shape of data distributions and can reflect the steepness and tail thickness of a data distribution. Kurtosis is defined as the fourth-order centroid moment of a random variable divided by the fourth power of the standard deviation and is a dimensionless parameter. In battery fault diagnosis, the kurtosis value of the voltage data for each single cell 10 minutes before the diagnosis is calculated to characterize the comprehensive degree of battery voltage change over a longer time window.

[0130] The ALOF algorithm used in S4 is an algorithm that is optimized based on the LOF algorithm for detecting abnormalities in multiple single cells at the same time and in multiple adjacent single cells at the same time. Its basic algorithm, LOF, is an unsupervised learning algorithm commonly used for anomaly detection. This algorithm identifies outliers by comparing the local density difference between the target data point and its adjacent data points. This algorithm is widely used in battery fault diagnosis, but it is greatly affected by the choice of neighborhood size k, and the diagnostic results are easily interfered with by data noise. Most importantly, when multiple anomalies occur simultaneously or multiple anomalies occur adjacent to each other, their neighborhoods are highly overlapping, making them easily misjudged as normal. Therefore, this solution is modified based on the original algorithm to dynamically adjust the neighborhood size k of each object based on the local density of the area where the object is located and include the object x itself in its neighborhood;

[0131] Using the ALOF algorithm to judge abnormalities of different voltage characteristic values can not only detect potential abnormalities more accurately and comprehensively, but also diagnose some potential faults at the same time; When the value exceeds the corresponding threshold, it indicates that the voltage of the single battery has abnormal fluctuations; when When the value exceeds the corresponding threshold, it indicates that the voltage of the single cell has experienced abnormal fluctuations in the recent period; When the value exceeds the corresponding threshold, it indicates that the single battery has an inconsistency risk;

[0132] The Mahalanobis distance is used to further diagnose the abnormal single battery located, and the S4 、 、 All cells that do not exceed the corresponding threshold are considered normal cells for that period. Their historical voltage data are averaged at each sampling moment to obtain voltage sequence data that represents the general state of normal cells for that period. This is used as a comparison target for abnormal cells to more accurately diagnose faults. The abnormal cell has already been located in step S5. This step further diagnoses the fault of the abnormal cell. The Mahalanobis distance can accurately and efficiently calculate the difference between two sequences. Calculating the Mahalanobis distance between the historical voltage data of the abnormal cell and the average voltage of the normal cells can effectively diagnose the actual fault condition of the abnormal cell.

[0133] Specifically, in this embodiment, the voltage-time data of normal vehicles and faulty vehicles during the charging process are obtained from the new energy vehicle big data platform. The voltage-time data of normal vehicles is as follows: Figure 2 As shown, the voltage-time data of the faulty vehicle is as follows Figure 3 As shown; Figure 4 shows the voltage-time data of a normal vehicle after noise reduction, Figure 5 The voltage-time data of the faulty vehicle after noise reduction is shown.

[0134] For the faulty vehicle, the time corresponding to the 3500th data point is selected as the diagnosis moment, and then the voltage matrix is constructed using historical data First, the voltage data at the time of diagnosis is used to make a preliminary judgment based on the upper and lower limits of the nominal voltage to determine whether there is an abnormality or serious fault in the voltage sensor; then the corresponding characteristics of each single battery are calculated separately and , and get the corresponding eigenvector 、 , then calculate 、 And the voltage sequence at the time of diagnosis Corresponding 、 、 , by comparing with the corresponding threshold, the specific ALOF value is as follows Figure 6 , it can be found that the No. 12 single cell has an inconsistency risk, and the No. 53, No. 54, and No. 55 single cells have abnormal voltage fluctuations and have recently experienced abnormal voltage rises and falls; therefore, the No. 12, No. 53, No. 54, and No. 55 single cells are determined to be abnormal batteries and require further internal short circuit diagnosis.

[0135] All cells except cells 12, 53, 54, and 55 were selected as representative of normal cells during this period. The voltage mean series of these normal cells was calculated. The Mahalanobis distances between the voltage mean series and the first 60 voltage data points for these four cells were then calculated. The values were 1.71, 2.37, 2.11, and 2.25, respectively, exceeding the threshold of 1.12. Therefore, these four cells were diagnosed as having internal short circuit failures. Furthermore, the degree of deviation of their voltage data from the normal cell voltage mean series varied. This suggests that cells with greater deviations from the normal cell voltage mean series are more likely to have a more severe failure, which is consistent with the observed deviations in the fault voltage curve.

[0136] The fault diagnosis method of the present invention is suitable for real-time monitoring of battery status and multi-fault diagnosis under real-world operating conditions. This method employs a two-tiered diagnostic strategy. The first tier extracts multi-level features from battery voltage data and uses an improved Advanced Local Outlier Factor (ALOF) algorithm to accurately identify various abnormalities in abnormal cells within a battery pack. The second tier uses Mahalanobis distance to further diagnose internal short-circuit faults in abnormal cells. To extract fault features at a more granular level, kurtosis and rate of change are extracted from different time windows of voltage data after eliminating environmental and sensor errors. The ALOF algorithm is then used to perform preliminary anomaly screening based on voltage variations at different time scales for each cell. This method not only diagnoses multiple anomalies but also locates the abnormal cell. Because voltage variation trends change with battery operating conditions and aging, fixed thresholds and comparison targets can lead to missed diagnoses. Therefore, the method uses the average voltage data of normal cells, screened out after the first tier of diagnosis, which reflects the current battery operating conditions and aging, as a comparison target. Mahalanobis distance is then used to further diagnose internal short-circuit faults in cells identified as abnormal in the first tier. This invention takes into account the impact of battery operating conditions and aging changes on fault diagnosis, and has stronger adaptability in actual operating conditions. At the same time, the dual-layer diagnosis strategy that integrates the improved outlier algorithm can realize multi-fault diagnosis and abnormal battery location. The algorithm is efficient and effectively solves the problems of misdiagnosis caused by changes in operating conditions and aging, and missed diagnosis of adjacent abnormal single cells.

[0137] In summary, the new energy vehicle power battery fault diagnosis method provided by the embodiment of the present invention uses multi-time-scale features to characterize the battery status. Combined with the improved ALOF algorithm, it can more accurately locate abnormal single cells and identify multiple abnormal conditions. The battery fault diagnosis method further diagnoses the internal short-circuit fault based on the location of the abnormal single cell. The Mahalanobis distance algorithm based on the mean voltage data of normal batteries in the same period can more efficiently and accurately diagnose the fault and reduce misdiagnosis. The overall solution has the advantages of diagnostic efficiency, accuracy, and robustness.

[0138] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A multi-fault diagnosis method for power batteries of new energy vehicles, characterized by: The following steps are involved: S1: Obtain the voltage data of all single cells in the real vehicle battery pack during operation before the sampling time, and construct the voltage matrix U of the battery pack n,c , c represents the cth single cell in the battery pack, n represents n sampling times, V0 is used as the voltage sequence at the detection time, and the voltage data at the detection time is subjected to preliminary abnormality detection according to the nominal upper and lower limit voltage values of the battery when it leaves the factory; S2: Using voltage matrix U n,c The voltage data of 60 moments before the current moment are included in the calculation of the kurtosis of the voltage data of each single cell in the time window, and then the kurtosis values of all single cells are combined to generate the kurtosis feature vector K; S3: Calculate the voltage change rate of each single cell in the time window using the voltage data of the six moments before the current moment in the voltage matrix, and then combine the voltage change rates of all single cells to generate a voltage change rate feature vector R; S4: Bring the kurtosis eigenvector K, the voltage change rate eigenvector R and the voltage sequence V0 at the detection time into the ALOF algorithm for calculation, and obtain the corresponding ALOF values, namely ALOF K 、ALOF R 、ALOF V , judge ALOF K 、ALOF R 、ALOF V Whether any item exceeds the corresponding threshold T K 、T R 、T V , if it exceeds the corresponding threshold, execute step S5; S5: Calculate the Mahalanobis distance M by comparing the historical voltage sequence of the abnormal single cell with the voltage mean of the normal cells in the same period. i Greater than the threshold T M When , it is determined that this single battery has an internal short circuit fault. The specific expression is as follows: Among them, M i The Mahalanobis distance between the historical voltage sequence of abnormal single cells and the mean voltage of normal cells in the same period is calculated, V abnormal represents the historical voltage data of abnormal single cells, the superscript T represents transposition, μ represents the historical voltage mean vector of normal cells, and Σ represents the historical voltage covariance matrix of normal cells. Its expression is as follows: Among them, V b Represents the voltage value of the bth sampling point, and N represents the total number of voltage data.

2. The multi-fault diagnosis method for a new energy vehicle power battery according to claim 1, characterized in that: In step S1, before constructing the voltage matrix, the acquired data is preprocessed. The preprocessing algorithm uses a wavelet threshold noise reduction algorithm to perform wavelet decomposition on the original voltage data to obtain the wavelet coefficients of each layer, where the wavelet coefficient formula is: Where z represents the number of sampling points, and log(z) represents the logarithm of the number of sampling points; Then, the abnormal data points are judged and removed according to the threshold value, and the soft threshold function W is selected as the threshold function, and its expression is: Among them, w is the wavelet coefficient, sgn() is the sign function, and λ is the critical threshold.

3. The multi-fault diagnosis method for a new energy vehicle power battery according to claim 1, characterized in that: The kurtosis k of the cth single cell in the kurtosis eigenvector K is c The specific expression is as follows: Among them, k c represents the kurtosis of the cth single cell, n l Indicates the window size selected for the long time scale, v i Represents the voltage value of the i-th sampling in the time window starting from the current moment, Indicates the mean voltage in the selected window, μ c represents the mean value of the voltage data of the cth single cell in the time window, σ c Indicates the standard deviation of the voltage data of the cth single cell in the time window.

4. The multi-fault diagnosis method for a new energy vehicle power battery according to claim 1, characterized in that: The voltage change rate r of the cth single cell in the voltage change rate characteristic vector R is c The calculation formula is: Among them, n s Indicates the window size selected for the short time scale, v s Indicates the voltage value at the left end of the time window, v t Indicates the voltage value at the current moment.

5. The multi-fault diagnosis method for a new energy vehicle power battery according to claim 1, characterized in that: The specific steps of the ALOF algorithm in step S4 are as follows: S4.1: Calculate the reachable distance d(y,x) between objects y and x: Among them, y m and x m are the mth eigenvalues of objects y and x, respectively, j is the total number of eigenvalues, Indicates the sum of j data; S4.2: Set an initial domain size k, and the reachable distance of object x to y within the domain size is: d k (x)=d(y k ,x) At this time, the initial neighborhood N of object x k (x) is expressed as: N k (x)={y∈D|d(y,x)≤d k (x)} Where D represents the entire data set, d k (x) represents the reachable distance of object x to y within the domain size, d(y k ,x) represents the distance between any object in domain k and object x; S4.3: Calculate the local density of object x as: Among them, d local (x) represents the average distance between object x and the objects in its initial neighborhood, which is used to reflect the local density, |N k (x)| represents the number of objects contained in the initial neighborhood of object x, represents the sum of the distances between object x and all objects y in its initial neighborhood; S4.4: Calculate the local distance d of all objects in the entire dataset local Average value of (x) Where |D| refers to the total number of objects in the dataset, ∑ x∈D Indicates summation; The dynamically changing neighborhood size k′(x) corresponding to each object is determined based on the local density and average local density of the data: Among them, α is the proportional coefficient, which is used to control the change range of the neighborhood size. Indicates rounding up to ensure that k′(x) is an integer; S4.5: For the new domain value k′, the reachable distance of object x to y within the domain size is: d k′ (x)=d(y k′ ,x) Among them, d(y k′ ,x) represents the distance between any object in the domain k′ and object x. The neighborhood including object x itself under the new neighborhood value k′ is expressed as: N k′ (x)={y∈D∣d(y,x)≤d k′ (x)}∪{x} The new local density ρ(x) is: where |N k′ (x)| represents the number of objects contained in the corresponding dynamic neighborhood of object x, d k′ (y,x) represents the reachable distance in the dynamically adjusted neighborhood, For all y∈N k′ (x) is summed, N k′ (x) represents the neighborhood of the object x itself under the new neighborhood value k′, and its expression is as follows: For object y, repeat the same process as object x to obtain the new local density ρ(y); S4.6 The revised ALOF value is updated as follows: Where ALOF(x) represents the ALOF value of x; The threshold T K 、T R 、T V The specific expression is as follows: T p =μ p +2s p Among them, p represents one of the eigenvalues K, R, V, ALOF p is the ALOF value corresponding to p, μ p ALOF of all single cells under characteristic value p p The mean of the values, σ p ALOF of all single cells under this characteristic value p The standard deviation of the values.

Images