A new energy vehicle power battery fault diagnosis method and system

Abstract

The present invention discloses a method and system for diagnosing faults in power batteries of new energy vehicles, and belongs to the technical field of battery monitoring of new energy vehicles. The method comprises: obtaining continuous time series data of voltages of each single cell in a power battery pack of a new energy vehicle to form a voltage matrix A; performing feature enhancement on the voltage matrix A to obtain a feature matrix B; sliding a sliding window with a preset length of h and a step length of 1 from top to bottom across the feature matrix B to obtain a feature matrix E under the current window; calculating the entropy value of each column of the feature value sequence in the feature matrix E under the current window by an entropy value method to obtain an entropy sequence under the current window; using an improved local distance-based anomaly detection method combined with a threshold method to identify and locate the values ​​in the entropy sequence under the current window; the present invention amplifies the abnormal data in the voltage signal by feature enhancement, converts the battery fault problem into a battery voltage outlier detection problem, and improves the battery fault diagnosis capability.

Description

Technical Field

[0001] The present invention relates to the technical field of battery monitoring for new energy vehicles, and in particular to a method and system for diagnosing faults of power batteries for new energy vehicles. Background Art

[0002] At present, voltage data combined with entropy method has been applied to the field of battery fault diagnosis and has achieved certain results, such as Shannon entropy method and sample entropy method. However, there are also some disadvantages. Shannon entropy method is easily affected by random errors, resulting in false alarms by identifying fault-free single cells as faulty cells; the diagnostic effect of sample entropy method is affected by scale factors, which is specifically related to the reconstruction vector size m and similarity r. The calculation window or reconstruction vector m and similarity r that play a decisive role in fault diagnosis are mainly obtained in the form of trial and error, which is labor-intensive. Summary of the invention

[0003] In order to solve the problems of the prior art, the present invention provides a method and system for diagnosing power battery faults in new energy vehicles.

[0004] On the one hand, a method for diagnosing a fault of a power battery of a new energy vehicle is provided, the method comprising:

[0005] Obtain the continuous time series data of the voltage of each single cell in the power battery pack of the new energy vehicle to form a voltage matrix A;

[0006] Performing feature enhancement on the voltage matrix A to obtain a feature matrix B;

[0007] According to a sliding window with a preset length of h and a step size of 1, the feature matrix B is drawn from top to bottom to obtain a feature matrix E under the current window;

[0008] Calculate the entropy value of each column of the eigenvalue sequence in the feature matrix E under the current window by using the entropy method to obtain the entropy sequence under the current window;

[0009] The improved local distance-based anomaly detection method combined with the threshold method is used to identify and locate the outliers in the entropy sequence under the current window to complete the power battery fault diagnosis.

[0010] Furthermore, the feature enhancement uses local gain or global gain.

[0011] Furthermore, the global gain formula is as follows:

[0012] ;

[0013] in, It represents the median value of all single battery voltages at the tth sampling time. represents the voltage value of the ith single cell at the tth sampling time, It represents the voltage value of the ith single cell after global gain at the tth sampling time;

[0014] The specific steps of the global gain are as follows:

[0015] Determine the median of all single cell voltage values ​​at each sampling moment;

[0016] Determine the absolute value of the difference between the voltage values ​​of all single cells and the median of the voltage values ​​of all single cells at each sampling moment, and define the absolute value of the difference as , represents the absolute value of the difference between the voltage value of single cell i at the tth sampling time and the median of the voltage values ​​of all single cells at the tth sampling time;

[0017] by As the calculation formula, the characteristic enhancement of the voltage of the single cell i at the tth sampling time obtained by the global gain can be obtained.

[0018] Furthermore, the local gain formula is as follows:

[0019] ;

[0020] in, It represents the cumulative sum of the voltage difference between single cell i and its first nearest neighbor single cell and the voltage difference between single cell i and its (K-1)th nearest neighbor single cell. represents the voltage difference between the single cell i and its nearest neighboring single cell k, k = 1, 2, ..., K;

[0021] The specific steps of the local gain are as follows:

[0022] Determine the median of all single cell voltage values ​​at each sampling moment;

[0023] Select the K value, calculate the local gain of all single cell voltage values ​​and the local gain of the median of all single cell voltage values ​​at each sampling moment;

[0024] Feature enhancement can be obtained by taking the absolute value of the difference between the local gain of all single cell voltage values ​​at each sampling moment and the local gain of the median of all single cell voltage values.

[0025] Furthermore, the feature matrix E is as follows:

[0026] ;

[0027] Where c is the number of windows, c=T-h+1, T is the number of samples, It represents the voltage of all single cells after feature enhancement in the cth window with a length of h. It represents the voltage value of the nth single cell at the tth time in the cth window with a length of h after characteristic enhancement.

[0028] Furthermore, the entropy value method adopts an improved fuzzy entropy method, and the specific steps include:

[0029] From the feature matrix E, each column of the feature value sequence is reorganized into a vector sequence with a dimension of m ;

[0030] in, ;

[0031] for The median value of the series;

[0032] In the vector sequence Define two m-dimensional vectors in and The distance between ;

[0033] Define fuzzy membership m, measure and Similarity ,Right now

[0034] ;

[0035] definition ;

[0036] Repeat the above steps, increase the dimension to (m+1), and calculate and The distance between , define fuzzy membership (m+1), measure and Similarity ,Right now

[0037] ;

[0038] Define fuzzy entropy, .

[0039] Furthermore, the distance Choose one of the Manhattan distance, Euclidean distance, and cosine distance.

[0040] Furthermore, the method of using the improved local distance-based anomaly detection method combined with the threshold method to identify and locate the values ​​in the entropy sequence under the current window specifically includes:

[0041] The clustering algorithm is used to classify the improved fuzzy entropy values ​​in the entropy sequence of all columns in the feature matrix E under the current window, and the improved fuzzy entropy values ​​corresponding to each category are obtained;

[0042] Calculate the average distance from all improved fuzzy entropy values ​​in each class under the current window to the current cluster center ;

[0043] Eliminate the clusters whose distance from each class to the current cluster center is less than the average distance in the current window. Improved fuzzy entropy value of;

[0044] The remaining improved fuzzy entropy values ​​retained in all classes under the current window are used as local anomaly factors;

[0045] The threshold is set by the threshold method: the sliding windows with lengths in the interval [10, 30] are set respectively, the local abnormal factors of the improved fuzzy entropy values ​​of normal single cells in the power battery pack under each length window are calculated, and the maximum value of the local abnormal factors under all length windows is taken as the set threshold;

[0046] Through sliding window iteration, the local abnormality factor based on the improved fuzzy entropy value of each battery cell in each window is obtained, and compared with the local abnormality factor based on the improved fuzzy entropy value of each battery cell in the current window through the set threshold. If no local abnormality factor exceeds the threshold, it is judged that there is no abnormal single cell in the current window, and the sliding window continues to move downward. The local abnormality factor of the improved fuzzy entropy value of each battery cell in the next window is judged according to the above method. If there is a local abnormality factor of the improved fuzzy entropy value in the current window that exceeds the threshold, the single cell corresponding to the local abnormality factor of the improved fuzzy entropy value is the faulty single cell.

[0047] Furthermore, the clustering algorithm is any one of K-Means clustering, DBSCAN clustering, and GMM clustering.

[0048] On the other hand, a new energy vehicle power battery fault diagnosis system is provided, comprising: a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor can implement the steps of the new energy vehicle power battery fault diagnosis method when executing the computer program.

[0049] The beneficial effects of the technical solution provided by the embodiment of the present invention are as follows: in the present invention, the voltage matrix A is formed by acquiring the continuous time series data of the voltage of each single cell in the battery pack, the features of the voltage matrix A are enhanced, the abnormal data in the voltage signal is amplified, and the battery fault problem is converted into a battery voltage outlier detection problem, thereby improving the battery fault diagnosis capability and preventing a single cell without fault from being identified as a faulty cell to generate a false alarm.

[0050] Secondly, the present invention can realize real-time battery status diagnosis in combination with the sliding window method, which has no obvious excessive dependence on the sliding window length, thereby reducing the misdiagnosis rate.

[0051] In addition, the present invention proposes to use an improved local distance-based anomaly detection method and a clustering algorithm to screen out suspicious entropy values ​​to be further diagnosed for local anomaly coefficient calculation, which greatly shortens the amount of calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0053] Figure 1 It is a flow chart of a method for diagnosing a fault of a power battery of a new energy vehicle provided by the present invention;

[0054] Figure 2 The present invention provides a voltage timing diagram of a vehicle when a battery failure occurs on the same day;

[0055] Figure 3 The present invention provides a method based on Figure 2 A partial enlarged view of the voltage time series data of all single cells (95 cells) at the time of the fault, i.e., 80 frames before and after the 4848th frame of data;

[0056] Figure 4 It is a battery fault diagnosis result diagram of an improved fuzzy entropy method based on local feature enhancement provided by the present invention;

[0057] Figure 5 This is a local enlarged diagram of the battery fault diagnosis result of an improved fuzzy entropy method based on local feature enhancement provided by the present invention;

[0058] Figure 6 It is a battery fault diagnosis result diagram based on the common fuzzy entropy method provided by the present invention;

[0059] Figure 7 This is a local enlarged diagram of battery fault diagnosis based on a common fuzzy entropy method provided by the present invention. DETAILED DESCRIPTION

[0060] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0061] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0062] Embodiment 1

[0063] A method for diagnosing a fault in a power battery of a new energy vehicle comprises the following steps:

[0064] Step (1): Obtain the continuous time series data of the voltage of each single cell in the power battery pack of the new energy vehicle and form a voltage matrix A as shown below:

[0065] ;

[0066] Among them, T is the number of samples, N is the total number of single cells in the power battery pack, Indicates the voltage value of the Nth single cell at the Tth sampling time.

[0067] Step (2): Perform feature enhancement on the voltage matrix A to obtain the feature matrix B.

[0068] Among them, feature enhancement uses one of global gain or local gain, and the global gain formula is as follows:

[0069] ;

[0070] in, It represents the median of all single cell voltage values ​​at the tth sampling time. represents the voltage value of the ith single cell at the tth sampling time, It represents the voltage value of the ith single cell after global gain at the tth sampling time.

[0071] The specific steps of global gain are as follows:

[0072] Determine the median of all single cell voltage values ​​at each sampling moment;

[0073] Determine the absolute value of the difference between the voltage values ​​of all single cells and the median of the voltage values ​​of all single cells at each sampling moment, and define the absolute value of the difference as , represents the absolute value of the difference between the voltage value of single cell i at the tth sampling time and the median of the voltage values ​​of all single cells at the tth sampling time;

[0074] by As the calculation formula, the characteristic enhancement of the voltage value of the single cell i at the tth sampling time obtained by the global gain can be obtained.

[0075] The local gain formula is as follows:

[0076] ;

[0077] in, It represents the cumulative sum of the voltage difference between single cell i and its first nearest neighbor single cell and the voltage difference between single cell i and its (K-1)th nearest neighbor single cell. represents the voltage difference between single cell i and its nearest neighboring single cell k, where k=1, 2, ..., K.

[0078] It should be noted that the nearest neighbor single cell specifically refers to: except for the single cell itself, the absolute values ​​of the voltage difference between the single cell and other single cells at the current sampling moment are arranged from small to large, and the first K values ​​are taken, among which the single cell corresponding to the first value is the first nearest neighbor single cell of the single cell, ..., the single cell corresponding to the last value is the Kth nearest neighbor single cell of the single cell. When determining the nearest neighbor single cell of each single cell, all single cell voltages at the current sampling moment can be first clustered into two clusters, and then the nearest neighbor single cells of each single cell under the cluster can be found in each cluster. Such a pre-classification method can effectively avoid global calculation of each single cell and improve calculation efficiency.

[0079] In this embodiment, the local gain method is improved by introducing the concept of local gain of the median battery voltage, and the abnormal data is further amplified by comparing the local gain of all battery voltages with the local gain of the median.

[0080] The method of feature enhancement using local gain is as follows:

[0081] Determine the median of all single cell voltage values ​​at each sampling moment; select a suitable K value to calculate the local gain of all single cell voltages at each sampling moment and the local gain of the median of all single cell voltage values, in this embodiment K=5; subtract the local gain of all single cell voltage values ​​at each sampling moment from the local gain of the median of all single cell voltage values ​​and take the absolute value to obtain feature enhancement.

[0082] Step (3): According to the sliding window with a preset length of h and a step size of 1, slide across the feature matrix B from top to bottom to obtain the feature matrix E under the current window, as shown below:

[0083] ;

[0084] Where c is the number of windows, c=T-h+1, It represents the voltage of all single cells after feature enhancement in the cth window with a length of h. It represents the voltage value of the nth single cell at the tth time in the cth window with a length of h after characteristic enhancement.

[0085] Step (4): Calculate the entropy value of each column of the feature value sequence in the feature matrix E under the current window by using the entropy method to obtain the entropy sequence under the current window. The entropy method adopts the improved fuzzy entropy method. The specific steps of the improved fuzzy entropy method are as follows:

[0086] S41: Extract the feature matrix E under the current window in step (3).

[0087] S42: Reorganize the eigenvalue sequence of each column from the feature matrix E into a vector sequence of dimension m .

[0088] in, , for The median value of the series.

[0089] S43: In vector sequence Define two m-dimensional vectors in and The distance between , where the distance Choose one of the Manhattan distance, Euclidean distance, and cosine distance.

[0090] Manhattan distance That is, find the vector and The absolute value of the difference between the values ​​of each element in each row is as follows:

[0091] ;

[0092] Euclidean distance That is, find the vector and Each row in the table corresponds to the square root of the sum of the squares of the differences in the values ​​of each element. The formula is as follows:

[0093] ;

[0094] Cosine distance That is, find the vector and The cosine value of the angle is as follows:

[0095] ;

[0096] In this embodiment, Manhattan distance is selected as the two m-dimensional vectors and The distance between .

[0097] S44: Define fuzzy membership m, measure and The similarity of .

[0098] S45: Definition .

[0099] S46: Repeat steps S42 to S45, increase the dimension to (m+1), and calculate and Manhattan distance between , define fuzzy membership (m+1), measure and Similarity ,definition .

[0100] S47: Define fuzzy entropy, .

[0101] Step (5): Use the improved local distance-based anomaly detection method combined with the threshold method to identify and locate the outliers in the entropy sequence under the current window to complete the power battery fault diagnosis.

[0102] The improved local distance-based anomaly detection method is specifically a local outlier factor detection method based on clustering. The clustering algorithm is any one of K-Means clustering, DBSCAN clustering, and GMM clustering. In this embodiment, the clustering algorithm is K-Means clustering.

[0103] The local outlier factor detection method based on clustering is as follows:

[0104] Use the clustering algorithm to classify the improved fuzzy entropy values ​​in the entropy sequence of all columns in the feature matrix E under the current window, and obtain the improved fuzzy entropy values ​​under each category; calculate the average distance from all improved fuzzy entropy values ​​in each class under the current window to the current cluster center. ; Eliminate the distance from each class in the current window to the current cluster center that is less than the average distance The improved fuzzy entropy value of the current window is used as the local anomaly factor for the remaining improved fuzzy entropy values ​​retained in all classes under the current window.

[0105] The threshold setting method in the threshold method is specifically as follows: set the sliding window with a length in the interval [10, 30] respectively, calculate the local abnormality factor of the improved fuzzy entropy value of each normal single battery in the power battery pack under each length window, count the maximum value of the local abnormality factor of each length window under all windows, and take the maximum value of the local abnormality factor under all length windows as the set threshold.

[0106] The specific strategy for identifying and locating outliers for the improved fuzzy entropy values ​​in the current window entropy sequence is as follows: through step (4) sliding window iteration, the local abnormality factor based on the improved fuzzy entropy value of each battery cell in each window is obtained, and the local abnormality factor based on the improved fuzzy entropy value of each battery cell in the current window is compared with the local abnormality factor based on the improved fuzzy entropy value of each battery cell in the current window through a set threshold. If no local abnormality factor exceeds the threshold, it is judged that there is no abnormal single cell in the current window, and the sliding window continues to move downward, and the local abnormality factor of the improved fuzzy entropy value of each battery cell in the next window is judged according to the above method; if there is a local abnormality factor of the improved fuzzy entropy value in the current window that exceeds the threshold, the single cell corresponding to the local abnormality factor of the improved fuzzy entropy value is the faulty single cell.

[0107] Embodiment 2

[0108] A new energy vehicle power battery fault diagnosis system includes: a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the steps of the new energy vehicle power battery fault diagnosis method in embodiment one can be implemented.

[0109] It should be noted that Figure 2 This is the voltage sequence diagram of all cells in the power battery one hour before the failure. There are 95 cells in the power battery, all connected in series. There are 5263 frames of data, each with an interval of 10s. It can be seen that the voltage sequence data is in the normal voltage amplitude fluctuation range for most of the time. The data trend includes the voltage rise interval in the charging stage and the voltage drop interval in the discharge stage. It can be seen that the voltage consistency of each cell is strong. Only a small number of cells have drastic voltage fluctuations in the early stage of the terminal, and the voltage of a large range of batteries increases and decreases sharply at the last moment.

[0110] Figure 3 Based on Figure 2From the partial enlarged diagram of the voltage timing data of all single cells (95 cells) at the time of the fault, i.e., the 80 frames before and after the 4848th frame data, it can be clearly seen that in the 39 frames of voltage timing data before the 4848th frame, all voltages were within the normal operating voltage range, i.e., between 4.1 and 4.2V. At the 4848th frame, the voltages of single cells No. 59, No. 60, No. 61, No. 63, and No. 65 all fluctuated violently, among which the voltages of single cells No. 59, No. 60, and No. 63 were all lower than the average voltage value of all battery voltages, and the voltage of single cell No. 63 was lower than 1V, indicating that there was a short circuit inside the battery; the voltages of single cells No. 61 and No. 65 were both higher than the standard operating voltage range of the battery, and further, the voltage of single cell No. 65 remained at around 4.9V, indicating that single cell No. 65 was faulty.

[0111] Figure 4 This is the battery fault diagnosis result diagram of the improved fuzzy entropy method based on local feature enhancement; Figure 5 This is a local enlarged view of the battery fault diagnosis results of the improved fuzzy entropy method based on local feature enhancement; Figure 6 This is the battery fault diagnosis result diagram based on the common fuzzy entropy method; Figure 7 This is a local enlarged view of battery fault diagnosis based on ordinary fuzzy entropy method.

[0112] Figure 4 and Figure 6 Both fuzzy entropy methods use the same sliding window length of 50, the reconstructed vector size m dimension is 2, the similarity r=0.25, and the number of neighbors for calculating the local anomaly factor is 15; Figure 4 It can be seen from the figure that the maximum local outlier factor before iteration 4799 appears at iteration 4015, with a value of 65.824. Figure 5 It can be seen that in the 4799th iteration window, the local outlier factor values ​​of five monomers deviate from the local outlier factors of most monomers. In this embodiment, the threshold is selected as 100, then in the current 4799th iteration window, only monomers No. 60 and No. 61 exceed the threshold, but after the 4800th iteration, monomers No. 59, No. 60, No. 61, No. 63, and No. 65 all exceed the threshold, and the threshold-exceeding state continues.

[0113] from Figure 6 It can be seen from the figure that the local outlier factors of each single cell before the 4799th iteration are all below 20. In this embodiment, 20 is selected as the safety threshold. Figure 7It can be seen that at the 4799th iteration, cells 59, 60, 61, 63, and 65 all exceeded the threshold of 20, and the local outlier factor of cell 63 was 27.7877. However, it can be seen that at the 4800th iteration, the local outlier factors of all cells dropped below 20. At the 4802nd iteration, cells 60 and 61 exceeded the threshold and continued to do so. During this period, cells 59 and 65 also exceeded the threshold and continued to do so. Other normal cells also exceeded the threshold. At the 4817th iteration, cell 63 exceeded the threshold again. However, in reality, after 4848 frames of data, the voltage of cell 63 remained at 0.9, indicating an internal short circuit in the battery.

[0114] In this embodiment, although the improved fuzzy entropy method battery fault diagnosis result based on feature enhancement diagnosed two fewer faulty cells at the 4799th iteration than the battery fault diagnosis result based on the ordinary fuzzy entropy method, the diagnosis result of the present invention was very stable after the 4800th iteration, and there was no missed diagnosis or misjudgment. Although the battery fault diagnosis method based on the ordinary fuzzy entropy method diagnosed five faulty cells at the 4799th iteration, no faulty cells were diagnosed at the 4800-4801 iterations, and missed diagnosis and misdiagnosis occurred at the 4801-4816 iterations, indicating that the accuracy and stability of this method are not as good as the improved fuzzy entropy method battery fault diagnosis result based on feature enhancement.

[0115] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0116] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for diagnosing faults in power batteries of new energy vehicles, characterized in that: The method comprises: Obtain the continuous time series data of the voltage of each single cell in the power battery pack of the new energy vehicle to form a voltage matrix A; Performing feature enhancement on the voltage matrix A to obtain a feature matrix B, wherein the feature enhancement uses a local gain or a global gain; The local gain formula is as follows: ； in, It represents the cumulative sum of the voltage difference between single cell i and its first nearest neighbor single cell and the voltage difference between single cell i and its (K-1)th nearest neighbor single cell. represents the voltage difference between the single cell i and its nearest neighboring single cell k, k = 1, 2, ..., K; The specific steps of the local gain are as follows: Determine the median of all single cell voltage values ​​at each sampling moment; Select the K value, calculate the local gain of all single cell voltage values ​​and the local gain of the median of all single cell voltage values ​​at each sampling moment; The feature enhancement can be obtained by taking the absolute value of the difference between the local gain of all single-cell voltage values ​​at each sampling moment and the local gain of the median of all single-cell voltage values; According to a sliding window with a preset length of h and a step size of 1, the feature matrix B is drawn from top to bottom to obtain a feature matrix E under the current window; For each column of the eigenvalue sequence in the feature matrix E under the current window, an improved fuzzy entropy value is calculated by an improved fuzzy entropy method to obtain an entropy sequence under the current window; The improved local distance-based anomaly detection method combined with the threshold method is used to identify and locate the values ​​in the entropy sequence under the current window to complete the power battery fault diagnosis, which specifically includes: The clustering algorithm is used to classify the improved fuzzy entropy values ​​in the entropy sequence of all columns in the feature matrix E under the current window, and the improved fuzzy entropy values ​​corresponding to each category are obtained; Calculate the average distance from all improved fuzzy entropy values ​​in each class under the current window to the current cluster center ; Eliminate the clusters whose distance from each class to the current cluster center is less than the average distance in the current window. Improved fuzzy entropy value of; The remaining improved fuzzy entropy values ​​retained in all classes under the current window are used as local anomaly factors; The threshold is set by the threshold method: the sliding windows with lengths in the interval [10, 30] are set respectively, the local abnormal factors of the improved fuzzy entropy values ​​of normal single cells in the power battery pack under each length window are calculated, and the maximum value of the local abnormal factors under all length windows is taken as the set threshold; Through sliding window iteration, the local abnormality factor based on the improved fuzzy entropy value of each battery cell in each window is obtained, and compared with the local abnormality factor based on the improved fuzzy entropy value of each battery cell in the current window through the set threshold. If no local abnormality factor exceeds the threshold, it is judged that there is no abnormal single cell in the current window, and the sliding window continues to move downward. The local abnormality factor of the improved fuzzy entropy value of each battery cell in the next window is judged according to the above method. If there is a local abnormality factor of the improved fuzzy entropy value in the current window that exceeds the threshold, the single cell corresponding to the local abnormality factor of the improved fuzzy entropy value is the faulty single cell.

2. The new energy vehicle power battery fault diagnosis method according to claim 1, characterized in that: The global gain formula is as follows: ； in, It represents the median of all single cell voltage values ​​at the tth sampling time. represents the voltage value of the ith single cell at the tth sampling time, It represents the voltage value of the ith single cell after global gain at the tth sampling time; The specific steps of the global gain are as follows: Determine the median of all single cell voltage values ​​at each sampling moment; Determine the absolute value of the difference between the voltage values ​​of all single cells and the median of the voltage values ​​of all single cells at each sampling moment, and define the absolute value of the difference as , represents the absolute value of the difference between the voltage value of single cell i at the tth sampling time and the median of the voltage values ​​of all single cells at the tth sampling time; by As the calculation formula, the characteristic enhancement of the voltage value of the single cell i at the tth sampling time obtained by the global gain can be obtained.

3. The new energy vehicle power battery fault diagnosis method according to claim 2, characterized in that: The feature matrix E is as follows: ； Where c is the number of windows, c=T-h+1, T is the number of samples, It represents the voltage of all single cells after feature enhancement in the cth window with a length of h. It represents the voltage value of the nth single cell at the tth time in the cth window with a length of h after characteristic enhancement.

4. The new energy vehicle power battery fault diagnosis method according to claim 3, characterized in that: The improved fuzzy entropy method specifically comprises the following steps: From the feature matrix E, each column of the feature value sequence is reorganized into a vector sequence with a dimension of m ; in, ; for The median value of the series; In the vector sequence Define two m-dimensional vectors in and The distance between ; Define fuzzy membership m, measure and Similarity ,Right now ； definition ; Repeat the above steps, increase the dimension to (m+1), and calculate and The distance between , define fuzzy membership (m+1), measure and Similarity ,Right now ； Define fuzzy entropy, .

5. The new energy vehicle power battery fault diagnosis method according to claim 4, characterized in that: The distance Choose one of the Manhattan distance, Euclidean distance, and cosine distance.

6. The new energy vehicle power battery fault diagnosis method according to claim 5, characterized in that: The clustering algorithm is selected from any one of K-Means clustering, DBSCAN clustering, and GMM clustering.

7. A new energy vehicle power battery fault diagnosis system, characterized in that: include: A memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, the steps of the new energy vehicle power battery fault diagnosis method as described in any one of claims 1 to 6 can be implemented.

Images