Battery fault diagnosis method and device based on multi-feature fusion, equipment and medium

Abstract

The invention relates to a battery fault diagnosis method and device based on multi-feature fusion, equipment and a medium. The method comprises the following steps: performing feature extraction according to electrochemical impedance spectroscopy data of each to-be-detected battery to obtain a multi-dimensional electrochemical impedance spectroscopy feature of each to-be-detected battery, and fusing the multi-dimensional electrochemical impedance spectroscopy feature of each to-be-detected battery to determine a fused feature vector of each to-be-detected battery, acquiring a clustering parameter range of each fusion feature vector, constructing a plurality of groups of initial parameters according to the clustering parameter range, performing clustering processing on each fusion feature vector according to each group of initial parameters, performing clustering parameter optimization processing according to each clustering processing result to obtain optimized parameters, and performing clustering processing on each fusion feature vector according to the optimized parameters. And performing fault diagnosis on each to-be-detected battery according to the optimized parameters and each fusion feature vector to determine a fault diagnosis type of each to-be-detected battery. By adopting the method, the battery fault diagnosis can be realized based on the multi-dimensional electrochemical impedance spectroscopy characteristics of each battery to be detected, and the correctness of the battery fault diagnosis is improved.

Description

Technical Field

[0001] The present application relates to the field of power electronics technology, and in particular to a battery fault diagnosis method, apparatus, device and medium based on multi-feature fusion. Background Art

[0002] With the development of electronic technology, batteries are widely used in various fields, such as electronics, transportation, energy storage, medical care, and industry. Frequent battery failures are common during long-term use. To save battery maintenance costs, regular battery fault diagnosis and timely repair of faulty batteries are necessary.

[0003] In the related art, the frequency domain characteristic detection of the battery is mainly used to diagnose the battery fault. However, the battery fault diagnosis in the related art has the problem of low accuracy of the diagnosis result. Summary of the Invention

[0004] Based on this, it is necessary to provide a battery fault diagnosis method, device, equipment and medium based on multi-feature fusion to address the above technical problems.

[0005] In a first aspect, the present application provides a battery fault diagnosis method based on multi-feature fusion, comprising:

[0006] Extracting features from the electrochemical impedance spectroscopy data of each battery to be tested to obtain multi-dimensional electrochemical impedance spectroscopy features of each battery to be tested, and fusing the multi-dimensional electrochemical impedance spectroscopy features of each battery to be tested to determine a fused feature vector of each battery to be tested;

[0007] Obtaining the clustering parameter range of the fused feature vector of each battery to be detected, and constructing multiple groups of initial parameters based on the clustering parameter range; each group of initial parameters includes different clustering parameters;

[0008] Clustering is performed on the fused feature vectors of each battery to be tested according to each group of initial parameters, and clustering parameter optimization is performed based on the clustering processing results to obtain optimized parameters;

[0009] According to the optimized parameters and the fused feature vectors of each battery to be detected, fault diagnosis is performed on each battery to be detected, and the fault diagnosis type of each battery to be detected is determined.

[0010] In one embodiment, the multi-dimensional electrochemical impedance spectroscopy characteristics of each battery to be tested are fused to determine the fused feature vector of each battery to be tested, including:

[0011] For any battery to be tested, an initial feature vector of the battery to be tested is obtained according to the multi-dimensional electrochemical impedance spectroscopy characteristics of the battery to be tested;

[0012] Normalizing the initial feature vector of the battery to be tested to obtain a standardized feature vector of the battery to be tested;

[0013] The standardized feature vector of the battery to be detected is subjected to dimensionality reduction processing to obtain a fused feature vector of the battery to be detected.

[0014] In one embodiment, the multidimensional electrochemical impedance spectroscopy characteristics include frequency domain characteristics, equivalent circuit characteristics, and statistical characteristics; obtaining an initial feature vector of the battery to be tested based on the multidimensional electrochemical impedance spectroscopy characteristics of the battery to be tested includes:

[0015] Obtaining a first weight coefficient of the frequency domain characteristics of the battery to be detected, a second weight coefficient of the equivalent circuit characteristics of the battery to be detected, and a third weight coefficient of the statistical characteristics of the battery to be detected;

[0016] Weighted feature concatenation is performed according to the frequency domain features, the equivalent circuit features, the statistical features, and the first weight coefficient, the second weight coefficient, and the third weight coefficient to obtain an initial feature vector of the battery to be detected.

[0017] In one embodiment, the clustering parameter range includes a neighborhood radius range and a minimum neighborhood point number range; multiple sets of initial parameters are constructed based on the clustering parameter range, including:

[0018] Obtaining multiple initial neighborhood radii according to a neighborhood radius range; and obtaining multiple initial minimum neighborhood point counts according to a minimum neighborhood point count range;

[0019] Multiple groups of initial parameters are obtained by combining multiple initial neighborhood radii and multiple initial minimum neighborhood point numbers.

[0020] In one embodiment, clustering parameter optimization is performed based on the clustering processing results to obtain optimized parameters, including:

[0021] Obtaining the silhouette coefficient of each clustering processing result, and determining the fitness score between the fusion feature vector of each battery to be tested and the corresponding clustering processing result based on each silhouette coefficient and the number of clusters corresponding to each clustering processing result;

[0022] Determine the probability of each group of initial parameters being selected based on each fitness score;

[0023] Select multiple benchmark parameters from each group of initial parameters according to each probability; and perform weighted summation processing on each benchmark parameter to determine candidate parameters;

[0024] The candidate parameters are iteratively optimized to obtain the optimized parameters.

[0025] In one embodiment, iterative optimization is performed on the candidate parameters to obtain optimized parameters, including:

[0026] According to the initial pheromone concentration of the candidate parameters, the candidate parameters are optimized to obtain the initial optimized parameters;

[0027] When the number of optimizations is greater than or equal to the preset number of optimizations, the initial optimization parameters are determined as the optimized parameters; when the number of optimizations is less than the preset number of optimizations, the initial pheromone concentration is updated to obtain the updated pheromone concentration;

[0028] According to the updated pheromone concentration, the initial optimization parameters are optimized until the number of optimizations is greater than or equal to the preset number of optimizations, and the current optimization parameters are determined as the optimized parameters.

[0029] In one embodiment, the optimized parameters include an optimized neighborhood radius and an optimized minimum number of neighborhood points. Based on the optimized parameters and the fused feature vector of each battery to be detected, fault diagnosis is performed on each battery to be detected, and the fault diagnosis type of each battery to be detected is determined, including:

[0030] According to the preset fault judgment rules, based on the optimized neighborhood radius and the optimized minimum number of neighborhood points, the fused feature vectors of each battery to be detected are clustered to obtain the fault diagnosis type of each battery to be detected.

[0031] In a second aspect, the present application also provides a battery fault diagnosis device based on multi-feature fusion, comprising:

[0032] A feature fusion module is used to extract features from the electrochemical impedance spectroscopy data of each battery to be tested, obtain multi-dimensional electrochemical impedance spectroscopy features of each battery to be tested, and fuse the multi-dimensional electrochemical impedance spectroscopy features of each battery to be tested to determine a fusion feature vector of each battery to be tested;

[0033] A construction module is used to obtain the clustering parameter range of the fusion feature vector of each battery to be detected, and to construct multiple groups of initial parameters according to the clustering parameter range; each group of initial parameters includes different clustering parameters;

[0034] The optimization processing module is used to cluster the fused feature vectors of each battery to be detected according to each group of initial parameters, and optimize the clustering parameters according to the clustering processing results to obtain the optimized neighborhood radius and the optimized minimum number of neighborhood points;

[0035] The fault diagnosis module is used to perform fault diagnosis on each battery to be detected based on the optimized neighborhood radius, the optimized minimum number of neighborhood points and the fused feature vector of each battery to be detected, and determine the fault diagnosis type of each battery to be detected.

[0036] In a third aspect, the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the method of any embodiment of the first aspect are implemented.

[0037] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method of any embodiment of the first aspect above.

[0038] In a fifth aspect, the present application also provides a computer program product, comprising a computer program, which, when executed by a processor, implements the steps of the method of any embodiment of the first aspect above.

[0039] The battery fault diagnosis method, device, equipment and medium based on multi-feature fusion include: extracting features from the electrochemical impedance spectrum data of each battery to be detected to obtain the multi-dimensional electrochemical impedance spectrum features of each battery to be detected, fusing the multi-dimensional electrochemical impedance spectrum features of each battery to be detected to determine the fusion feature vector of each battery to be detected, obtaining the clustering parameter range of the fusion feature vector of each battery to be detected, and constructing multiple groups of initial parameters based on the clustering parameter range; clustering the fusion feature vector of each battery to be detected according to each group of initial parameters, and optimizing the clustering parameters according to each clustering processing result to obtain optimized parameters, and optimizing the fusion feature vector of each battery to be detected according to the optimized parameters and the fusion feature vector of each battery to be detected. Battery fault diagnosis is performed to determine the fault diagnosis type of each battery to be tested; the above method can extract multidimensional electrochemical impedance spectroscopy features based on the electrochemical impedance spectroscopy data of each battery to be tested, and fuse the multidimensional electrochemical impedance spectroscopy features to comprehensively reflect the multi-scale chemical state changes inside each battery to be tested. On this basis, battery fault diagnosis can be realized based on the multi-dimensional electrochemical impedance spectroscopy features of each battery to be tested, thereby improving the accuracy, practicality and reliability of battery fault diagnosis; at the same time, the above method can optimize the clustering parameters based on the preliminary clustering processing results to improve the adaptation of the optimized parameters to the current application scenario, thereby greatly improving the accuracy of battery fault diagnosis based on the optimized parameters and reducing the missed diagnosis rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 1 is a flow chart of a battery fault diagnosis method based on multi-feature fusion in one embodiment;

[0041] Figure 2 is a flowchart of a battery fault diagnosis method based on multi-feature fusion in another embodiment;

[0042] Figure 3 is a flowchart of a battery fault diagnosis method based on multi-feature fusion in another embodiment;

[0043] Figure 4 is a flowchart of a battery fault diagnosis method based on multi-feature fusion in another embodiment;

[0044] Figure 5 is a flowchart of a battery fault diagnosis method based on multi-feature fusion in another embodiment;

[0045] Figure 6 is a flowchart of a battery fault diagnosis method based on multi-feature fusion in another embodiment;

[0046] Figure 7 1 is a structural block diagram of a battery fault diagnosis device based on multi-feature fusion in one embodiment;

[0047] Figure 8 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0049] The battery fault diagnosis method based on multi-feature fusion provided in the embodiments of the present application can be applied to a battery fault diagnosis system. The battery fault diagnosis system includes multiple batteries to be tested and a computer device. The multiple batteries to be tested may include normal batteries and faulty batteries, and each battery to be tested may be, but is not limited to, a primary battery, a secondary battery, or a reserve battery. The computer device may be, but is not limited to, various personal computers, laptops, smartphones, tablet computers, Internet of Things devices, and portable wearable devices.

[0050] In an exemplary embodiment, Figure 1 As shown, a battery fault diagnosis method based on multi-feature fusion is provided. The method is applied to computer equipment as an example, and includes the following steps:

[0051] S100 , extracting features from the electrochemical impedance spectroscopy data of each battery to be tested to obtain a multi-dimensional electrochemical impedance spectroscopy feature of each battery to be tested, and fusing the multi-dimensional electrochemical impedance spectroscopy features of each battery to be tested to determine a fused feature vector of each battery to be tested.

[0052] In practical applications, an electrochemical workstation can be used to collect electrochemical impedance spectroscopy (EIS) data for each battery under test. Correspondingly, a computer device can retrieve pre-stored EIS data for each battery under test, or receive real-time EIS data from the electrochemical workstation.

[0053] Specifically, the computer device can extract the features of the electrochemical impedance spectroscopy data of each battery to be detected to obtain the multi-dimensional electrochemical impedance spectroscopy features of each battery to be detected, and fuse the multi-dimensional electrochemical impedance spectroscopy features of each battery to be detected to obtain the fusion feature vector of each battery to be detected.

[0054] In one implementation, a feature extraction method can be used to extract features from the electrochemical impedance spectroscopy data of each battery to be tested, thereby obtaining multi-dimensional electrochemical impedance spectroscopy features for each battery to be tested. Optionally, the feature extraction method can be based on a graph-based intuitive analysis method, an equivalent circuit fitting method, a mathematical transformation and statistical analysis method, a machine learning algorithm, or the like.

[0055] In another implementation, an algorithm model may be pre-trained, and then the electrochemical impedance spectroscopy data of each battery to be tested is input into the algorithm model, and the algorithm model outputs the multi-dimensional electrochemical impedance spectroscopy characteristics of each battery to be tested.

[0056] The multi-dimensional electrochemical impedance spectroscopy characteristics of the batteries to be tested may be integrated by simply splicing or combining the multi-dimensional electrochemical impedance spectroscopy characteristics of the batteries to be tested in any order.

[0057] In the embodiment of the present application, the multi-dimensional electrochemical impedance spectroscopy characteristics of the battery to be tested include the frequency domain characteristics of the battery to be tested, the equivalent circuit characteristics of the battery to be tested, and the statistical characteristics of the battery to be tested. It should be noted that the electrochemical impedance spectroscopy data of the battery to be tested can be understood as the electrochemical impedance spectroscopy curve of the battery to be tested. Specifically, the computer device can extract the real part of the key frequency points (such as 0.1Hz, 1Hz, 100Hz, 1000Hz, etc.) in the electrochemical impedance spectroscopy data of each battery to be tested. and the imaginary part , as the frequency domain characteristics of each battery to be tested.

[0058] At the same time, the computer equipment can use the Randles circuit model to extract the equivalent circuit parameters from the electrochemical impedance spectroscopy data of each battery to be tested 、 and , as the equivalent circuit characteristics of each battery to be tested.

[0059] In addition, the computer device can use the electrochemical impedance spectroscopy data of each battery to be tested to invert the DRT curve of each battery to be tested, and then determine the relaxation time distribution width of each battery to be tested based on the DRT curve of each battery to be tested. , and obtain the peak frequency corresponding to the electrochemical impedance spectroscopy data of each battery to be tested, and then calculate the Nyquist curvature of each battery to be tested according to the peak frequency corresponding to the electrochemical impedance spectroscopy data of each battery to be tested, and calculate the relaxation time distribution width of each battery to be tested and Nyquist curvature The statistical characteristics of each battery to be tested are determined.

[0060] S200: Obtain the clustering parameter range of the fused feature vector of each battery to be detected, and construct multiple groups of initial parameters based on the clustering parameter range, wherein each group of initial parameters includes different clustering parameters.

[0061] In practical applications, the fused feature vectors of each battery to be detected can be analyzed and processed to obtain the clustering parameter range of the fused feature vectors of each battery to be detected. Alternatively, the clustering parameter range of the fused feature vectors of each battery to be detected can be determined based on historical experience.

[0062] Optionally, the above-mentioned clustering parameter range may include a cluster number range and a distance metric range. Correspondingly, the computer device may construct multiple sets of initial parameters based on the cluster number range and the distance metric range. Optionally, the above-mentioned initial parameters may include clustering parameters, which may include the cluster number and the distance metric value, and the cluster number and distance metric value in each set of initial parameters are both within the cluster number range and the distance metric range.

[0063] S300 , clustering the fused feature vectors of each battery to be detected according to each group of initial parameters, and optimizing the clustering parameters according to each clustering result to obtain optimized parameters.

[0064] Specifically, the computer device may use a clustering algorithm to perform clustering processing on the fusion feature vectors of each battery to be detected for each group of initial parameters, and perform clustering parameter optimization processing according to each clustering processing result to obtain optimized parameters.

[0065] In the embodiment of the present application, the clustering algorithm corresponds to clustering parameters. The clustering algorithm can be a partition-based algorithm, a hierarchical clustering method, a model-based algorithm, etc. In the embodiment of the present application, the clustering algorithm can be a DBSCAN clustering method. At the same time, each set of initial parameters corresponds to a clustering processing result. Optionally, the clustering processing result can include the number of clusters and the category to which the fused feature vector of each battery to be tested belongs.

[0066] In practical applications, a computer device can pre-train a parameter optimization model, then input each clustering processing result into the parameter optimization model. The parameter optimization model then performs clustering parameter optimization based on each clustering processing result and inputs the optimized parameters. Optionally, the parameter optimization model can be at least one of a convolutional neural network model, a fully connected neural network model, a recurrent neural network model, a residual neural network model, etc. The optimized parameters are of the same type as the clustering parameters.

[0067] S400 , performing fault diagnosis on each battery to be detected based on the optimized parameters and the fused feature vector of each battery to be detected, and determining the fault diagnosis type of each battery to be detected.

[0068] Specifically, the computer device can cluster the fused feature vectors of each battery to be tested based on the optimized parameters to perform fault diagnosis on each battery to be tested, and determine the fault diagnosis type of each battery to be tested based on the clustering processing results. Optionally, the fault diagnosis type of the battery to be tested can be a normal type or a fault type, and the fault type can be electrochemical aging fault, manufacturing defect fault, abuse condition fault, progressive fault, sudden fault, intermittent fault, etc.

[0069] The technical solution in the embodiment of the present application is to extract features from the electrochemical impedance spectrum data of each battery to be detected to obtain the multi-dimensional electrochemical impedance spectrum features of each battery to be detected, and fuse the multi-dimensional electrochemical impedance spectrum features of each battery to be detected to determine the fused feature vector of each battery to be detected, obtain the clustering parameter range of the fused feature vector of each battery to be detected, and construct multiple groups of initial parameters according to the clustering parameter range; cluster the fused feature vector of each battery to be detected according to each group of initial parameters, and optimize the clustering parameters according to each clustering processing result to obtain optimized parameters, and perform fault diagnosis and confirmation on each battery to be detected according to the optimized parameters and the fused feature vector of each battery to be detected. Determine the fault diagnosis type of each battery to be tested; the above method can extract multidimensional electrochemical impedance spectroscopy features based on the electrochemical impedance spectroscopy data of each battery to be tested, and fuse the multidimensional electrochemical impedance spectroscopy features to comprehensively reflect the multi-scale chemical state changes inside each battery to be tested. On this basis, battery fault diagnosis can be realized based on the multi-dimensional electrochemical impedance spectroscopy features of each battery to be tested, thereby improving the accuracy, practicality and reliability of battery fault diagnosis; at the same time, the above method can optimize the clustering parameters based on the preliminary clustering processing results to improve the adaptation of the optimized parameters to the current application scenario, thereby greatly improving the accuracy of battery fault diagnosis based on the optimized parameters and reducing the missed diagnosis rate.

[0070] The following describes the process of fusing the multi-dimensional electrochemical impedance spectroscopy characteristics of each battery to be tested to determine the fused feature vector of each battery to be tested. Figure 2 As shown, the step of fusing the multi-dimensional electrochemical impedance spectroscopy characteristics of each battery to be tested to determine the fused feature vector of each battery to be tested in the above S100 may include the following steps:

[0071] S110 , for any battery to be detected, obtaining an initial characteristic vector of the battery to be detected according to the multi-dimensional electrochemical impedance spectroscopy characteristics of the battery to be detected.

[0072] In practical applications, for any battery to be tested, the multi-dimensional electrochemical impedance spectroscopy characteristics of the battery to be tested can be spliced ​​in a fixed splicing order to obtain the initial characteristic vector of the battery to be tested.

[0073] S120 , normalize the initial feature vector of the battery to be detected to obtain a normalized feature vector of the battery to be detected.

[0074] Furthermore, the computer device can use a normalization method to normalize the multi-dimensional electrochemical impedance spectroscopy features in the initial feature vector of the battery to be tested, thereby obtaining a normalized feature vector of the battery to be tested. Optionally, the normalization method can be Min-Max normalization, robust normalization, normalization, etc. In the embodiment of the present application, the normalization method can be Z-score normalization.

[0075] S130 , performing dimensionality reduction processing on the standardized feature vector of the battery to be detected to obtain a fused feature vector of the battery to be detected.

[0076] Specifically, the computer device can use a dimensionality reduction algorithm to reduce the dimensionality of the standardized feature vector of the battery to be tested to obtain a fused feature vector of the battery to be tested. Optionally, the dimensionality reduction algorithm can be a linear dimensionality reduction algorithm or a nonlinear dimensionality reduction algorithm. In the embodiment of the present application, the dimensionality reduction algorithm can be a t-SNE dimensionality reduction method.

[0077] The following describes the process of obtaining the initial feature vector of the battery to be tested based on the multi-dimensional electrochemical impedance spectroscopy characteristics of the battery to be tested. In one embodiment, the multi-dimensional electrochemical impedance spectroscopy characteristics include frequency domain characteristics, equivalent circuit characteristics and statistical characteristics; Figure 3 As shown, the steps in S110 above may include:

[0078] S111 , obtaining a first weight coefficient of a frequency domain characteristic of a battery to be detected, a second weight coefficient of an equivalent circuit characteristic of a battery to be detected, and a third weight coefficient of a statistical characteristic of a battery to be detected.

[0079] In practical applications, the first weight coefficient, the second weight coefficient and the third weight coefficient may be user-defined or determined based on historical experience, but the sum of the first weight coefficient, the second weight coefficient and the third weight coefficient is equal to 1.

[0080] S112 , performing weighted feature concatenation based on the frequency domain features, the equivalent circuit features, the statistical features, and the first weight coefficient, the second weight coefficient, and the third weight coefficient to obtain an initial feature vector of the battery to be detected.

[0081] In an embodiment of the present application, weighted feature concatenation may be performed on the frequency domain features, equivalent circuit features, statistical features, and the first weight coefficient, the second weight coefficient, and the third weight coefficient to obtain an initial feature vector of the battery to be detected.

[0082] Specifically, the process of weighted feature splicing can be understood as splicing the product of the frequency domain feature and the first weight coefficient, the product of the equivalent circuit feature and the second weight coefficient, and the product of the statistical feature and the third weight coefficient to obtain the initial feature vector of the battery to be detected. In the embodiment of the present application, the initial feature vector of the battery to be detected is It can be expressed by the following formula (1):

[0083] (1)

[0084] in, Represents the frequency domain characteristics of the battery to be tested, Represents the equivalent circuit characteristics of the battery to be tested, Indicates the statistical characteristics of the battery to be tested, represents the first weight coefficient, represents the second weight coefficient, represents the third weight coefficient; ; ; .

[0085] The technical solution in the embodiment of the present application is as follows: for any battery to be detected, based on the multi-dimensional electrochemical impedance spectrum characteristics of the battery to be detected, the initial feature vector of the battery to be detected is obtained, the initial feature vector of the battery to be detected is standardized to obtain the standardized feature vector of the battery to be detected, and the standardized feature vector of the battery to be detected is subjected to dimensionality reduction processing to obtain a fused feature vector of the battery to be detected; before the clustering processing, the above method can first perform standardization and dimensionality reduction processing on the initial feature vector of the battery to be detected, which can prepare for the subsequent improvement of the accuracy, efficiency and interpretability of the clustering processing.

[0086] The following describes the process of constructing multiple sets of initial parameters based on the clustering parameter range. In one embodiment, the clustering parameter range includes a neighborhood radius range and a minimum neighborhood point range; Figure 4 As shown, the step of constructing multiple groups of initial parameters according to the clustering parameter range in S200 can be implemented in the following ways:

[0087] S210 , obtaining a plurality of initial neighborhood radii according to a neighborhood radius range; and obtaining a plurality of initial minimum neighborhood points according to a minimum neighborhood point range.

[0088] Specifically, the computer device can be used in the neighborhood radius range according to the interval method or random method. Select multiple different neighborhood radii between the maximum neighborhood radius and the minimum neighborhood radius as multiple initial neighborhood radii, and select the initial neighborhood radii according to the interval method or random method within the minimum neighborhood point range. Select multiple different minimum neighborhood points between the largest minimum neighborhood point number and the smallest minimum neighborhood point number as multiple initial minimum neighborhood points.

[0089] In the embodiment of the present application, the number of selected initial neighborhood radii and the number of initial minimum neighborhood points may be equal or unequal; wherein the selected nth initial neighborhood radius and the nth initial minimum neighborhood point number may be respectively expressed as:

[0090] (2)

[0091] (3)

[0092] S220 , performing combination processing according to multiple initial neighborhood radii and multiple initial minimum neighborhood point numbers to obtain multiple groups of initial parameters.

[0093] In one embodiment, any initial neighborhood radius can be selected from multiple initial neighborhood radii, and any initial minimum neighborhood point number can be selected from multiple initial minimum neighborhood point numbers, and the two can be combined to form a set of initial parameters, and so on, to construct multiple sets of different initial parameters.

[0094] The initial neighborhood radius in each set of initial parameters may be the same or different, and the initial minimum number of neighborhood points in each set of initial parameters may be the same or different. For example, the initial parameter formed by the combination of the nth initial neighborhood radius and the nth initial minimum number of neighborhood points can be expressed as .

[0095] The technical solution in the embodiment of the present application obtains multiple initial neighborhood radii according to the neighborhood radius range; and obtains multiple initial minimum neighborhood points according to the minimum neighborhood point range, and combines the multiple initial neighborhood radii and the multiple initial minimum neighborhood points to obtain multiple groups of initial parameters; the above method does not require a complicated processing process, thereby speeding up the construction of the initial parameters, improving the efficiency of the construction of the initial parameters, and reducing the complexity of constructing the initial parameters.

[0096] The following describes the process of performing clustering parameter optimization based on the clustering processing results to obtain optimized parameters. Figure 5 As shown, the step of performing clustering parameter optimization processing according to each clustering processing result to obtain optimized parameters in the above S300 can be implemented in the following manner:

[0097] S310 , obtaining the silhouette coefficient of each clustering processing result, and determining the fitness score between the fused feature vector of each battery to be detected and the corresponding clustering processing result according to each silhouette coefficient and the number of clusters corresponding to each clustering processing result.

[0098] In practical applications, for any clustering result, the computer device can calculate the number of samples i to cluster C k The average distance of all other samples in the cluster (i.e., the distance within the cluster) a(i), and the distance between sample i and all other clusters C j The minimum average distance (i.e., the shortest distance between clusters) b(i) of the clusters (j≠k) is obtained by subtracting b(i) from a(i) and then dividing it by the maximum value between b(i) and a(i) to obtain the silhouette coefficient s(i) of sample i in the clustering processing result. The silhouette coefficient s(i) can be expressed by the following formula (4). It should be noted that the above sample i can be any one of the fused feature vectors of each battery to be tested.

[0099] (4)

[0100] Furthermore, for any silhouette coefficient, an arithmetic operation can be performed based on the silhouette coefficient and the number of clusters in the corresponding clustering processing result to obtain a fitness score between the fused feature vector of the battery to be tested and the clustering processing result. Optionally, the arithmetic operation can be at least one of addition, subtraction, multiplication, division, logarithm, exponential operation, etc.

[0101] In the embodiment of the present application, any of the above batteries to be tested The fitness score between the fusion feature vector and the corresponding clustering processing result This can be achieved through formula (5):

[0102] (5)

[0103] in, represents the actual number of clusters obtained after clustering processing, Indicates the expected number of clusters after clustering processing, Indicates the number of clusters penalty.

[0104] S320: Determine the probability of each group of initial parameters being selected according to each fitness score.

[0105] Specifically, the computer device may perform analysis processing, arithmetic operation processing and / or comparison processing on each fitness score to obtain the probability of each group of initial parameters being selected.

[0106] In an embodiment of the present application, the computer device may use a selection operation in a genetic algorithm (ie, a GA algorithm) to calculate the probability of each group of initial parameters being selected according to each fitness score.

[0107] It should be noted here that for any set of initial parameters, the fitness score corresponding to the set of initial parameters can be obtained by summing the fitness scores between the fusion feature vectors of each battery to be tested and the corresponding clustering processing results. , and then the fitness score corresponding to the initial parameters can be Fitness scores corresponding to all groups of initial parameters The probability of the initial parameters being selected is obtained by taking the sum as the quotient Among them, the probability of any set of initial parameters being selected is It can be expressed by the following formula (6):

[0108] (6)

[0109] S330 , selecting multiple benchmark parameters from each group of initial parameters according to each probability; and performing weighted summation processing on each benchmark parameter to determine candidate parameters.

[0110] In practical applications, computer equipment can compare or extreme value each probability, and select multiple benchmark parameters with larger probabilities from each group of initial parameters based on the processing results. Furthermore, each benchmark parameter can be weighted and summed to obtain candidate parameters.

[0111] In the embodiment of the present application, the computer device can use the crossover operation in the genetic algorithm to perform weighted summation processing based on each benchmark parameter to obtain the candidate parameter. For example, if the multiple benchmark parameters include two benchmark parameters, namely benchmark parameters 1 and 2, 1 , 2 , correspondingly, the following formulas (7) and (8) can be used to calculate the reference parameters 1 and 2 Perform weighted summation, that is

[0112] = + (7)

[0113] (8)

[0114] in, , Represents a random weight coefficient, and the candidate parameters can be expressed as Candidate neighborhood radius, Indicates the minimum number of candidate neighborhood points.

[0115] S340: Perform iterative optimization processing on the candidate parameters to obtain optimized parameters.

[0116] In one embodiment, the computer device may pre-train an iterative optimization model, then input candidate parameters into the iterative optimization model. The iterative optimization model then iteratively optimizes the candidate parameters and outputs optimized parameters. Optionally, the iterative optimization model may be at least one of a convolutional neural network model, a fully connected neural network model, a recurrent neural network model, a residual neural network model, a long short-term memory neural network model, and the like.

[0117] The technical solution in the embodiment of the present application obtains the silhouette coefficient of each clustering processing result, and determines the fitness score between the fusion feature vector of each battery to be tested and the corresponding clustering processing result based on each silhouette coefficient and the number of clusters corresponding to each clustering processing result, determines the probability of each group of initial parameters being selected based on each fitness score, and selects multiple benchmark parameters from each group of initial parameters based on each probability; and determines the candidate parameters by weighted summation processing based on each benchmark parameter, and iteratively optimizes the candidate parameters to obtain optimized parameters; the above method can optimize the clustering parameters based on the clustering processing results obtained by the preliminary clustering processing, so as to provide dependent information for the subsequent accurate battery fault diagnosis, and the processing process is simple and does not require the participation of complex algorithms, thereby reducing the complexity of the clustering parameter optimization processing and improving the speed of the clustering parameter optimization processing; at the same time, the above method does not require manual participation when optimizing the clustering parameters, thereby accelerating the optimization speed and optimization efficiency of the clustering parameters.

[0118] The following describes the process of iteratively optimizing the candidate parameters to obtain the optimized parameters. Figure 6 As shown, the steps in S340 above can be implemented in the following ways:

[0119] S341. Optimize the candidate parameters according to their initial pheromone concentrations to obtain initial optimized parameters.

[0120] The computer device may adjust the candidate parameters using the initial pheromone concentrations of the candidate parameters to achieve optimization processing of the candidate parameters and obtain corresponding initial optimization parameters.

[0121] In the embodiment of the present application, the computer device can use the mutation operation introduced in the ant colony algorithm (i.e., the ACO algorithm) to apply the initial pheromone concentration weighted perturbation to the candidate parameters to optimize the candidate parameters and obtain the corresponding initial optimized parameters. It should be noted that the implementation process of applying the initial pheromone concentration weighted perturbation to the candidate parameters can be expressed by formulas (9) and (10):

[0122] (9)

[0123] (10)

[0124] in, represents the learning rate (used to control the perturbation amplitude). The initial pheromone concentration of the above candidate parameters may include the initial pheromone concentration corresponding to the candidate neighborhood radius. The initial pheromone concentration corresponding to the candidate minimum number of neighborhood points At the same time, the above initial optimization parameters can be expressed as .

[0125] S342. When the number of optimizations is greater than or equal to the preset number of optimizations, the initial optimization parameters are determined as the optimized parameters; when the number of optimizations is less than the preset number of optimizations, the initial pheromone concentration is updated to obtain the updated pheromone concentration.

[0126] Specifically, when the current optimization times are less than the preset optimization times, the initial pheromone concentration may be updated to obtain an updated pheromone concentration.

[0127] In one embodiment, the initial pheromone concentration may be updated by adding or subtracting the initial pheromone concentration from a preset adjustment amount to obtain an updated pheromone concentration. In this embodiment of the present application, the initial pheromone concentration may be updated by the following formulas (11) and (12):

[0128] (11)

[0129] (12)

[0130] in, Indicates the current number of optimizations. Indicates the number of optimizations for the next time. represents the increment of the initial pheromone concentration of the neighborhood radius, represents the increment of the initial pheromone concentration corresponding to the minimum number of neighborhood points, represents the volatilization rate of the initial pheromone concentration.

[0131] S343. Optimize the initial optimization parameters according to the updated pheromone concentration until the number of optimizations is greater than or equal to the preset number of optimizations, and determine the current optimization parameters as the optimized parameters.

[0132] It should be noted that, during the iterative optimization process, each optimization process is to optimize the parameters after the previous optimization process again using the latest updated pheromone concentration.

[0133] The technical solution in the embodiment of the present application optimizes the candidate parameters according to their initial pheromone concentrations to obtain initial optimized parameters. When the number of optimizations is greater than or equal to a preset number of optimizations, the initial optimized parameters are determined as optimized parameters. When the number of optimizations is less than the preset number of optimizations, the initial pheromone concentration is updated to obtain an updated pheromone concentration. The initial optimized parameters are optimized according to the updated pheromone concentration until the number of optimizations is greater than or equal to the preset number of optimizations, and the current optimized parameters are determined as optimized parameters. The above method can obtain the optimal clustering parameters through iterative optimization, so as to prepare for the subsequent improvement of the accuracy of battery fault diagnosis results.

[0134] In one embodiment, the optimized parameters include an optimized neighborhood radius and an optimized minimum number of neighborhood points. The step of performing fault diagnosis on each battery to be detected based on the optimized parameters and the fused feature vector of each battery to be detected in S400 to determine the fault diagnosis type of each battery to be detected may include: clustering the fused feature vector of each battery to be detected based on the optimized neighborhood radius and the optimized minimum number of neighborhood points in accordance with a preset fault judgment rule to obtain the fault diagnosis type of each battery to be detected.

[0135] In an embodiment of the present application, the computer device can use the DBSCAN clustering method to cluster the fused feature vectors of each battery to be detected according to the preset fault judgment rules based on the optimized neighborhood radius and the optimized minimum number of neighborhood points to obtain the fault diagnosis type of each battery to be detected.

[0136] It should be noted here that the above fault determination rule may be an association rule formed by combining Rule 1 and Rule 2, or an association rule formed by combining Rule 3 and Rule 2.

[0137] Among them, Rule 1 may include calculating the mean of the frequency domain characteristics, the mean of the equivalent circuit characteristics, and the mean of the statistical characteristics of all batteries to be tested in each cluster, and for any cluster, when the mean of the frequency domain characteristics of all batteries to be tested is less than -2.0 and the mean of the equivalent circuit characteristics of all batteries to be tested is less than -1.0, it is determined that the cluster is an internal short circuit fault type; when the mean of the frequency domain characteristics of all batteries to be tested is less than -3.0 and the mean of the statistical characteristics of all batteries to be tested is less than -2.0, it is determined that the cluster is a lithium plating fault type; when the mean of the statistical characteristics of all batteries to be tested is greater than 3.0 and the mean of the equivalent circuit characteristics of all batteries to be tested is greater than 2.0, it is determined that the cluster is an aging fault type; Rule 2 may determine that the battery to be tested that cannot be included in any cluster is a new type of fault battery or transient interference based on the fused feature vector of the battery to be tested.

[0138] At the same time, the above rule 3 may include calculating the mean of the frequency domain characteristics, the mean of the equivalent path characteristics, and the mean of the statistical characteristics of all batteries to be detected in each cluster, and then matching the preset frequency domain characteristic thresholds, preset equivalent path characteristic thresholds, and preset statistical characteristic thresholds corresponding to different fault types with the mean of the frequency domain characteristics, the mean of the equivalent path characteristics, and the mean of the statistical characteristics of all batteries to be detected in each cluster to obtain the fault type corresponding to each cluster.

[0139] The technical solution in the embodiment of the present application can cluster the fused feature vectors of each battery to be detected according to the preset fault judgment rules, based on the optimized neighborhood radius and the optimized minimum number of neighborhood points, to obtain the fault diagnosis type of each battery to be detected. This process can accurately locate the faulty battery and improve the reliability of the battery fault diagnosis results.

[0140] In one embodiment, the present application also provides a battery fault diagnosis method based on multi-feature fusion, which is applied to a computer device. The method includes the following process:

[0141] (1) Extracting features from the electrochemical impedance spectroscopy data of each battery to be tested to obtain multidimensional electrochemical impedance spectroscopy features of each battery to be tested; the multidimensional electrochemical impedance spectroscopy features include frequency domain features, equivalent circuit features, and statistical features;

[0142] (2) For any battery to be detected, obtain a first weight coefficient of the frequency domain characteristics of the battery to be detected, a second weight coefficient of the equivalent circuit characteristics of the battery to be detected, and a third weight coefficient of the statistical characteristics of the battery to be detected;

[0143] (3) Perform weighted feature concatenation based on frequency domain features, equivalent circuit features, statistical features, and the first weight coefficient, the second weight coefficient, and the third weight coefficient to obtain an initial feature vector of the battery to be tested;

[0144] (4) Standardizing the initial feature vector of the battery to be tested to obtain a standardized feature vector of the battery to be tested;

[0145] (5) Perform dimensionality reduction processing on the standardized feature vector of the battery to be tested to obtain the fused feature vector of the battery to be tested;

[0146] (6) Obtain the clustering parameter range of the fusion feature vector of each battery to be tested; the clustering parameter range includes the neighborhood radius range and the minimum neighborhood point number range;

[0147] (7) obtaining a plurality of initial neighborhood radii according to the neighborhood radius range; and obtaining a plurality of initial minimum neighborhood points according to the minimum neighborhood point range;

[0148] (8) Combining multiple initial neighborhood radii and multiple initial minimum neighborhood point numbers to obtain multiple sets of initial parameters;

[0149] (9) Clustering the fusion feature vectors of each battery to be tested according to each group of initial parameters;

[0150] (10) Obtaining the silhouette coefficient of each clustering processing result, and determining the fitness score between the fusion feature vector of each battery to be tested and the corresponding clustering processing result based on each silhouette coefficient and the number of clusters corresponding to each clustering processing result;

[0151] (11) Determine the probability of each group of initial parameters being selected based on each fitness score;

[0152] (12) Select multiple benchmark parameters from each group of initial parameters according to each probability; and perform weighted summation processing on each benchmark parameter to determine the candidate parameters;

[0153] (13) Optimizing the candidate parameters according to their initial pheromone concentrations to obtain initial optimized parameters;

[0154] (14) When the number of optimizations is greater than or equal to the preset number of optimizations, the initial optimization parameters are determined as the optimized parameters; when the number of optimizations is less than the preset number of optimizations, the initial pheromone concentration is updated to obtain the updated pheromone concentration;

[0155] (15) According to the updated pheromone concentration, the initial optimization parameters are optimized until the number of optimizations is greater than or equal to the preset number of optimizations, and the current optimization parameters are determined as the optimized parameters; the optimized parameters include the optimized neighborhood radius and the optimized minimum number of neighborhood points;

[0156] (16) According to the preset fault judgment rules, based on the optimized neighborhood radius and the optimized minimum number of neighborhood points, the fused feature vectors of each battery to be tested are clustered to obtain the fault diagnosis type of each battery to be tested.

[0157] The execution process of the above (1) to (16) can be specifically referred to the description of the above embodiment. The implementation principles and technical effects are similar and will not be repeated here.

[0158] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0159] Based on the same inventive concept, the embodiments of the present application also provide a battery fault diagnosis device based on multi-feature fusion for implementing the battery fault diagnosis method based on multi-feature fusion mentioned above. The implementation solution provided by this device is similar to the implementation solution described in the above method. Therefore, the specific limitations of one or more embodiments of the battery fault diagnosis device based on multi-feature fusion provided below can be found in the limitations of the battery fault diagnosis method based on multi-feature fusion above, and will not be repeated here.

[0160] In an exemplary embodiment, Figure 7 As shown, a battery fault diagnosis device based on multi-feature fusion is provided, comprising: a feature fusion module 11, a construction module 12, an optimization processing module 13 and a fault diagnosis module 14, wherein:

[0161] A feature fusion module 11 is used to extract features from the electrochemical impedance spectroscopy data of each battery to be tested, obtain multi-dimensional electrochemical impedance spectroscopy features of each battery to be tested, and fuse the multi-dimensional electrochemical impedance spectroscopy features of each battery to be tested to determine a fused feature vector for each battery to be tested;

[0162] A construction module 12 is used to obtain the clustering parameter range of the fusion feature vector of each battery to be detected, and construct multiple groups of initial parameters according to the clustering parameter range; each group of initial parameters includes different clustering parameters;

[0163] The optimization processing module 13 is used to cluster the fused feature vectors of each battery to be detected according to each set of initial parameters, and optimize the clustering parameters according to the clustering processing results to obtain the optimized neighborhood radius and the optimized minimum number of neighborhood points;

[0164] The fault diagnosis module 14 is configured to perform fault diagnosis on each battery to be detected based on the optimized neighborhood radius, the optimized minimum number of neighborhood points, and the fused feature vector of each battery to be detected, and determine the fault diagnosis type of each battery to be detected.

[0165] The battery fault diagnosis device based on multi-feature fusion provided in the embodiment of the present application can be used to execute the technical solution in the above-mentioned battery fault diagnosis method embodiment based on multi-feature fusion of the present application. Its implementation principle and technical effects are similar and will not be repeated here.

[0166] In one embodiment, the feature fusion module 11 includes: a first acquisition unit, a normalization processing unit, and a dimensionality reduction processing unit, wherein:

[0167] A first acquisition unit is configured to acquire, for any battery to be detected, an initial feature vector of the battery to be detected based on the multi-dimensional electrochemical impedance spectroscopy characteristics of the battery to be detected;

[0168] A standardization processing unit, used to perform standardization processing on the initial feature vector of the battery to be tested to obtain a standardized feature vector of the battery to be tested;

[0169] The dimensionality reduction processing unit is used to perform dimensionality reduction processing on the standardized feature vector of the battery to be detected to obtain a fused feature vector of the battery to be detected.

[0170] The battery fault diagnosis device based on multi-feature fusion provided in the embodiment of the present application can be used to execute the technical solution in the above-mentioned battery fault diagnosis method embodiment based on multi-feature fusion of the present application. Its implementation principle and technical effects are similar and will not be repeated here.

[0171] In one embodiment, the multidimensional electrochemical impedance spectroscopy features include frequency domain features, equivalent circuit features, and statistical features; the first acquisition unit is specifically configured to:

[0172] Obtaining a first weight coefficient of the frequency domain characteristics of the battery to be detected, a second weight coefficient of the equivalent circuit characteristics of the battery to be detected, and a third weight coefficient of the statistical characteristics of the battery to be detected;

[0173] Weighted feature concatenation is performed according to the frequency domain features, the equivalent circuit features, the statistical features, and the first weight coefficient, the second weight coefficient, and the third weight coefficient to obtain an initial feature vector of the battery to be detected.

[0174] The battery fault diagnosis device based on multi-feature fusion provided in the embodiment of the present application can be used to execute the technical solution in the above-mentioned battery fault diagnosis method embodiment based on multi-feature fusion of the present application. Its implementation principle and technical effects are similar and will not be repeated here.

[0175] In one embodiment, the clustering parameter range includes a neighborhood radius range and a minimum neighborhood point range; the construction module 12 includes: a second acquisition unit and a combination processing unit, wherein:

[0176] A second acquiring unit is configured to acquire a plurality of initial neighborhood radii according to a neighborhood radius range; and acquire a plurality of initial minimum neighborhood point numbers according to a minimum neighborhood point number range;

[0177] The combination processing unit is used to perform combination processing according to multiple initial neighborhood radii and multiple initial minimum neighborhood point numbers to obtain multiple groups of initial parameters.

[0178] The battery fault diagnosis device based on multi-feature fusion provided in the embodiment of the present application can be used to execute the technical solution in the above-mentioned battery fault diagnosis method embodiment based on multi-feature fusion of the present application. Its implementation principle and technical effects are similar and will not be repeated here.

[0179] In one embodiment, the optimization processing module 13 includes: a first determination unit, a second determination unit, a weighted summation processing unit, and an iterative optimization processing unit, wherein:

[0180] A first determining unit is used to obtain a silhouette coefficient of each clustering processing result, and determine a fitness score between the fusion feature vector of each battery to be detected and the corresponding clustering processing result according to each silhouette coefficient and the number of clusters corresponding to each clustering processing result;

[0181] A second determining unit is used to determine the probability of each group of initial parameters being selected according to each fitness score;

[0182] A weighted summation processing unit is used to select multiple reference parameters from each group of initial parameters according to each probability; and perform weighted summation processing on each reference parameter to determine candidate parameters;

[0183] The iterative optimization processing unit is used to perform iterative optimization processing on the candidate parameters to obtain optimized parameters.

[0184] The battery fault diagnosis device based on multi-feature fusion provided in the embodiment of the present application can be used to execute the technical solution in the above-mentioned battery fault diagnosis method embodiment based on multi-feature fusion of the present application. Its implementation principle and technical effects are similar and will not be repeated here.

[0185] In one embodiment, the iterative optimization processing unit is specifically configured to:

[0186] According to the initial pheromone concentration of the candidate parameters, the candidate parameters are optimized to obtain the initial optimized parameters;

[0187] When the number of optimizations is greater than or equal to the preset number of optimizations, the initial optimization parameters are determined as the optimized parameters; when the number of optimizations is less than the preset number of optimizations, the initial pheromone concentration is updated to obtain the updated pheromone concentration;

[0188] According to the updated pheromone concentration, the initial optimization parameters are optimized until the number of optimizations is greater than or equal to the preset number of optimizations, and the current optimization parameters are determined as the optimized parameters.

[0189] The battery fault diagnosis device based on multi-feature fusion provided in the embodiment of the present application can be used to execute the technical solution in the above-mentioned battery fault diagnosis method embodiment based on multi-feature fusion of the present application. Its implementation principle and technical effects are similar and will not be repeated here.

[0190] In one embodiment, the optimized parameters include the optimized neighborhood radius and the optimized minimum number of neighborhood points; the fault diagnosis module 14 is specifically configured to:

[0191] According to the preset fault judgment rules, based on the optimized neighborhood radius and the optimized minimum number of neighborhood points, the fused feature vectors of each battery to be detected are clustered to obtain the fault diagnosis type of each battery to be detected.

[0192] The battery fault diagnosis device based on multi-feature fusion provided in the embodiment of the present application can be used to execute the technical solution in the above-mentioned battery fault diagnosis method embodiment based on multi-feature fusion of the present application. Its implementation principle and technical effects are similar and will not be repeated here.

[0193] Each module in the multi-feature fusion-based battery fault diagnosis device can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a computer device memory in the form of software, so that the processor can call and execute the corresponding operations of each module.

[0194] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 8As shown. The computer device includes a processor, memory, an input / output interface, a communication interface, a display unit, and an input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals via wired or wireless means. The wireless means can be implemented via Wi-Fi, mobile cellular networks, near-field communication (NFC), or other technologies. When executed by the processor, the computer program implements a battery fault diagnosis method based on multi-feature fusion. The display unit of the computer device is used to produce a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device casing, or an external keyboard, touchpad or mouse.

[0195] Those skilled in the art will understand that Figure 8 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0196] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0197] Extracting features from the electrochemical impedance spectroscopy data of each battery to be tested to obtain multi-dimensional electrochemical impedance spectroscopy features of each battery to be tested, and fusing the multi-dimensional electrochemical impedance spectroscopy features of each battery to be tested to determine a fused feature vector of each battery to be tested;

[0198] Obtaining the clustering parameter range of the fused feature vector of each battery to be detected, and constructing multiple groups of initial parameters based on the clustering parameter range; each group of initial parameters includes different clustering parameters;

[0199] Clustering is performed on the fused feature vectors of each battery to be tested according to each group of initial parameters, and clustering parameter optimization is performed based on the clustering processing results to obtain optimized parameters;

[0200] According to the optimized parameters and the fused feature vectors of each battery to be detected, fault diagnosis is performed on each battery to be detected, and the fault diagnosis type of each battery to be detected is determined.

[0201] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0202] Extracting features from the electrochemical impedance spectroscopy data of each battery to be tested to obtain multi-dimensional electrochemical impedance spectroscopy features of each battery to be tested, and fusing the multi-dimensional electrochemical impedance spectroscopy features of each battery to be tested to determine a fused feature vector of each battery to be tested;

[0203] Obtaining the clustering parameter range of the fused feature vector of each battery to be detected, and constructing multiple groups of initial parameters based on the clustering parameter range; each group of initial parameters includes different clustering parameters;

[0204] Clustering is performed on the fused feature vectors of each battery to be tested according to each group of initial parameters, and clustering parameter optimization is performed based on the clustering processing results to obtain optimized parameters;

[0205] According to the optimized parameters and the fused feature vectors of each battery to be detected, fault diagnosis is performed on each battery to be detected, and the fault diagnosis type of each battery to be detected is determined.

[0206] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0207] Extracting features from the electrochemical impedance spectroscopy data of each battery to be tested to obtain multi-dimensional electrochemical impedance spectroscopy features of each battery to be tested, and fusing the multi-dimensional electrochemical impedance spectroscopy features of each battery to be tested to determine a fused feature vector of each battery to be tested;

[0208] Obtaining the clustering parameter range of the fused feature vector of each battery to be detected, and constructing multiple groups of initial parameters based on the clustering parameter range; each group of initial parameters includes different clustering parameters;

[0209] Clustering is performed on the fused feature vectors of each battery to be tested according to each group of initial parameters, and clustering parameter optimization is performed based on the clustering processing results to obtain optimized parameters;

[0210] According to the optimized parameters and the fused feature vectors of each battery to be detected, fault diagnosis is performed on each battery to be detected, and the fault diagnosis type of each battery to be detected is determined.

[0211] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments. In particular, any reference to a memory, database, or other medium used in the embodiments provided in this application can include at least one of a non-volatile memory and a volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), data processing logic based on quantum computing, artificial intelligence (AI) processors, and the like.

[0212] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0213] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A battery fault diagnosis method based on multi-feature fusion, characterized in that: The method comprises: Extracting features from the electrochemical impedance spectroscopy data of each battery to be tested to obtain multidimensional electrochemical impedance spectroscopy features of each battery to be tested, and fusing the multidimensional electrochemical impedance spectroscopy features of each battery to be tested to determine a fused feature vector for each battery to be tested; Obtaining a clustering parameter range of the fused feature vector of each battery to be detected, and constructing multiple groups of initial parameters according to the clustering parameter range; each group of initial parameters includes different clustering parameters; Clustering the fused feature vectors of the batteries to be tested according to the initial parameters of each group, and optimizing the clustering parameters according to the clustering results to obtain optimized parameters; According to the optimized parameters and the fused feature vector of each battery to be detected, fault diagnosis is performed on each battery to be detected to determine the fault diagnosis type of each battery to be detected.

2. The method according to claim 1, characterized in that The step of fusing the multi-dimensional electrochemical impedance spectroscopy characteristics of each battery to be detected to determine a fusion feature vector of each battery to be detected includes: For any battery to be tested, obtaining an initial feature vector of the battery to be tested according to the multi-dimensional electrochemical impedance spectroscopy characteristics of the battery to be tested; Normalizing the initial feature vector of the battery to be detected to obtain a standardized feature vector of the battery to be detected; Performing dimensionality reduction processing on the standardized feature vector of the battery to be detected to obtain a fused feature vector of the battery to be detected.

3. The method according to claim 2, characterized in that The multidimensional electrochemical impedance spectroscopy characteristics include frequency domain characteristics, equivalent circuit characteristics, and statistical characteristics; and obtaining the initial characteristic vector of the battery to be detected based on the multidimensional electrochemical impedance spectroscopy characteristics of the battery to be detected includes: Obtaining a first weight coefficient of the frequency domain characteristics of the battery to be detected, a second weight coefficient of the equivalent circuit characteristics of the battery to be detected, and a third weight coefficient of the statistical characteristics of the battery to be detected; Weighted feature concatenation is performed according to the frequency domain features, the equivalent circuit features, the statistical features, and the first weight coefficient, the second weight coefficient, and the third weight coefficient to obtain an initial feature vector of the battery to be detected.

4. The method according to any one of claims 1 to 3, characterized in that The clustering parameter range includes a neighborhood radius range and a minimum neighborhood point range; and constructing multiple sets of initial parameters based on the clustering parameter range includes: Acquire multiple initial neighborhood radii according to the neighborhood radius range; and acquire multiple initial minimum neighborhood points according to the minimum neighborhood point range; The multiple groups of initial parameters are obtained by combining the multiple initial neighborhood radii and the multiple initial minimum neighborhood point numbers.

5. The method according to any one of claims 1 to 3, characterized in that The clustering parameter optimization process is performed according to the clustering process results to obtain the optimized parameters, including: Obtaining a silhouette coefficient of each clustering processing result, and determining a fitness score between the fused feature vector of each battery to be tested and the corresponding clustering processing result according to each silhouette coefficient and the number of clusters corresponding to each clustering processing result; Determining the probability of each group of initial parameters being selected according to each fitness score; Selecting multiple reference parameters from each group of the initial parameters according to each of the probabilities; and performing weighted summation processing on each of the reference parameters to determine candidate parameters; An iterative optimization process is performed on the candidate parameters to obtain optimized parameters.

6. The method according to claim 5, characterized in that The iterative optimization process of the candidate parameters to obtain optimized parameters includes: Optimizing the candidate parameters according to the initial pheromone concentrations of the candidate parameters to obtain initial optimized parameters; When the number of optimizations is greater than or equal to the preset number of optimizations, the initial optimization parameters are determined as the optimized parameters; when the number of optimizations is less than the preset number of optimizations, the initial pheromone concentration is updated to obtain an updated pheromone concentration; The initial optimization parameters are optimized according to the updated pheromone concentration until the number of optimizations is greater than or equal to the preset number of optimizations, and the current optimization parameters are determined as the optimized parameters.

7. The method according to any one of claims 1 to 3, characterized in that The optimized parameters include an optimized neighborhood radius and an optimized minimum number of neighborhood points; and performing fault diagnosis on each battery to be detected based on the optimized parameters and the fused feature vector of each battery to be detected, and determining the fault diagnosis type of each battery to be detected, including: According to preset fault judgment rules, based on the optimized neighborhood radius and the optimized minimum number of neighborhood points, clustering processing is performed on the fused feature vectors of each battery to be detected to obtain a fault diagnosis type of each battery to be detected.

8. A battery fault diagnosis device based on multi-feature fusion, characterized in that: The device comprises: a feature fusion module for extracting features from the electrochemical impedance spectroscopy data of each battery to be detected, obtaining multidimensional electrochemical impedance spectroscopy features of each battery to be detected, and fusing the multidimensional electrochemical impedance spectroscopy features of each battery to be detected to determine a fused feature vector for each battery to be detected; A construction module, configured to obtain a clustering parameter range of the fused feature vector of each battery to be detected, and construct multiple groups of initial parameters according to the clustering parameter range; each group of initial parameters includes different clustering parameters; An optimization processing module, configured to cluster the fused feature vectors of the batteries to be detected according to the initial parameters of each group, and optimize the clustering parameters according to the clustering processing results to obtain an optimized neighborhood radius and an optimized minimum number of neighborhood points; A fault diagnosis module is used to perform fault diagnosis on each battery to be detected based on the optimized neighborhood radius, the optimized minimum number of neighborhood points and the fused feature vector of each battery to be detected, and determine the fault diagnosis type of each battery to be detected.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images