Vehicle battery fault diagnosis method, device, equipment and storage medium

Abstract

The application discloses a vehicle battery fault diagnosis method and device, equipment and storage medium, relates to the new energy battery monitoring technical field, and the vehicle battery fault diagnosis method comprises: carrying out data expansion on battery associated data to determine battery expansion data; a target entropy value model, a battery temperature model and a battery fluctuation model are respectively established according to the battery expansion data; according to the target entropy value model, the battery temperature model and the battery fluctuation model, feature screening and fusion are carried out, and battery key features are determined;A target battery diagnosis model is established according to the battery key features;The battery data to be diagnosed is input into the target battery diagnosis model, and the battery diagnosis result of the target vehicle is obtained. By the method of multi-index fusion, the abnormal condition of the battery is accurately identified, measures are taken in advance to maintain and repair, so as to prolong the service life of the battery, improve the reliability and safety of the new energy vehicle.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of new energy battery monitoring, in particular to a vehicle battery fault diagnosis method, device, equipment and storage medium. BACKGROUND

[0002] With the vigorous development of the new energy vehicle industry, the safety and reliability of power batteries have become the focus of public attention. Battery fault diagnosis and risk assessment are crucial for ensuring the safe operation of vehicles, but existing technologies focus on specific faults or single abnormality detection, lack overall risk assessment, and are usually only applicable to specific working conditions or vehicle models. In addition, these technologies often start from the algorithm design perspective without a deep understanding of the nature of battery faults, resulting in low accuracy in identifying risk vehicles.

[0003] In the face of the reality that battery performance is influenced by multiple factors, single parameter evaluation is obviously insufficient. Actual faults usually involve multiple characteristics, so an overall diagnosis and risk assessment system is particularly necessary. Therefore, how to solve the problem of the lack of overall risk assessment in existing new energy battery monitoring technology has become an urgent problem to be solved. SUMMARY

[0004] The main purpose of the present application is to provide a vehicle battery fault diagnosis method, device, equipment and storage medium, which aims to solve the technical problem of the lack of overall risk assessment in existing new energy battery monitoring technology.

[0005] To achieve the above-mentioned purpose, the present application provides a vehicle battery fault diagnosis method, which comprises:

[0006] determining battery expansion data by data expansion of battery correlation data;

[0007] establishing a target entropy value model, a battery temperature model and a battery fluctuation model according to the battery expansion data respectively;

[0008] determining battery key features according to feature screening and fusion of the target entropy value model, the battery temperature model and the battery fluctuation model;

[0009] establishing a target battery diagnosis model according to the battery key features;

[0010] inputting the battery data to be diagnosed into the target battery diagnosis model to obtain the battery diagnosis result of the target vehicle.

[0011] In an embodiment, the battery expansion data is determined by data expansion of the battery correlation data, which comprises:

[0012] preprocessing the battery correlation data to obtain preprocessed battery correlation data;

[0013] The outlier data in the preprocessed battery correlation data is removed to obtain battery cleaning data;

[0014] The battery cleaning data is normalized to obtain normalized battery cleaning data;

[0015] The normalized battery cleaning data is data augmented to obtain battery augmented data.

[0016] In an embodiment, before the outlier data in the preprocessed battery correlation data is removed to obtain battery cleaning data, it further includes:

[0017] According to a preset abnormal factor strategy and the preprocessed battery correlation data, the data outlier degree is determined;

[0018] According to the outlier degree threshold and the data outlier degree, data screening is performed to obtain a data screening result;

[0019] According to the data screening result, the outlier data in the preprocessed battery correlation data is determined.

[0020] In an embodiment, the normalized battery cleaning data is data augmented to obtain battery augmented data, including:

[0021] According to the target category data in the normalized battery cleaning data, the corresponding target data distance is determined;

[0022] The target data distance is standardized to obtain a target standard distance;

[0023] According to the target standard distance, the data distance ratio of the target category data is determined;

[0024] According to the data distance ratio of the target category data, the battery augmented data is determined.

[0025] In an embodiment, the battery augmented data is determined according to the data distance ratio of the target category data, including:

[0026] According to the data distance ratio of the target category data, the reference sample data is determined;

[0027] According to a linear interpolation strategy, the target category data, and the reference sample data, target generated data is obtained;

[0028] According to the target generated data and the target category data, the battery augmented data is determined.

[0029] In one embodiment, the step of determining key battery features by feature filtering and fusion based on the target entropy model, the battery temperature model, and the battery fluctuation model includes:

[0030] Feature extraction is performed based on the target entropy model, the battery temperature model, and the battery fluctuation model to obtain extracted feature data;

[0031] The extracted feature data is then filtered to obtain filtered feature data;

[0032] The selected feature data is fused to obtain the key features of the battery.

[0033] In one embodiment, establishing a target battery diagnostic model based on the key battery features includes:

[0034] An initial index fusion model is established based on the key characteristics of the battery.

[0035] The initial index fusion model is trained according to the preset optimization strategy to obtain the model training results;

[0036] The target battery diagnostic model is obtained based on the training results of the model.

[0037] Furthermore, to achieve the above objectives, this application also proposes a vehicle battery fault diagnosis device, which includes:

[0038] The processing module is used to augment the battery-related data and determine the augmented battery data.

[0039] A module is established to build a target entropy model, a battery temperature model, and a battery fluctuation model based on the battery expansion data.

[0040] The filtering module is used to perform feature filtering and fusion based on the target entropy model, the battery temperature model, and the battery fluctuation model to determine key battery features;

[0041] The establishment module is also used to establish a target battery diagnostic model based on the key battery features;

[0042] The diagnostic module is used to input the battery data to be diagnosed into the target battery diagnostic model to obtain the battery diagnostic results of the target vehicle.

[0043] In addition, to achieve the above objectives, this application also proposes a vehicle battery fault diagnosis device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the vehicle battery fault diagnosis method as described above.

[0044] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the vehicle battery fault diagnosis method described above.

[0045] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the vehicle battery fault diagnosis method described above.

[0046] This application determines battery augmentation data by augmenting battery-related data; establishes a target entropy model, a battery temperature model, and a battery fluctuation model based on the augmented battery data; performs feature filtering and fusion based on the target entropy model, the battery temperature model, and the battery fluctuation model to determine key battery features; establishes a target battery diagnostic model based on the key battery features; inputs the battery data to be diagnosed into the target battery diagnostic model to obtain the battery diagnostic results for the target vehicle. Through multi-indicator fusion, abnormal battery conditions are accurately identified, allowing for proactive maintenance and repair measures, thereby extending battery life and improving the reliability and safety of new energy vehicles. Attached Figure Description

[0047] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0048] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 This is a flowchart illustrating an embodiment of the vehicle battery fault diagnosis method of this application.

[0050] Figure 2 This is a schematic diagram of the architecture of a new energy vehicle battery monitoring system based on multi-index fusion, provided in Embodiment 1 of the vehicle battery fault diagnosis method of this application.

[0051] Figure 3 This is a schematic diagram of the LSTM-based multi-index fusion model provided in Embodiment 1 of the vehicle battery fault diagnosis method of this application;

[0052] Figure 4 This is a flowchart illustrating Embodiment 2 of the vehicle battery fault diagnosis method of this application.

[0053] Figure 5 This is a schematic diagram of the module structure of the vehicle battery fault diagnosis device according to an embodiment of this application;

[0054] Figure 6 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the vehicle battery fault diagnosis method in this application embodiment.

[0055] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0056] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0057] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0058] The main solution of this application embodiment is as follows: Battery-related data is augmented to determine augmented battery data; a target entropy model, a battery temperature model, and a battery fluctuation model are established based on the augmented battery data; feature filtering and fusion are performed based on the target entropy model, the battery temperature model, and the battery fluctuation model to determine key battery features; a target battery diagnostic model is established based on the key battery features; the battery data to be diagnosed is input into the target battery diagnostic model to obtain the battery diagnostic results for the target vehicle.

[0059] With the booming development of the new energy vehicle industry, the safety and reliability of power batteries have become a focus of public attention. Battery fault diagnosis and risk assessment are crucial for ensuring the safe operation of vehicles, but existing technologies mostly focus on the detection of specific faults or individual cell anomalies, lacking a holistic risk assessment, and are usually only applicable to specific operating conditions or vehicle models. In addition, these technologies often start from the perspective of algorithm design without a deep understanding of the nature of battery faults, resulting in low accuracy in detecting high-risk vehicles.

[0060] Given that battery performance is influenced by multiple factors, single-parameter evaluation is clearly insufficient. Real-world faults often involve multiple characteristics, making a holistic diagnostic and risk assessment system particularly necessary. Therefore, addressing the lack of a comprehensive risk assessment mechanism in existing new energy battery monitoring technologies has become an urgent problem to be solved.

[0061] This application determines battery augmentation data by augmenting battery-related data; establishes a target entropy model, a battery temperature model, and a battery fluctuation model based on the augmented battery data; performs feature filtering and fusion based on the target entropy model, the battery temperature model, and the battery fluctuation model to determine key battery features; establishes a target battery diagnostic model based on the key battery features; inputs the battery data to be diagnosed into the target battery diagnostic model to obtain the battery diagnostic results for the target vehicle. Through multi-indicator fusion, abnormal battery conditions are accurately identified, allowing for proactive maintenance and repair measures, thereby extending battery life and improving the reliability and safety of new energy vehicles.

[0062] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or a vehicle battery fault diagnosis device capable of performing the above functions. The following description uses a vehicle battery fault diagnosis device as the executing entity to illustrate this embodiment and the subsequent embodiments.

[0063] Based on this, embodiments of this application provide a method for diagnosing vehicle battery faults, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the vehicle battery fault diagnosis method of this application.

[0064] In this embodiment, the vehicle battery fault diagnosis method includes steps S10 to S50:

[0065] Step S10: Expand the battery-related data to determine the expanded battery data;

[0066] It should be noted that, as Figure 2As shown, this embodiment improves data quality by removing outliers from preprocessed data and expanding the sample to address uneven distribution of samples in risk scenarios. Based on this, three models are established: an entropy model reflecting outlier anomalies in single-cell voltage, a temperature consistency diagnostic model reflecting battery temperature anomalies, and a capacity consistency diagnostic model reflecting battery capacity anomalies. These models extract intrinsic feature data characterizing entropy, temperature consistency, and capacity consistency. To further improve the accuracy and robustness of the models, the feature parameters extracted from the three models are filtered and fused using the Pearson correlation coefficient. This step helps eliminate redundant features and retain the most representative feature parameters, thereby improving the training efficiency and predictive performance of the models. Finally, a new energy vehicle battery monitoring model based on multi-indicator fusion is constructed. This model comprehensively utilizes multiple indicator information to accurately monitor the battery status and provide timely alarm and warning functions. This multi-indicator fusion method can more accurately identify battery anomalies, allowing for proactive maintenance and repair measures, thereby extending battery life and improving the reliability and safety of new energy vehicles.

[0067] It is understandable that battery-related data includes key parameters such as battery voltage, temperature, current, charge / discharge status, and health status, while battery expansion data refers to battery data after data type balancing.

[0068] In practice, outlier removal and data normalization are performed sequentially on the original data. A dataset consisting of training, validation, and test sets is then constructed using data samples from a certain number of vehicles. However, in actual operation, the data suffers from class imbalance. This embodiment employs the Safe-Level SMOTE algorithm to balance the dataset by synthesizing new minority class samples, resulting in battery data with balanced data types.

[0069] It should be noted that new energy vehicles utilize the advanced monitoring capabilities of their onboard terminals and sensor devices to achieve real-time monitoring of the vehicle's power battery pack and its individual battery cells. These devices accurately collect data on key parameters, including but not limited to battery voltage, temperature, current, charge / discharge status, and health status, thus providing a detailed usage record for each battery cell. As the vehicle operates daily, this data gradually accumulates, forming a complete lifecycle data record, which is automatically uploaded to a pre-set big data platform via the onboard communication system.

[0070] Step S20: Establish the target entropy model, battery temperature model, and battery fluctuation model based on the battery expansion data.

[0071] It is understandable that the target entropy model refers to the battery entropy model, the battery temperature model refers to the temperature fluctuation consistency diagnostic model, and the battery fluctuation model refers to the capacity fluctuation consistency diagnostic model.

[0072] In practical implementation, multiple feature parameters are extracted: An entropy model reflecting outlier anomalies in single-cell voltage, a temperature consistency diagnostic model reflecting battery temperature anomalies, and a capacity consistency diagnostic model reflecting battery capacity anomalies are established for the dataset. Intrinsic feature data characterizing entropy, temperature consistency, and capacity consistency are extracted for each model. The entropy model is based on the Shannon information entropy and Z-score of the single-cell voltage in the dataset, used to identify numerical outlier features and trend dispersion features of abnormal single-cell voltage data. The intrinsic entropy features specifically include the extreme values, mean, and variance of the Z-score values ​​of abnormal single cells. The temperature fluctuation consistency diagnostic model is based on the variance and median of the battery temperature window in the dataset, aiming to identify battery temperature fluctuation consistency faults. Specifically, this model selects the extreme values, mean, and variance of the variance values ​​of the abnormal battery temperature window as indicators of fluctuation consistency. The capacity fluctuation consistency diagnostic model is based on the variance and median of the battery capacity window in the dataset, used to identify battery capacity fluctuation consistency faults. Similarly, this model selects the extreme values, mean, and variance of the variance values ​​of the abnormal battery capacity window as feature parameters of fluctuation consistency.

[0073] Step S30: Based on the target entropy model, the battery temperature model, and the battery fluctuation model, feature filtering and fusion are performed to determine the key features of the battery;

[0074] Understandably, key battery features refer to the feature vector synthesized from features that truly contribute to model training and prediction performance.

[0075] In practice, a series of feature parameters were extracted using the three models mentioned above. However, before using these features for subsequent model training, the features that truly contribute to model training and prediction performance were selected, thus obtaining the key battery features.

[0076] In one feasible implementation, step S30 may include steps A31 to A33:

[0077] Step A31: Perform feature extraction based on the target entropy model, the battery temperature model, and the battery fluctuation model to obtain extracted feature data;

[0078] It is understandable that extracting feature data refers to all feature data extracted through the target entropy model, battery temperature model, and battery fluctuation model.

[0079] In practice, the target entropy model, battery temperature model, and battery fluctuation model are used to extract battery-related feature data to obtain the extracted feature data.

[0080] Step A32: Perform feature filtering on the extracted feature data to obtain filtered feature data;

[0081] It is understandable that selecting feature data refers to selecting feature data that truly contributes to model training and prediction performance.

[0082] In practice, all feature data extracted through the target entropy model, battery temperature model, and battery fluctuation model are used to calculate the corresponding correlation coefficient matrix using the Pearson correlation coefficient. Features with an absolute correlation value greater than 0.8 are then selected to obtain the feature data that truly contributes to model training and prediction performance.

[0083] Step A33: Perform feature fusion on the selected feature data to obtain key battery features.

[0084] Understandably, a series of feature parameters were extracted using the battery entropy model, temperature fluctuation consistency diagnostic model, and capacity fluctuation consistency diagnostic model. However, before using these features for subsequent model training, a crucial step is to select those features that truly contribute to model training and prediction performance. This embodiment employs correlation analysis for selection: Pearson correlation coefficients are used to calculate the pairwise correlation coefficient matrices for the aforementioned risk features, and features with an absolute correlation value greater than 0.8 are selected. Features with high correlation between two features are discarded. For two features X and Y, their Pearson correlation coefficient r... XY The calculation formula is:

[0085]

[0086] Among them, X i and Y i These are the observed values ​​of features X and Y, where X and Y are the features observed. Their average value is n, where n is the number of observations. The filtered features are combined into a single feature vector. This feature vector will contain the most valuable information from the three models, providing an optimized set of inputs for subsequent neural network models.

[0087] Step S40: Establish a target battery diagnostic model based on the key battery features;

[0088] It is understandable that the target battery diagnostic model refers to a multi-indicator fusion diagnostic model based on deep learning.

[0089] In the specific implementation, a battery entropy model, a temperature fluctuation consistency diagnostic model, and a capacity fluctuation consistency diagnostic model are constructed on the preprocessed dataset. Relevant feature parameters are extracted, and correlation analysis is performed for comprehensive screening. Features that are meaningless to the fusion of the three models and the training of the neural network are removed. The multiple risks that are retained are merged into a feature vector as the input of a deep learning-based multi-index fusion diagnostic model to determine whether the vehicle has a risk. Considering that vehicle data collection is carried out within a life cycle, the extracted features are related to time. Therefore, this embodiment uses an LSTM-based multi-index fusion model to identify risky vehicles.

[0090] In one feasible implementation, step S40 may include steps A41 to A43:

[0091] Step A41: Establish an initial index fusion model based on the key battery characteristics;

[0092] It is understandable that the initial index fusion model refers to the battery diagnostic neural network model before optimization.

[0093] Step A42: Train the initial index fusion model according to the preset optimization strategy to obtain the model training result;

[0094] It is understandable that the preset optimization strategy refers to the Adam optimization algorithm, and the model training result refers to whether the model has been successfully trained.

[0095] In practice, the diagnostic results of the neural network (whether it is a risky vehicle) are compared with the actual weight values. In this embodiment, the Adam optimization algorithm is used to optimize and train the LSTM neural network model to obtain the result of whether the model has been trained successfully.

[0096] Step A43: Obtain the target battery diagnostic model based on the model training results.

[0097] Understandably, the target battery diagnostic model is obtained after training the unoptimized battery diagnostic neural network model.

[0098] It should be noted that this embodiment constructs a multi-index fusion model based on LSTM, and the model architecture is as follows: Figure 3As shown, the input layer: The feature vector obtained by fusing the above features is input into the network, therefore the dimension of the input layer is consistent with the length of the vector obtained after feature fusion. LSTM layer: The LSTM layer is used to process sequential data and capture temporal dependencies. Multiple LSTM units are stacked to build a deeper network, extracting higher-dimensional relevant features to enhance model performance. Dropout is used to mitigate the risk of overfitting. Fully connected layer: After the LSTM layer, a fully connected layer is added to further extract features and learn non-linear relationships, and the ReLU activation function is used to increase non-linearity. Output layer: The output layer is used to predict whether a vehicle is at risk. A sigmoid activation function is used to map a single output value between 0 and 1, thereby diagnosing whether a vehicle is at risk. Loss function: In this embodiment, the focal cross-entropy loss function is used to correct the neural network.

[0099] FL(p)=-α(1-p) γ log(p)-(1-α)p γ log(1-p)

[0100] Where: p is the probability that the model predicts a positive class (such as a faulty vehicle), α is the reciprocal of the proportion of positive class samples in the dataset, used to balance the contributions of positive and negative classes, and γ is the focus parameter, which reduces the loss contribution of easily classified (risk-free) samples, thereby making the model pay more attention to samples that are difficult to classify (risky), thus improving the model's performance in identifying risks.

[0101] It should be noted that, as shown in Table 1, this embodiment uses hit rate (POD), false alarm rate (FAR), and critical success index (CSI) to test the effectiveness of the model's diagnosis.

[0102] Table 1:

[0103]

[0104] The formula for hit rate POD is as follows: FAR represents the proportion of incorrect model diagnoses out of the total number of risk-free vehicles in real-world data. It can be seen that the more times the model makes incorrect diagnoses, the larger the FAR and the lower the model's accuracy; conversely, the smaller the FAR, the higher the model's accuracy. This embodiment uses the Critical Success Index (CSI) to represent the combined effect of POD and FAR. The formula for CSI is shown below: CSI represents the proportion of correctly diagnosed vehicles that may be at risk. The closer the CSI value is to 1, the higher the model's accuracy. Therefore, in practical validation, we hope that the values ​​of POD and CSI can be as close to 1 as possible, and the value of FAR can be as close to 0 as possible. At this point, the model's diagnostic results for risky and non-risky vehicles are more accurate.

[0105] Based on the aforementioned model evaluation metrics, the diagnostic results of the neural network (whether it is a risky vehicle) are compared with the actual weight values. In this embodiment, the Adam optimization algorithm is used to optimize and train the LSTM neural network model. Adam combines the advantages of both AdaGrad and RMSProp optimization algorithms, and can adaptively adjust the learning rate, which is very useful when dealing with non-stationary objective functions. Adam is a first-order optimization algorithm that can replace the traditional stochastic gradient descent process. Compared with other stochastic gradient descent optimization algorithms, it has the following advantages: a. Efficiency: Adam usually requires less memory and has proven to be very efficient in practice; b. Applicability to big data and high-dimensional spaces: Adam often performs better when facing high-dimensional parameter spaces or large-scale datasets; c. Robustness: Compared with other optimization algorithms, Adam is less likely to get trapped in local optima, especially in complex, high-dimensional optimization problems.

[0106] It should be understood that risk characteristic data collected from actual vehicles and uploaded to a big data platform are input into a trained neural network to assess the risk status of the vehicle's power battery. In this way, fault diagnosis results not only provide precise guidance for vehicle maintenance but also help prevent potential faults in advance, thereby reducing the risk of unexpected downtime and improving vehicle safety and reliability. Furthermore, these diagnostic results can provide valuable feedback to battery manufacturers, helping them improve product design and manufacturing processes, and extend battery life.

[0107] Step S50: Input the battery data to be diagnosed into the target battery diagnostic model to obtain the battery diagnostic results of the target vehicle.

[0108] It is understandable that the battery data to be diagnosed refers to the key input data of the battery to be diagnosed, and the battery diagnosis results include whether the battery has a fault.

[0109] In practice, the key input data of the vehicle battery to be diagnosed is input into the target battery diagnostic model to diagnose battery faults, thereby obtaining the result of whether the target vehicle has a fault.

[0110] This embodiment determines battery augmentation data by augmenting battery-related data; based on the augmented battery data, it establishes a target entropy model, a battery temperature model, and a battery fluctuation model; based on the target entropy model, the battery temperature model, and the battery fluctuation model, it performs feature filtering and fusion to determine key battery features; based on the key battery features, it establishes a target battery diagnostic model; and it inputs the battery data to be diagnosed into the target battery diagnostic model to obtain the battery diagnostic results for the target vehicle. By using a multi-indicator fusion method, it accurately identifies abnormal battery conditions, allowing for proactive maintenance and repair measures, thereby extending battery life and improving the reliability and safety of new energy vehicles.

[0111] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 4 The vehicle battery fault diagnosis method further includes steps S11 to S14 in step S10:

[0112] Step S11: Preprocess the battery association data to obtain preprocessed battery association data;

[0113] Understandably, preprocessing methods include deduplication, smoothing, and removal of null values.

[0114] In practice, the big data platform performs preprocessing on the data, such as deduplication, smoothing, and removal of null and missing values, to obtain the original data, which is the preprocessed battery-related data.

[0115] Step S12: Remove outlier data from the preprocessed battery association data to obtain battery cleaning data;

[0116] It is understandable that outlier data refers to outlier values ​​in the preprocessed battery association data, while battery cleansing data refers to the battery data after outlier data has been removed.

[0117] In this implementation, the Local Outlier Factor (LOF) algorithm is used to process outlier data. Using the LOF algorithm not only eliminates the influence of test errors on the regularity of feature quantities, but also yields battery data after removing outliers, i.e., battery cleaned data.

[0118] In one feasible implementation, steps A121 to A123 may be included before step S12:

[0119] Step A121: Determine the degree of data outlier based on the preset anomaly factor strategy and the preprocessed battery correlation data;

[0120] It is understandable that the preset anomaly factor strategy refers to the Local Outlier Factor (LOF) algorithm, and the data outlier degree refers to the outlier degree of each data point in the preprocessed battery correlation data.

[0121] In practice, the outlier degree of each data point in the preprocessed battery-related data is calculated using the Local Outlier Factor (LOF) algorithm, thereby obtaining the outlier degree of the data.

[0122] Step A122: Filter the data based on the outlier threshold and the degree of outlier in the data to obtain the data filtering results;

[0123] It is understandable that the outlier threshold refers to the critical value used to judge the outlier degree of outlier data, and the data filtering result refers to the filtering result of outlier data.

[0124] In practice, the outlier degree of each data point in the preprocessed battery association data is compared with the outlier degree threshold used to judge outlier data, and the result of the magnitude relationship between the outlier degree of each data point and the outlier degree threshold used to judge outlier data is obtained.

[0125] Step A123: Determine outlier data in the preprocessed battery association data based on the data filtering results.

[0126] In practice, based on the relationship between the outlier degree of each data point and the outlier degree threshold used to determine outlier data, outlier data in the battery-related data is selected by filtering out data whose outlier degree is greater than the outlier degree threshold used to determine outlier data.

[0127] It should be noted that the big data platform performs preprocessing such as deduplication, smoothing, and removal of null and missing values ​​to obtain the original data. However, during the data acquisition process, real battery characteristic data may be mixed with error interference data, resulting in obvious outliers in the original data. This may be due to data acquisition errors and battery degradation characteristics. To solve this problem, this embodiment uses the Local Outlier Factor (LOF) algorithm to process outlier data. Using the LOF algorithm can not only eliminate the influence of test errors on the regularity of feature quantities, but also retain the features reflected by the physicochemical characteristics during the lithium battery degradation process, thereby effectively improving the information content of feature quantities and the learning efficiency of the model. LOF is a density-based outlier detection method that determines the degree of outlier status by calculating the LOF of each sample. Based on the set threshold, invalid outliers can be accurately located, avoiding the problem of incorrect cleaning of valid data by outlier discrimination methods based on data distribution. In the feature quantities, the ratio of the local reachability density of a point to the average local reachability density of its neighboring points is expressed as the LOF of point n, and the specific calculation formula is as follows:

[0128]

[0129] Where N K (p) represents the k-distance neighborhoods of point n, where point m is one of the points in the n-neighborhood. k (p) and lrd k (m) represents the local reachability density of point n and point m, respectively. When the LOF value is close to 1, it indicates that the surrounding density of point n and its neighboring points is basically the same, and the probability of n and its neighboring points being in the same cluster is higher. Conversely, n may be an outlier.

[0130] Step S13: Normalize the battery cleaning data to obtain normalized battery cleaning data.

[0131] Understandably, data normalization is used: In this embodiment, the Z-score method is employed to normalize the data by converting it to a standard normal distribution (mean of 0, standard deviation of 1). The principle of Z-score normalization is to subtract the mean of each feature column from the data in that column, and then divide by the standard deviation of that column. The specific calculation formula is as follows:

[0132]

[0133] Z-score processing ensures that the data in each feature column conforms to a standard normal distribution, i.e., a mean of 0 and a standard deviation of 1. This approach eliminates the influence of different dimensions on each feature, giving them the same scale and better adapting to the requirements of certain deep learning algorithms.

[0134] Step S14: Perform data augmentation on the normalized battery cleaning data to obtain battery augmentation data.

[0135] Understandably, after performing outlier removal and data normalization on the original data, a dataset consisting of training, validation, and test sets is constructed by selecting a certain number of vehicle data samples. However, in actual operation, the data suffers from class imbalance. This embodiment employs the Safe-Level SMOTE algorithm to balance the dataset by synthesizing new minority class samples, while preserving the data distribution characteristics and avoiding the risk of overfitting as much as possible.

[0136] In one feasible implementation, steps A141 to A144 may be included before step S14:

[0137] Step A141: Determine the corresponding target data distance based on the target type data in the normalized battery cleaning data;

[0138] It is understandable that target class data refers to minority class sample types. For example, if the number of temperature samples is 5, the number of voltage samples is 6, and the number of current samples is 2, then the current sample is determined to be a minority class sample type. Target data distance refers to the ratio between each minority class sample and all other minority class samples.

[0139] In practice, for each minority class sample x i Calculate its relationship with all other minority class samples x. j The Euclidean distance d(x) between them i ,x j The formula is as follows:

[0140]

[0141] Where n is the number of features.

[0142] Step A142: Standardize the target data distance to obtain the target standard distance;

[0143] It is understandable that the target standard distance refers to the distance of the target data after standardization.

[0144] In practice, the distance matrix is ​​standardized to ensure that all distance metrics have the same weight. The standardized distance calculation formula is as follows:

[0145]

[0146] Wherein, min(d(x) i ,·)) and max(d(x i ,·)) are respectively x i The minimum and maximum distances to all other samples.

[0147] Step A143: Determine the data distance ratio of the target type data based on the target standard distance;

[0148] Understandably, the data distance ratio refers to the ratio of the nearest distance from each minority class sample to another minority class sample to the nearest distance to the majority class sample.

[0149] In practice, for each minority class sample x i Calculate its Safe-Level, which is the ratio of the nearest distance to the minority class sample to the nearest distance to the majority class sample. The formula for calculating Safe-Level is:

[0150]

[0151] in, It is x i The minimum standardized distance to other minority class samples. It is x i The minimum standardized distance to other majority class samples.

[0152] Step A144: Determine battery expansion data based on the data distance ratio of the target type data.

[0153] Understandably, linear interpolation is performed based on the data distance ratio (Safe-Level) of the minority class data to obtain synthetic samples, and thus battery expansion data.

[0154] In one possible implementation, steps B1441 to B1443 may be included before step A144:

[0155] Step B1441: Determine reference sample data based on the data distance ratio of the target type data;

[0156] It is understandable that the reference sample data refers to the smallest nearest neighbor as the reference sample data.

[0157] In practice, based on the calculated data distance ratio (Safe-Level), the nearest neighbor with the smallest Safe-Level is selected as the reference sample for the current sample.

[0158] Step B1442: Based on the linear interpolation strategy, the target type data, and the reference sample data, the target generation data is obtained;

[0159] It is understandable that linear interpolation strategy refers to the method of linear interpolation, and target generated data refers to the synthetic data generated by linear interpolation.

[0160] In practice, the number of synthetic samples (N) to be generated is determined based on the Safe-Level value, and the calculation formula is as follows:

[0161] N = (Safe-Level(x) i )-1)×α

[0162] Where α is a parameter that controls the number of samples generated. In the current sample x i and the selected reference sample x r N synthetic samples are generated using linear interpolation. The formula for linear interpolation is:

[0163] x new =x i +β×(x r -x i ),β∈[0,1]

[0164] Where β is the interpolation ratio determined based on N and α.

[0165] Step B1443: Determine battery expansion data based on the target generation data and the target type data.

[0166] In practice, the generated synthetic samples are added to the minority class sample set to increase its quantity, thus obtaining the battery expansion data.

[0167] It should be noted that the Safe-Level SMOTE algorithm adds consideration to the local structure of the data on top of the traditional SMOTE algorithm. By introducing the concept of Safe-Level, the oversampling process becomes more refined and safer. This algorithm is particularly suitable for situations where minority class samples are surrounded by majority class samples. It can effectively increase the number of minority class samples without destroying the original data structure, thereby improving the model's generalization ability.

[0168] This embodiment preprocesses battery-related data to obtain preprocessed battery-related data; outliers are removed from the preprocessed battery-related data to obtain cleaned battery data; the cleaned battery data is then normalized to obtain normalized battery-related data; and the normalized battery-related data is augmented to obtain augmented battery data. By sequentially performing outlier removal and data normalization on the original data, a dataset including a training set, validation set, and test set is constructed by selecting a certain number of vehicle data samples. Then, new minority class samples are synthesized to balance the dataset, while preserving the data distribution characteristics and avoiding the risk of overfitting as much as possible.

[0169] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the vehicle battery fault diagnosis method of this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0170] This application also provides a vehicle battery fault diagnosis device; please refer to... Figure 5 The vehicle battery fault diagnosis device includes:

[0171] Processing module 10 is used to augment battery-related data and determine battery augmentation data;

[0172] Module 20 is used to establish a target entropy model, a battery temperature model, and a battery fluctuation model based on the battery expansion data.

[0173] The filtering module 30 is used to perform feature filtering and fusion based on the target entropy model, the battery temperature model and the battery fluctuation model to determine the key features of the battery.

[0174] The establishment module 20 is also used to establish a target battery diagnostic model based on the key battery features;

[0175] The diagnostic module 40 is used to input the battery data to be diagnosed into the target battery diagnostic model to obtain the battery diagnostic results of the target vehicle.

[0176] Optionally, the processing module 10 is further configured to:

[0177] The battery association data is preprocessed to obtain preprocessed battery association data.

[0178] Outlier data is removed from the preprocessed battery association data to obtain battery cleaning data;

[0179] The battery cleaning data is normalized to obtain normalized battery cleaning data.

[0180] The normalized battery cleaning data is augmented to obtain battery augmented data.

[0181] Optionally, the processing module 10 is further configured to:

[0182] The degree of data outlier is determined based on the preset anomaly factor strategy and the preprocessed battery correlation data.

[0183] Data is filtered based on outlier threshold and outlier level to obtain data filtering results;

[0184] Outlier data in the preprocessed battery association data are determined based on the data filtering results.

[0185] Optionally, the processing module 10 is further configured to:

[0186] The corresponding target data distance is determined based on the target type data in the normalized battery cleaning data.

[0187] The target data distance is standardized to obtain the target standard distance;

[0188] Determine the data distance ratio of the target type data based on the target standard distance;

[0189] Battery expansion data is determined based on the data distance ratio of the target type data.

[0190] Optionally, the processing module 10 is further configured to:

[0191] Reference sample data is determined based on the data distance ratio of the target category data;

[0192] Target generation data is obtained based on the linear interpolation strategy, the target type data, and the reference sample data;

[0193] Battery expansion data is determined based on the target generation data and the target type data.

[0194] Optionally, the filtering module 30 is further configured to:

[0195] Feature extraction is performed based on the target entropy model, the battery temperature model, and the battery fluctuation model to obtain extracted feature data;

[0196] The extracted feature data is then filtered to obtain filtered feature data;

[0197] The selected feature data is fused to obtain the key features of the battery.

[0198] Optionally, the establishment module 20 is further configured to:

[0199] An initial index fusion model is established based on the key characteristics of the battery.

[0200] The initial index fusion model is trained according to the preset optimization strategy to obtain the model training results;

[0201] The target battery diagnostic model is obtained based on the training results of the model.

[0202] The vehicle battery fault diagnosis device provided in this application, employing the vehicle battery fault diagnosis method described in the above embodiments, can solve the technical problem of the lack of a comprehensive risk assessment in existing new energy battery monitoring technologies. Compared with the prior art, the beneficial effects of the vehicle battery fault diagnosis device provided in this application are the same as those of the vehicle battery fault diagnosis method provided in the above embodiments, and other technical features in the vehicle battery fault diagnosis device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0203] This application provides a vehicle battery fault diagnosis device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the vehicle battery fault diagnosis method in the above embodiment 1.

[0204] The following is for reference. Figure 6 The diagram illustrates a structural schematic suitable for implementing a vehicle battery fault diagnosis device according to embodiments of this application. The vehicle battery fault diagnosis device in these embodiments may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 6 The vehicle battery fault diagnosis device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0205] like Figure 6 As shown, the vehicle battery fault diagnosis device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the vehicle battery fault diagnosis device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. The communication device 1009 allows the vehicle battery fault diagnosis equipment to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows a vehicle battery fault diagnosis equipment with various systems, it should be understood that implementation or possession of all the systems shown is not required. More or fewer systems may be implemented alternatively.

[0206] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0207] The vehicle battery fault diagnosis device provided in this application, employing the vehicle battery fault diagnosis method described in the above embodiments, can solve the technical problem of the lack of a comprehensive risk assessment in existing new energy battery monitoring technologies. Compared with the prior art, the beneficial effects of the vehicle battery fault diagnosis device provided in this application are the same as those of the vehicle battery fault diagnosis method provided in the above embodiments, and other technical features of this vehicle battery fault diagnosis device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0208] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0209] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0210] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the vehicle battery fault diagnosis method in the above embodiments.

[0211] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0212] The aforementioned computer-readable storage medium may be included in the vehicle battery fault diagnosis equipment; or it may exist independently and not be installed in the vehicle battery fault diagnosis equipment.

[0213] The aforementioned computer-readable storage medium carries one or more programs. When these programs are executed by a vehicle battery fault diagnosis device, the vehicle battery fault diagnosis device performs the following actions: augments battery-related data to determine augmented battery data; establishes a target entropy model, a battery temperature model, and a battery fluctuation model based on the augmented battery data; performs feature filtering and fusion based on the target entropy model, the battery temperature model, and the battery fluctuation model to determine key battery features; establishes a target battery diagnosis model based on the key battery features; and inputs the battery data to be diagnosed into the target battery diagnosis model to obtain the battery diagnosis result for the target vehicle.

[0214] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0215] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0216] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0217] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described vehicle battery fault diagnosis method, thereby solving the technical problem that existing new energy battery monitoring technologies lack a comprehensive risk assessment. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the vehicle battery fault diagnosis method provided in the above embodiments, and will not be repeated here.

[0218] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the vehicle battery fault diagnosis method described above.

[0219] The computer program product provided in this application can solve the technical problem that existing new energy battery monitoring technologies lack a comprehensive risk assessment. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the vehicle battery fault diagnosis method provided in the above embodiments, and will not be repeated here.

[0220] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for diagnosing vehicle battery faults, characterized in that, The vehicle battery fault diagnosis method includes: The battery association data is preprocessed to obtain preprocessed battery association data. Outlier data is removed from the preprocessed battery association data to obtain battery cleaning data; The battery cleaning data is normalized to obtain normalized battery cleaning data. The corresponding target data distance is determined based on the target type data in the normalized battery cleaning data. The target data distance is standardized to obtain the target standard distance; Determine the data distance ratio of the target type data based on the target standard distance; Reference sample data is determined based on the data distance ratio of the target category data; Target generation data is obtained based on the linear interpolation strategy, the target type data, and the reference sample data; Battery expansion data is determined based on the target generation data and the target type data; Based on the battery expansion data, establish a target entropy model, a battery temperature model, and a battery fluctuation model respectively. Based on the target entropy model, the battery temperature model, and the battery fluctuation model, feature selection and fusion are performed to determine the key features of the battery; Establish a target battery diagnostic model based on the key battery characteristics described; Input the battery data to be diagnosed into the target battery diagnostic model to obtain the battery diagnostic results for the target vehicle; The step of determining key battery features by feature filtering and fusion based on the target entropy model, the battery temperature model, and the battery fluctuation model includes: Feature extraction is performed based on the target entropy model, the battery temperature model, and the battery fluctuation model to obtain extracted feature data; The extracted feature data is then filtered to obtain filtered feature data; The selected feature data is fused to obtain the key features of the battery.

2. The method as described in claim 1, characterized in that, Before removing outliers from the preprocessed battery association data to obtain battery cleaning data, the process further includes: The degree of data outlier is determined based on the preset anomaly factor strategy and the preprocessed battery correlation data. Data is filtered based on outlier threshold and outlier level to obtain data filtering results; Outlier data in the preprocessed battery association data are determined based on the data filtering results.

3. The method as described in claim 1, characterized in that, The step of establishing a target battery diagnostic model based on the key battery features includes: An initial index fusion model is established based on the key characteristics of the battery. The initial index fusion model is trained according to the preset optimization strategy to obtain the model training results; The target battery diagnostic model is obtained based on the training results of the model.

4. A vehicle battery fault diagnosis device, characterized in that, The device includes: The processing module is used to preprocess battery-related data to obtain preprocessed battery-related data; remove outlier data from the preprocessed battery-related data to obtain battery cleaning data; normalize the battery cleaning data to obtain normalized battery cleaning data; determine the corresponding target data distance based on the target type data in the normalized battery cleaning data; standardize the target data distance to obtain target standard distance; determine the data distance ratio of the target type data based on the target standard distance; determine reference sample data based on the data distance ratio of the target type data; obtain target generation data based on a linear interpolation strategy, the target type data, and the reference sample data; and determine battery expansion data based on the target generation data and the target type data. A module is established to build a target entropy model, a battery temperature model, and a battery fluctuation model based on the battery expansion data. The filtering module is used to perform feature filtering and fusion based on the target entropy model, the battery temperature model, and the battery fluctuation model to determine key battery features; The establishment module is also used to establish a target battery diagnostic model based on the key battery features; The diagnostic module is used to input the battery data to be diagnosed into the target battery diagnostic model to obtain the battery diagnostic results of the target vehicle. The filtering module is further configured to extract features based on the target entropy model, the battery temperature model, and the battery fluctuation model to obtain extracted feature data; filter the extracted feature data to obtain filtered feature data; and fuse the filtered feature data to obtain key battery features.

5. A vehicle battery fault diagnosis device, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the vehicle battery fault diagnosis method as described in any one of claims 1 to 3.

6. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the vehicle battery fault diagnosis method as described in any one of claims 1 to 3.

Images