Vehicle early warning method, device and equipment and storage medium

Abstract

The application provides a vehicle early warning method, device and equipment and a storage medium, relates to the technical field of vehicle early warning, and comprises the following steps: acquiring a battery pack early warning parameter time sequence set of each vehicle, wherein the early warning parameter time sequence set comprises a pressure difference time sequence set, a temperature difference time sequence set and a pressure difference entropy time sequence set; determining a feature vector corresponding to each vehicle according to the early warning parameter time sequence set, wherein the feature vector comprises a pressure difference dimension component, a temperature difference dimension component and a pressure difference entropy dimension component; dividing the feature vector corresponding to each vehicle into a plurality of sets, and each set comprises the feature vector corresponding to the vehicle of the same battery pack type; and determining an abnormal vehicle based on an anomaly detection algorithm and according to the feature vector corresponding to the vehicle in each set. The abnormality of different vehicle battery packs is evaluated by comprehensively considering multi-dimensional battery pack performance parameters, which is more comprehensive, helps to improve the accuracy of abnormal battery pack identification, and thus improves the accuracy of vehicle early warning.

Description

Technical Field

[0001] This application relates to the field of vehicle warning technology, and in particular to a vehicle warning method, device, equipment and storage medium. Background Technology

[0002] With the explosive growth of the new energy vehicle market, the safety of new energy vehicles has gradually become an increasingly important issue for the industry and users. As the core component of new energy vehicles, especially electric vehicles, the safety of the power battery directly determines the safety of the entire vehicle.

[0003] Currently, each vehicle performs power battery anomaly detection based on the differences between individual cells within its battery pack, identifying abnormal cells and issuing warnings accordingly. For example, the cloud can design voltage thresholds based on voltage variations of different individual cells, using expert experience and statistical algorithms, to identify abnormal cells whose voltages are outside the threshold range.

[0004] However, identifying abnormal individual batteries from a single dimension results in low accuracy, which in turn leads to low accuracy in vehicle warnings. Summary of the Invention

[0005] This application provides a vehicle warning method, device, equipment, and storage medium, which helps to improve the accuracy of vehicle warnings.

[0006] In a first aspect, this application provides a vehicle early warning method, which includes: acquiring a time series set of battery pack early warning parameters for each vehicle, wherein the time series set of early warning parameters includes a pressure difference time series set, a temperature difference time series set, and a pressure difference entropy time series set;

[0007] The feature vectors for each vehicle are determined based on the time series set of warning parameters. The feature vectors include pressure difference dimension components, temperature difference dimension components, and pressure difference entropy dimension components.

[0008] The feature vectors corresponding to each vehicle are divided into multiple sets, and each set includes the feature vectors corresponding to vehicles with the same battery pack type.

[0009] Based on the anomaly detection algorithm, abnormal vehicles are identified according to the feature vectors corresponding to vehicles in each set.

[0010] In one possible implementation, the timing set of battery pack warning parameters for each vehicle is obtained, including:

[0011] Acquire battery data for each vehicle's battery pack, including individual cell voltage and probe temperature;

[0012] The pressure difference value of the vehicle at each time moment is calculated based on the individual unit voltage, and the pressure difference time series set is obtained;

[0013] The temperature difference of the vehicle at each time point is calculated based on the probe temperature, and the temperature difference time series is obtained.

[0014] Based on the pressure difference values ​​at each time point, a pressure difference curve is plotted. Then, for the pressure difference curve, an over-coverage sliding operation is performed according to a preset time window length to calculate the pressure difference entropy within each window, thus obtaining a time series set of pressure difference entropy.

[0015] In one possible implementation, the feature vector corresponding to each vehicle is determined based on the time series set of warning parameters, including:

[0016] The feature vector is calculated from the time series set of warning parameters according to a preset algorithm. The preset algorithm includes any one of the following: the average value method per unit time and the median method per unit time.

[0017] In one possible implementation, based on an anomaly detection algorithm, abnormal vehicles are identified according to the feature vectors corresponding to vehicles in each set, including:

[0018] An outlier identification is performed on each set based on an anomaly detection algorithm. Outliers are the feature vectors corresponding to abnormal vehicles.

[0019] Output the abnormal vehicle identifiers corresponding to outliers.

[0020] In one possible implementation, the anomaly detection algorithm includes a density-based noise-applied spatial clustering algorithm, which identifies outliers in each set based on the anomaly detection algorithm, including:

[0021] Obtain multiple parameter sets, each parameter set including any neighborhood radius within a preset neighborhood radius range and any minimum number of points within a preset quantity range;

[0022] For each set, multiple parameter groups are input into density-based noise and spatial clustering algorithm for calculation to determine and store the abnormal feature vector and normal feature vector corresponding to each parameter group in the set.

[0023] Outliers are identified based on anomalous and normal feature vectors.

[0024] In one possible implementation, the anomaly detection algorithm includes the Isolation Forest algorithm, which identifies outliers in each set based on the anomaly detection algorithm, including:

[0025] For each set, multiple different abnormal data proportion values ​​within a preset proportion range are input into the Isolation Forest algorithm for calculation, and the abnormal feature vectors and normal feature vectors in the set are determined and stored.

[0026] Outliers are identified based on anomalous and normal feature vectors.

[0027] In one possible implementation, outliers are determined based on anomalous and normal feature vectors, including:

[0028] Calculate the distance between abnormal clusters and normal clusters to obtain distance parameters. Abnormal clusters include abnormal feature vectors, and normal clusters include normal feature vectors.

[0029] And calculate the radius of a normal cluster to obtain the radius parameter;

[0030] Calculate the ratio of the distance parameter to the radius parameter to obtain the distance factor;

[0031] The outlier is identified by the abnormal feature vector corresponding to the maximum distance factor.

[0032] Secondly, this application provides a vehicle warning device, which includes an acquisition module, a first determination module, a division module, and a second determination module, wherein...

[0033] The acquisition module is used to acquire the time series set of battery pack warning parameters for each vehicle. The time series set of warning parameters includes the time series set of differential pressure, the time series set of temperature difference, and the time series set of differential pressure entropy.

[0034] The first determination module is used to determine the feature vector corresponding to each vehicle based on the time series set of early warning parameters. The feature vector includes pressure difference dimension component, temperature difference dimension component and pressure difference entropy dimension component.

[0035] The partitioning module is used to divide the feature vectors corresponding to each vehicle into multiple sets, each set including the feature vectors corresponding to vehicles with the same battery pack type.

[0036] The second determination module is used to determine abnormal vehicles based on the feature vectors corresponding to vehicles in each set, using an anomaly detection algorithm.

[0037] In one possible implementation, the acquisition module is specifically used for:

[0038] Acquire battery data for each vehicle's battery pack, including individual cell voltage and probe temperature;

[0039] The pressure difference value of the vehicle at each time moment is calculated based on the individual unit voltage, and the pressure difference time series set is obtained;

[0040] The temperature difference of the vehicle at each time point is calculated based on the probe temperature, and the temperature difference time series is obtained.

[0041] Based on the pressure difference values ​​at each time point, a pressure difference curve is plotted. Then, for the pressure difference curve, an over-coverage sliding operation is performed according to a preset time window length to calculate the pressure difference entropy within each window, thus obtaining a time series set of pressure difference entropy.

[0042] In one possible implementation, the first determining module is specifically used for:

[0043] The feature vector is calculated from the time series set of warning parameters according to a preset algorithm. The preset algorithm includes any one of the following: the average value method per unit time and the median method per unit time.

[0044] In one possible implementation, the second determining module is specifically used for:

[0045] An outlier identification is performed on each set based on an anomaly detection algorithm. Outliers are the feature vectors corresponding to abnormal vehicles.

[0046] Output the abnormal vehicle identifiers corresponding to outliers.

[0047] In one possible implementation, the anomaly detection algorithm includes a density-based noise-applied spatial clustering algorithm, and a second determination module specifically used for:

[0048] Obtain multiple parameter sets, each parameter set including any neighborhood radius within a preset neighborhood radius range and any minimum number of points within a preset quantity range;

[0049] For each set, multiple parameter groups are input into density-based noise and spatial clustering algorithm for calculation to determine and store the abnormal feature vector and normal feature vector corresponding to each parameter group in the set.

[0050] Outliers are identified based on anomalous and normal feature vectors.

[0051] In one possible implementation, the anomaly detection algorithm includes the Isolation Forest algorithm, and the second determination module is specifically used for:

[0052] For each set, multiple different abnormal data proportion values ​​within a preset proportion range are input into the Isolation Forest algorithm for calculation, and the abnormal feature vectors and normal feature vectors in the set are determined and stored.

[0053] Outliers are identified based on anomalous and normal feature vectors.

[0054] In one possible implementation, the second determining module is specifically used for:

[0055] Calculate the distance between abnormal clusters and normal clusters to obtain distance parameters. Abnormal clusters include abnormal feature vectors, and normal clusters include normal feature vectors.

[0056] And calculate the radius of a normal cluster to obtain the radius parameter;

[0057] Calculate the ratio of the distance parameter to the radius parameter to obtain the distance factor;

[0058] The outlier is identified by the abnormal feature vector corresponding to the maximum distance factor.

[0059] Thirdly, this application provides an electronic device, including: a processor and a memory; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory, causing the processor to perform the vehicle warning method as described in the first aspect or any possible implementation of the first aspect.

[0060] Fourthly, this application provides a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, are used to implement the vehicle warning method as described in the first aspect or any possible implementation thereof.

[0061] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the vehicle warning method as described in the first aspect or any possible implementation thereof.

[0062] In this application, a time-series set of battery pack warning parameters for each vehicle is obtained. This time-series set includes a pressure difference time-series set, a temperature difference time-series set, and a pressure difference entropy time-series set. Based on the time-series set, a feature vector corresponding to each vehicle is determined. This feature vector includes pressure difference dimension components, temperature difference dimension components, and pressure difference entropy dimension components. The feature vectors corresponding to each vehicle are divided into multiple sets, each set including feature vectors corresponding to vehicles with the same battery pack type. Based on an anomaly detection algorithm, abnormal vehicles are identified according to the feature vectors corresponding to vehicles in each set. By combining the feature vectors representing the pressure difference, temperature difference, and pressure difference entropy of different vehicles, the anomaly of the battery packs of different vehicles is evaluated. This approach provides a more comprehensive consideration and helps improve the accuracy of abnormal battery pack identification, thereby improving the accuracy of vehicle warnings. Attached Figure Description

[0063] Figure 1 This is a schematic diagram illustrating the scenario to which the embodiments of this application apply;

[0064] Figure 2 A schematic flowchart illustrating a vehicle warning method provided in an embodiment of this application;

[0065] Figure 3 A flowchart illustrating yet another vehicle warning method provided in this application embodiment;

[0066] Figure 4 This is a schematic diagram of the structure of a vehicle warning device provided in an embodiment of this application;

[0067] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0068] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the embodiments of this application. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0069] In the embodiments of this application, terms such as "first" and "second" are used to distinguish identical or similar items with essentially the same function and purpose. For example, "first chip" and "second chip" are used only to distinguish different chips and do not limit their order of execution. Those skilled in the art will understand that terms such as "first" and "second" do not limit the quantity or execution order, and that "first" and "second" do not necessarily imply that they are different.

[0070] It should be noted that, in the embodiments of this application, the terms "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design scheme described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0071] In this application embodiment, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, ab, a--c, bc, or abc, where a, b, and c can be single or multiple.

[0072] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation portals are provided for users to choose to authorize or refuse.

[0073] It should be understood that although the steps in the flowcharts of this application's embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some of the steps in the figures may include at least one sub-step or at least one stage. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.

[0074] With the rapid development of power battery technology, new energy vehicles using lithium batteries as their power source have become an important direction for the green development and low-carbon transformation of the automotive industry. As the new energy vehicle market experiences explosive growth, the safety of these vehicles is also becoming an increasingly important issue for the industry and users.

[0075] As a core component of new energy vehicles, especially electric vehicles, the safety of the power battery directly determines the safety of the entire vehicle. Therefore, real-time fault warning and safety monitoring of the vehicle's power battery pack (hereinafter referred to as the battery pack) throughout its entire life cycle is of significant practical importance. Currently, a battery management system (BMS) is used to achieve real-time fault alarms based on vehicle-side monitoring signals. Vehicle-cloud collaboration is achieved using technologies such as the Internet and big data, enabling real-time fault warnings and safety monitoring of the battery in the cloud. For power batteries, the battery characteristic parameters monitored in the cloud can include signals such as individual cell voltage, probe temperature, and current.

[0076] The vehicle's battery pack is composed of multiple identical individual batteries connected in series and parallel. Individual batteries have certain inconsistencies in voltage, internal resistance, and capacity. As the number of charge and discharge cycles increases during use, the differences in internal parameters between individual batteries will gradually increase, resulting in some outliers in parameters such as voltage, voltage drop, voltage drop rate, temperature, and temperature difference of individual batteries.

[0077] In some implementations, each vehicle performs power battery anomaly detection based on the differences between individual cells within its own battery pack, identifying abnormal cells and issuing warnings accordingly. For example, the cloud-based system designs voltage thresholds based on voltage variations in different individual cells, using expert experience and statistical algorithms, to identify abnormal cells whose voltages are outside the threshold range.

[0078] However, in the early stages of the use of power batteries, the consistency of each individual battery cell is relatively good, meaning that the parameter differences between each individual battery cell are not significant. Therefore, the method of anomaly detection and analysis based on the differences between different individual batteries cells within the battery pack is prone to defining one or more normal individual batteries cells as abnormal individual batteries, that is, mistakenly identifying normal data as abnormal data.

[0079] Moreover, when using expert experience to design the parameter thresholds of the battery pack, the inconsistent characteristics of battery packs under different models and operating conditions make it impossible to achieve fine-grained differentiation in the design of the parameter thresholds.

[0080] Applying statistical algorithms to physical monitoring signals of power batteries, such as pressure difference and temperature difference, has certain limitations. Statistical algorithms can include methods such as standard deviation and box plots. The standard deviation method requires data to follow a normal distribution, but actual individual battery monitoring quantities such as pressure difference and temperature difference do not meet this assumption. Their data exhibit significant one-sidedness, skewed peaks, and long tails, leading to the failure of parameter threshold calculations for the battery pack. The box plot method requires threshold design based on expert experience and cannot be directly used for outlier identification, lacking support for outlier identification based on the inherent characteristics of the data.

[0081] Furthermore, regarding the inconsistent performance caused by cell aging in vehicle batteries, the characteristic parameters representing abnormalities are generally considered only from a single dimension such as voltage or temperature, without comprehensive judgment from multiple dimensions. Therefore, for example, assuming that the voltage of each individual cell in the vehicle battery pack is normal, but there is an individual cell with abnormal temperature, if the abnormality of the individual cell is judged from the voltage dimension, the abnormal individual cell cannot be detected, and the vehicle will not issue a warning. Therefore, identifying abnormal individual cells from a single dimension has low accuracy, which in turn leads to low accuracy of vehicle warnings.

[0082] In view of this, embodiments of this application provide a vehicle early warning method. This method calculates and analyzes the voltage consistency, voltage fluctuation, and temperature consistency of battery packs in different vehicles to identify abnormal battery packs and, consequently, abnormal vehicles. By comprehensively evaluating the anomalies of battery packs in different vehicles using multi-dimensional battery pack performance parameters, this method offers a more holistic approach, helping to improve the accuracy of abnormal battery pack identification and thus enhancing the accuracy of vehicle early warning.

[0083] For example, Figure 1 This is a schematic diagram illustrating the scenario to which the embodiments of this application apply. For example... Figure 1 As shown, the scenario includes a server 100 and multiple vehicles 200, where the server 100 can be a cloud server and the vehicles 200 can be, for example, electric vehicles equipped with power batteries.

[0084] Understandable Figure 1 The number of cars 200 in this example is merely illustrative. In actual application scenarios, the number of cars 200 can be greater, and this application embodiment does not impose a specific limitation on this.

[0085] Each of the multiple vehicles 200 can be equipped with a battery management system. Each battery management system is used to monitor the battery pack data of each vehicle 200 and upload the battery pack data to the server 100 in real time or periodically.

[0086] Server 100 can implement early warning of abnormal vehicles based on the vehicle early warning method provided in the embodiments of this application.

[0087] For example, the vehicle warning method may include: acquiring a time-series set of battery pack warning parameters for each vehicle, including a time-series set of differential pressure, a time-series set of temperature difference, and a time-series set of differential pressure entropy; determining a feature vector corresponding to each vehicle based on the time-series set of warning parameters, the feature vector including components of differential pressure, temperature difference, and differential pressure entropy; dividing the feature vectors corresponding to each vehicle into multiple sets, each set including feature vectors corresponding to vehicles of the same battery pack type; and determining abnormal vehicles based on an anomaly detection algorithm and the feature vectors corresponding to vehicles in each set. By combining the feature vectors representing differential pressure, temperature difference, and differential pressure entropy of battery packs from different vehicles, the method assesses the anomaly of battery packs in different vehicles more comprehensively, which helps improve the accuracy of abnormal battery pack identification and thus improves the accuracy of vehicle warnings.

[0088] The technical solutions shown in this application will now be described in detail through specific embodiments. It should be noted that the following embodiments may exist independently or in combination with each other; identical or identical content will not be repeated in different embodiments.

[0089] For example, Figure 2 A flowchart illustrating a vehicle warning method provided in an embodiment of this application is shown. The execution entity of this embodiment can be... Figure 1 The specific entity executing the action, such as server 100, can be determined based on the actual application scenario. Figure 2 As shown, the method may include:

[0090] S201: Obtain the time series set of battery pack warning parameters for each vehicle. The time series set of warning parameters includes the time series set of differential pressure, the time series set of temperature difference, and the time series set of differential pressure entropy.

[0091] Among them, the pressure difference in the pressure difference time series represents the maximum inconsistency of individual cells in the battery pack, and the pressure difference entropy in the pressure difference entropy time series represents the degree of disorder of this maximum inconsistency.

[0092] In this embodiment, the battery pack differential pressure time series set and differential pressure entropy time series set of each vehicle can be determined by the voltage of each individual cell in each battery pack and the timestamp corresponding to the voltage of each individual cell; the battery pack temperature difference time series set of each vehicle can be determined by the probe temperature of each individual cell in each battery pack and the timestamp corresponding to the probe temperature of each individual cell.

[0093] S202: Determine the feature vector corresponding to each vehicle based on the time series set of warning parameters. The feature vector includes pressure difference dimension component, temperature difference dimension component and pressure difference entropy dimension component.

[0094] The feature vector is used to characterize the overall features of the battery pack of the corresponding vehicle.

[0095] For example, the cloud server calculates each pressure difference in the time series set of battery pack pressure difference, each temperature difference in the time series set of temperature difference, and each pressure difference entropy in the time series set of pressure difference entropy for a single vehicle, respectively, to obtain the values ​​of the battery pack in the pressure difference dimension, temperature difference dimension, and pressure difference entropy dimension of the vehicle. These values ​​of pressure difference dimension, temperature difference dimension, and pressure difference entropy dimension are the pressure difference dimension component, temperature difference dimension component, and pressure difference entropy dimension component in the feature vector corresponding to the vehicle.

[0096] Understandably, the cloud server determines the feature vectors corresponding to each vehicle in the same way.

[0097] S203: Divide the feature vectors corresponding to each vehicle into multiple sets, each set including the feature vectors corresponding to vehicles with the same battery pack type.

[0098] For example, suppose S1, S2, S3...S k These are the feature vectors corresponding to vehicle 1, vehicle 2, vehicle 3... vehicle k. They are then divided into S1, S2, S3... S... according to a pre-defined partitioning logic. k The system is divided into multiple sets, with pre-defined partitioning logic including categories such as the same battery pack type, the same vehicle type, the same series-parallel connection type of the battery pack, vehicles from the same region, and vehicles from the same season. For example, vehicles can be partitioned into sets S1, S2, S3...S... based on their battery pack type. k The system is divided into multiple sets, each containing feature vectors corresponding to vehicles with the same battery pack type. Since vehicle battery data differs under different operating conditions, the battery data of vehicles under the same operating conditions is compared according to a pre-defined partitioning logic. This facilitates setting thresholds for the battery data of vehicles under the same operating conditions, thereby achieving accurate identification of battery pack anomalies.

[0099] Assuming k is 10, when classifying according to the same battery pack type, S1, S2, S3...S 10 It was divided into 3 sets: {S1, S2, S7} l{S3,S5,S} 10} l+1 {S4,S6,S8,S9} l+2 If the three feature vectors S1, S2, and S7 in the l-th set correspond to vehicles with the same battery pack type, then the three feature vectors S3, S5, and S7 in the (l+1)-th set correspond to vehicles with the same battery pack type. 10 The three feature vectors correspond to vehicles with the same battery pack type. Similarly, the four feature vectors S4, S6, S8, and S9 in the (l+2)th set correspond to vehicles with the same battery pack type. In other words, each set includes feature vectors corresponding to vehicles with the same battery pack type. Here, l is the set number.

[0100] S204: Based on the anomaly detection algorithm, anomaly vehicles are identified according to the feature vectors corresponding to vehicles in each set.

[0101] Anomaly detection algorithms can include K-nearest neighbor, density-based spatial clustering of applications with noise (DBSCAN), isolation forest (IForest), and local outlier factor (LOF) algorithms.

[0102] In a possible implementation, an anomaly detection algorithm is used to process the feature vectors corresponding to vehicles in each set to determine outlier feature vectors, thereby identifying the abnormal vehicles corresponding to the outlier feature vectors.

[0103] In this embodiment, a time-series set of battery pack warning parameters for each vehicle is obtained. This time-series set includes a pressure difference time-series set, a temperature difference time-series set, and a pressure difference entropy time-series set. Based on these time-series sets, a feature vector is determined for each vehicle. This feature vector includes pressure difference, temperature difference, and pressure difference entropy components. The feature vectors for each vehicle are divided into multiple sets, each set containing feature vectors for vehicles with the same battery pack type. Based on an anomaly detection algorithm, abnormal vehicles are identified according to the feature vectors for each vehicle in each set. By combining the feature vectors representing the pressure difference, temperature difference, and pressure difference entropy of different vehicle battery packs, the anomaly of different vehicle battery packs is evaluated more comprehensively, which helps improve the accuracy of abnormal battery pack identification and thus improves the accuracy of vehicle warnings.

[0104] Based on the above embodiments, in order to more clearly describe the technical solution of this application, please refer to the exemplary embodiments. Figure 3 , Figure 3This illustration shows a flowchart of yet another vehicle warning method provided in an embodiment of this application. The executing entity of this embodiment can be... Figure 1 The specific entity executing the action, such as server 100, can be determined based on the actual application scenario. Figure 3 As shown, the method may include:

[0105] S301: Obtain battery data for each vehicle's battery pack, including individual cell voltage and probe temperature; calculate the differential pressure value of the vehicle at each time step based on the individual cell voltage to obtain a differential pressure time series set; calculate the temperature difference value of the vehicle at each time step based on the probe temperature to obtain a temperature difference time series set; plot the differential pressure curve based on the differential pressure value at each time step, and perform non-overlapping sliding on the differential pressure curve according to a preset time window length to calculate the differential pressure entropy within each window to obtain a differential pressure entropy time series set.

[0106] In this embodiment, the cloud server can obtain battery pack data collected by the battery management system of each vehicle in real time, and clean the collected battery pack data. Cleaning may include steps such as removing outliers, processing missing values, and data format conversion. The battery pack data may include data such as the individual cell voltage and probe temperature of each battery cell.

[0107] The cloud server samples the cleaned battery pack data according to a preset sampling period to obtain the battery data of each vehicle's battery pack. The preset sampling period can be, for example, 10 seconds.

[0108] Understandably, the data in the battery pack data obtained by the cloud server from the battery management system of each vehicle carries a corresponding timestamp. Therefore, the battery data of each vehicle's battery pack also carries a corresponding timestamp.

[0109] In a possible implementation, the corresponding differential pressure time series set and differential pressure entropy time series set are determined based on the voltage of each individual cell and the corresponding timestamp, and the corresponding temperature difference time series set is determined based on the temperature of each probe and the corresponding timestamp.

[0110] For example, at time t, the differential pressure value ΔU of the battery pack t The maximum single-cell voltage U of the battery pack max and the smallest unit voltage U min The difference, the pressure difference ΔU at each time point. t Composition of differential pressure time series {ΔU t}; Based on the pressure difference ΔU at each moment t Plot the pressure differential curve, and for the pressure differential curve, perform an uncovered sliding motion according to a preset time window length, and calculate the pressure differential entropy within each window i.

[0111]

[0112] Among them, P ij The pressure difference value ΔU t The probability values ​​of different voltage intervals within the i-th sliding window, where there are N voltage intervals, and N is a positive integer.

[0113]

[0114] Among them, C ij The pressure difference value ΔU t The number of samples falling into pressure difference interval j within the i-th sliding window.

[0115] Temperature difference ΔT of the battery pack t The maximum probe temperature T of the battery pack max and minimum probe temperature T min The difference, the temperature difference ΔT at each time point. t Composition of temperature difference time series set {ΔT t}

[0116] In this embodiment, the abnormal characteristics of the battery pack are characterized from multiple dimensions such as pressure difference, temperature difference, and pressure difference entropy. Compared with a single dimension, this can more richly reflect the characteristics of the battery pack, which helps to improve the identification accuracy of abnormal battery packs, thereby improving the warning accuracy of vehicles corresponding to abnormal battery packs.

[0117] S302: Calculate the feature vector of the time series set of warning parameters according to the preset algorithm. The preset algorithm includes any one of the following: the method of calculating the average value per unit time and the method of calculating the median per unit time.

[0118] In a possible implementation, for each vehicle's battery pack, the unit time characteristics of each dimension are calculated according to a preset algorithm to obtain the pressure difference dimension component, temperature difference dimension component, and pressure difference entropy dimension component.

[0119] For example, assuming the preset algorithm is the unit time average method, and the unit time is a day, for the battery pack of vehicle k, the average of all pressure differences corresponding to a certain day in the pressure difference time series is calculated to obtain the pressure difference dimension component v1; the average of all temperature differences corresponding to that day in the temperature difference time series is calculated to obtain the temperature difference dimension component v2; and the average of all pressure difference entropies corresponding to that day in the pressure difference time series is calculated to obtain the pressure difference entropy dimension component v3. Then the feature vector S corresponding to vehicle k is... k = [v1, v2, v3].

[0120] S303: Divide the feature vectors corresponding to each vehicle into multiple sets, each set including the feature vectors corresponding to vehicles with the same battery pack type.

[0121] This step is similar to or the same as step S203 above, and will not be repeated here.

[0122] S304: Based on the anomaly detection algorithm, outlier identification is performed on each set. Outliers are the feature vectors corresponding to abnormal vehicles; the abnormal vehicle identifiers corresponding to the outliers are output.

[0123] For example, the feature vector S k = [v1, v2, v3] can be understood as coordinate points in three-dimensional space. Based on the anomaly detection algorithm, outlier points are identified in each set. That is, these coordinate points are labeled by the anomaly detection algorithm. The labels can include anomaly points and normal points. Anomaly points are outliers, which are the feature vectors corresponding to abnormal vehicles. After identifying outliers, the abnormal vehicle identifier corresponding to the outlier is output for subsequent vehicle battery management system warnings.

[0124] In this embodiment, the feature vectors representing the pressure difference, temperature difference, and pressure difference entropy of battery packs from different vehicles are combined to evaluate the anomalies of battery packs in different vehicles. This approach is more comprehensive and helps to improve the accuracy of identifying abnormal battery packs, thereby improving the accuracy of vehicle warnings.

[0125] The following section introduces the specific implementation of different anomaly detection algorithms for identifying outliers.

[0126] In one possible implementation, the anomaly detection algorithm includes a density-based noise-applied spatial clustering algorithm, which identifies outliers in each set based on the anomaly detection algorithm, including:

[0127] Multiple parameter sets are obtained, each parameter set including any neighborhood radius within a preset neighborhood radius range and any minimum number of points within a preset number range; for each set, the multiple parameter sets are respectively input into a density-based noise application spatial clustering algorithm for calculation, and the abnormal feature vector and normal feature vector corresponding to each parameter set in the set are determined and stored; outliers are determined based on the abnormal feature vector and normal feature vector.

[0128] The preset neighborhood radius range is the range of values ​​for the neighborhood radius, and the preset number range is the range of values ​​for the minimum number of points. The preset neighborhood radius range and the preset number range can be set according to the actual application scenario, and this application embodiment does not make specific limitations on them.

[0129] In a possible implementation, the input parameters for a density-based noise-based spatial clustering algorithm are a parameter set, which may include the neighborhood radius and the minimum number of points. The cloud server can store preset neighborhood radius ranges and preset number ranges. By taking any neighborhood radius from the preset neighborhood radius range and any minimum number of points from the preset number range as a parameter set, multiple parameter sets can be obtained.

[0130] For each set, the multiple parameter groups are input into the density-based noise application spatial clustering algorithm for calculation, and the abnormal feature vector and normal feature vector corresponding to each parameter group in the set are obtained.

[0131] For example, assuming a preset neighborhood radius includes M possible values ​​for the neighborhood radius, and a preset number of points includes N possible values ​​for the minimum number of points, then the density-based noise-based spatial clustering algorithm has M×N parameter sets. Each parameter set is input into the density-based noise-based spatial clustering algorithm to calculate for each set, obtaining the corresponding abnormal feature vector and normal feature vector for each parameter set in the set. Anomaly labels are then assigned to the abnormal feature vectors corresponding to each parameter set, and normal labels are assigned to the normal feature vectors. For example, for sets L and L'... +1 The calculation involves inputting M×N parameter sets into density-based noise and applying a spatial clustering algorithm to set L. This yields the corresponding anomalous and normal feature vectors for each parameter set within set L, resulting in M×N sets of anomalous and normal feature vectors. +1 Calculations are performed to obtain the parameter sets in set L. +1 Each of these features corresponds to an abnormal feature vector and a normal feature vector, resulting in M×N sets of abnormal feature vectors and normal feature vectors.

[0132] It is understandable that each parameter group corresponds to a set of abnormal feature vectors and normal feature vectors. Different parameter groups correspond to different abnormal feature vectors, and different parameter groups correspond to different normal feature vectors.

[0133] Furthermore, by combining the abnormal and normal feature vectors corresponding to the parameter sets with different neighborhood radii and minimum number of points in set L, outliers in set L are determined.

[0134] For example, for each set of parameters, the quality of the abnormal feature vectors and normal feature vectors can be evaluated by numerical values ​​such as distance factor, variance or standard deviation. The abnormal feature vectors in the set of abnormal feature vectors and normal feature vectors that are evaluated as the best are the outliers in the set L.

[0135] In this embodiment, the DBSCAN algorithm is used to identify the feature vector corresponding to the battery pack of an abnormal vehicle, avoiding a single reliance on expert experience and without too many restrictions on the distribution of battery data.

[0136] In one possible implementation, the anomaly detection algorithm includes the Isolation Forest algorithm, which identifies outliers in each set based on the anomaly detection algorithm, including:

[0137] For each set, multiple different outlier data proportions within a preset proportion range are input into the Isolation Forest algorithm for calculation, and the outlier feature vectors and normal feature vectors in the set are determined and stored; outliers are determined based on the outlier feature vectors and normal feature vectors.

[0138] The preset percentage range refers to the reasonable range of values ​​for the percentage of outlier data. This preset range can be configured on the cloud server; for example, it can be between 0 and 0.3. In some possible implementations, when configuring the preset range for the percentage of outlier data on the cloud server, the range can be narrowed based on the actual scenario. This reduces the number of times the outlier data percentage is input into the Isolation Forest algorithm for calculation, thereby quickly determining the preferred percentage of outlier data. Based on this preferred percentage, the corresponding outlier feature vector and normal feature vector are obtained to accurately identify outliers.

[0139] For example, for each set, such as set L, different abnormal data proportion values ​​are input into the Isolation Forest algorithm to calculate set L, and the abnormal feature vector and normal feature vector corresponding to each abnormal data proportion value are obtained. The abnormal feature vector corresponding to each abnormal data proportion value is labeled with an abnormal label, and the normal feature vector is labeled with a normal label.

[0140] It is understandable that each percentage of abnormal data corresponds to a set of abnormal feature vectors and normal feature vectors. Different percentages of abnormal data correspond to different abnormal feature vectors, and different percentages of abnormal data also correspond to different normal feature vectors.

[0141] Furthermore, by combining the abnormal feature vectors and normal feature vectors corresponding to different proportions of abnormal data in set L, outliers in the set are identified.

[0142] For example, for each abnormal data percentage, the abnormal feature vector and normal feature vector can be evaluated by numerical values ​​such as distance factor, variance or standard deviation. The abnormal feature vector in the set of abnormal feature vectors and normal feature vectors that are evaluated as the best is the outlier in the set L.

[0143] In this embodiment, the feature vectors corresponding to the battery packs of abnormal vehicles are identified by the isolated forest algorithm, which avoids a single reliance on expert experience and does not impose many restrictions on the distribution of battery data.

[0144] In one possible implementation, outliers are determined based on anomalous and normal feature vectors, including:

[0145] Calculate the distance between the abnormal cluster and the normal cluster to obtain the distance parameter. The abnormal cluster includes abnormal feature vectors, and the normal cluster includes normal feature vectors. Calculate the radius of the normal cluster to obtain the radius parameter. Calculate the ratio of the distance parameter to the radius parameter to obtain the distance factor. Determine the abnormal feature vector corresponding to the maximum distance factor as the outlier.

[0146] The distance factor is used to evaluate the quality of the anomaly detection algorithm's calculation results. The larger the distance factor, the better the anomaly detection algorithm's calculation results. The distance factor can be a floating-point number greater than 0.

[0147] For example, suppose that after partitioning according to a preset partitioning logic, we get a set {S1, S2, ..., S}. 10 Furthermore, assuming that after calculating the set using an anomaly detection algorithm with the first set of input parameters, the resulting anomaly feature vectors are S1 and S2, and the normal feature vectors are S3 to S4, the results are as follows: 10 Then, the abnormal feature vectors S1 and S2 corresponding to the first parameter group form an abnormal cluster, and the normal feature vectors S3 to S... 10 Form a normal cluster, calculate the distance between the center point of the normal cluster and the center point of the abnormal cluster to obtain the distance parameter, where the distance can be Euclidean distance; calculate the radius of the normal cluster to obtain the radius parameter; calculate the ratio of the distance parameter and the radius parameter to obtain the distance factor when the input parameters of the anomaly detection algorithm are the first parameter group.

[0148] Suppose that after calculating the set using an anomaly detection algorithm with the second set of input parameters, the resulting anomaly feature vectors are S1, S2, and S8, and the normal feature vectors are S3 to S7 and S9 to S8. 10 Then, the abnormal feature vectors S1, S2, and S8 corresponding to the second parameter group form an abnormal cluster, while the normal feature vectors S3 to S7 and S9 to S8 form an abnormal cluster. 10 Form a normal cluster, calculate the distance between the center point of the normal cluster and the center point of the abnormal cluster to obtain the distance parameter; calculate the radius of the normal cluster to obtain the radius parameter; calculate the ratio of the distance parameter and the radius parameter to obtain the distance factor when the input parameters of the anomaly detection algorithm are the second parameter group.

[0149] Similarly, when the input parameters are the third parameter group, the fourth parameter group, ... the Mth parameter group, the distance factors can be obtained based on the corresponding abnormal feature vectors and normal feature vectors, respectively.

[0150] It is understandable that the first parameter group, the second parameter group, ... the Mth parameter group are all within the input parameter threshold range of the anomaly detection algorithm.

[0151] Furthermore, the parameter set corresponding to the largest distance factor among the obtained distance factors is determined as the preferred input parameter of the anomaly detection algorithm. This preferred input parameter is stored, and the anomaly feature vector obtained by the anomaly detection algorithm after calculating this set is the outlier point.

[0152] In this embodiment, the input parameters of the anomaly detection algorithm are dynamically searched and optimized based on the distance factor used to determine the results of the anomaly detection algorithm. This achieves dynamic search of the anomaly judgment threshold, which helps improve the accuracy of anomaly vehicle identification. It meets the different requirements of different battery pack models and different operating conditions for the input parameters of the anomaly detection algorithm, improves the robustness of the method, and identifies anomaly vehicles based on the vehicle's own battery data, avoiding a single reliance on expert experience.

[0153] Figure 4 This is a schematic diagram of the structure of a vehicle warning device provided in an embodiment of this application, as shown below. Figure 4 As shown, the vehicle warning device 40 includes: an acquisition module 401, a first determination module 402, a division module 403, and a second determination module 404, wherein,

[0154] The acquisition module 401 is used to acquire the time series set of battery pack warning parameters for each vehicle. The time series set of warning parameters includes the time series set of differential pressure, the time series set of temperature difference, and the time series set of differential pressure entropy.

[0155] The first determining module 402 is used to determine the feature vector corresponding to each vehicle based on the time series set of early warning parameters. The feature vector includes pressure difference dimension component, temperature difference dimension component and pressure difference entropy dimension component.

[0156] The partitioning module 403 is used to divide the feature vectors corresponding to each vehicle into multiple sets, each set including the feature vectors corresponding to vehicles with the same battery pack type.

[0157] The second determination module 404 is used to determine abnormal vehicles based on the feature vectors corresponding to vehicles in each set, using an anomaly detection algorithm.

[0158] In one possible implementation, module 401 is specifically used for:

[0159] Acquire battery data for each vehicle's battery pack, including individual cell voltage and probe temperature;

[0160] The pressure difference value of the vehicle at each time moment is calculated based on the individual unit voltage, and the pressure difference time series set is obtained;

[0161] The temperature difference of the vehicle at each time point is calculated based on the probe temperature, and the temperature difference time series is obtained.

[0162] Based on the pressure difference values ​​at each time point, a pressure difference curve is plotted. Then, for the pressure difference curve, an over-coverage sliding operation is performed according to a preset time window length to calculate the pressure difference entropy within each window, thus obtaining a time series set of pressure difference entropy.

[0163] In one possible implementation, the first determining module 402 is specifically used for:

[0164] The feature vector is calculated from the time series set of warning parameters according to a preset algorithm. The preset algorithm includes any one of the following: the average value method per unit time and the median method per unit time.

[0165] In one possible implementation, the second determining module 404 is specifically used for:

[0166] An outlier identification is performed on each set based on an anomaly detection algorithm. Outliers are the feature vectors corresponding to abnormal vehicles.

[0167] Output the abnormal vehicle identifiers corresponding to outliers.

[0168] In one possible implementation, the anomaly detection algorithm includes a density-based noise-applied spatial clustering algorithm, and a second determination module 404, specifically used for:

[0169] Obtain multiple parameter sets, each parameter set including any neighborhood radius within a preset neighborhood radius range and any minimum number of points within a preset quantity range;

[0170] For each set, multiple parameter groups are input into density-based noise and spatial clustering algorithm for calculation to determine and store the abnormal feature vector and normal feature vector corresponding to each parameter group in the set.

[0171] Outliers are identified based on anomalous and normal feature vectors.

[0172] In one possible implementation, the anomaly detection algorithm includes the Isolation Forest algorithm, and the second determination module 404 is specifically used for:

[0173] For each set, multiple different abnormal data proportion values ​​within a preset proportion range are input into the Isolation Forest algorithm for calculation, and the abnormal feature vectors and normal feature vectors in the set are determined and stored.

[0174] Outliers are identified based on anomalous and normal feature vectors.

[0175] In one possible implementation, the second determining module 404 is specifically used for:

[0176] Calculate the distance between abnormal clusters and normal clusters to obtain distance parameters. Abnormal clusters include abnormal feature vectors, and normal clusters include normal feature vectors.

[0177] And calculate the radius of a normal cluster to obtain the radius parameter;

[0178] Calculate the ratio of the distance parameter to the radius parameter to obtain the distance factor;

[0179] The outlier is identified by the abnormal feature vector corresponding to the maximum distance factor. The vehicle warning device 40 provided in this application embodiment can execute the technical solution shown in the above-described vehicle warning method embodiment; its implementation principle and beneficial effects are similar and will not be repeated here.

[0180] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Please refer to... Figure 5 The electronic device 50 includes a memory 501, a processor 502, a communication component 503, and a bus 504. The memory 501, processor 502, and communication component 503 are interconnected via the bus 504.

[0181] Memory 501 stores computer-executed instructions;

[0182] The processor 502 executes the computer execution instructions stored in the memory 501, causing the processor 502 to execute the above-mentioned vehicle warning method;

[0183] The communication component 503 can be applied to, but is not limited to, transceiver devices such as transceivers, to enable communication between the electronic device 50 and other devices or communication networks;

[0184] Bus 504 may include a pathway for transmitting information between various components of electronic device 50 (e.g., memory 501, processor 502, communication component 503).

[0185] Electronic devices 50 can be chips, modules, integrated development environments (IDEs), etc.

[0186] Figure 5 The electronic device 50 shown in the embodiment can execute the technical solution shown in the above vehicle warning method embodiment. Its implementation principle and beneficial effects are similar, and will not be described again here.

[0187] This application also provides a computer-readable storage medium storing computer-executable instructions, which are used to implement the above-described vehicle warning method when executed by a processor.

[0188] This application also provides a computer program product, including a computer program that, when executed by a processor, can implement the above-described vehicle warning method.

[0189] The computer-readable storage medium and computer program product of this application embodiment can execute the above-described vehicle warning method. The specific implementation process and beneficial effects are described above and will not be repeated here.

[0190] All or part of the steps in the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a readable memory. When the program is executed, it performs the steps of the above-described method embodiments; and the aforementioned memory (storage medium) includes: read-only memory (ROM), random access memory (RAM), flash memory, hard disk, solid-state drive, magnetic tape, floppy disk, optical disc, and any combination thereof.

[0191] This application describes embodiments of methods, apparatus (systems), and computer program products according to flowchart illustrations and / or block diagrams. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processing unit of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processing unit of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0192] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0193] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0194] Obviously, those skilled in the art can make various modifications and variations to the embodiments of this application without departing from the spirit and scope of this application. Therefore, if these modifications and variations to the embodiments of this application fall within the scope of the claims of this application and their equivalents, this application also intends to include these modifications and variations.

Claims

1. A vehicle early warning method, characterized in that, include: Obtain the time series set of battery pack warning parameters for each vehicle. The time series set of warning parameters includes a pressure difference time series set, a temperature difference time series set, and a pressure difference entropy time series set. Among them, the pressure difference in the pressure difference time series set represents the maximum inconsistency of individual cells in the battery pack, and the pressure difference entropy in the pressure difference entropy time series set represents the degree of disorder of the maximum inconsistency. The feature vector corresponding to each vehicle is determined based on the time series set of the warning parameters. The feature vector includes pressure difference dimension component, temperature difference dimension component and pressure difference entropy dimension component. According to the preset partitioning logic, the battery data of vehicles under the same working conditions are compared, and the feature vectors corresponding to each vehicle are divided into multiple sets. Each set includes the feature vectors corresponding to vehicles of the same type. The preset partitioning logic includes the same battery pack type, the same vehicle type, the same battery pack series-parallel connection type, vehicles from the same region, or vehicles from the same season. Based on the anomaly detection algorithm, the abnormal feature vectors and normal feature vectors within the set are determined, and outliers are determined based on the abnormal feature vectors and normal feature vectors. The outliers are the feature vectors corresponding to the abnormal vehicles. Output the abnormal vehicle identifiers corresponding to the outliers; The anomaly detection algorithm includes a density-based noise-applied spatial clustering algorithm. The anomaly detection algorithm determines anomalous and normal feature vectors within the set, and identifies outliers based on the anomalous and normal feature vectors, including: Multiple parameter groups are obtained, and the number of each parameter group is the product of the neighborhood radius of M values ​​within a preset neighborhood radius range and the minimum number of points of N values ​​within a preset number range. For each set, the multiple parameter groups are respectively input into the density-based noise applied spatial clustering algorithm for calculation, and the M×N sets of abnormal feature vectors and normal feature vectors of each parameter group within their respective sets are determined and stored; Calculate the distance between the abnormal cluster and the normal cluster to obtain the distance parameter. The abnormal cluster includes the abnormal feature vector, and the normal cluster includes the normal feature vector. And calculate the radius of the normal cluster to obtain the radius parameter; Calculate the ratio of the distance parameter to the radius parameter to obtain the distance factor; The outlier in the set is determined by identifying the abnormal feature vector corresponding to the maximum distance factor.

2. The method according to claim 1, characterized in that, The acquisition of the time series set of battery pack warning parameters for each vehicle includes: Obtain battery data for each vehicle's battery pack, including individual cell voltage and probe temperature; The pressure difference value of the vehicle at each time moment is calculated based on the individual unit voltage to obtain the pressure difference time series set; The temperature difference value of the vehicle at each time moment is calculated based on the probe temperature to obtain the temperature difference time series set; Based on the pressure difference value at each time point, a pressure difference curve is plotted. Then, for the pressure difference curve, an uncovered sliding motion is performed according to a preset time window length to calculate the pressure difference entropy within each window, thus obtaining the time series set of pressure difference entropy.

3. The method according to claim 2, characterized in that, The feature vector corresponding to each vehicle is determined based on the time series set of warning parameters, including: The feature vector is calculated from the time series set of the warning parameters according to a preset algorithm, wherein the preset algorithm includes any one of the following: the average value method per unit time, and the median method per unit time.

4. The method according to claim 1, characterized in that, The anomaly detection algorithm includes the Isolation Forest algorithm. The step of determining anomalous and normal feature vectors in the set based on the anomaly detection algorithm, and determining outliers based on the anomalous and normal feature vectors, includes: For each set, multiple different abnormal data proportion values ​​within a preset proportion range are input into the isolated forest algorithm for calculation, and the abnormal feature vectors and normal feature vectors in the set are determined and stored. Outliers are determined based on the anomalous feature vector and the normal feature vector.

5. A vehicle warning device, characterized in that, The device includes an acquisition module, a first determination module, a division module, and a second determination module, wherein... The acquisition module is used to acquire the time series set of battery pack warning parameters for each vehicle. The time series set of warning parameters includes a pressure difference time series set, a temperature difference time series set, and a pressure difference entropy time series set. The first determining module is used to determine the feature vector corresponding to each vehicle based on the time series set of the early warning parameters. The feature vector includes a pressure difference dimension component, a temperature difference dimension component, and a pressure difference entropy dimension component. The partitioning module is used to compare the battery data of vehicles under the same working conditions according to a preset partitioning logic, and to divide the feature vectors corresponding to each vehicle into multiple sets. Each set includes the feature vectors corresponding to vehicles of the same type. The preset partitioning logic includes the same battery pack type, the same vehicle type, the same battery pack series-parallel connection type, vehicles in the same region, or vehicles in the same season. The second determining module is used to determine the abnormal feature vectors and normal feature vectors in the set based on the anomaly detection algorithm, determine outliers based on the abnormal feature vectors and normal feature vectors, wherein the outlier is the feature vector corresponding to the abnormal vehicle; and output the abnormal vehicle identifier corresponding to the outlier. The anomaly detection algorithm includes a density-based noise-applied spatial clustering algorithm. The second determining module is used to acquire multiple parameter groups, where the number of each parameter group is the product of the neighborhood radius of M values ​​within a preset neighborhood radius range and the minimum number of points with N values ​​within a preset number range. For each set, the multiple parameter groups are respectively input into the density-based noise-applied spatial clustering algorithm for calculation, determining and storing M×N sets of abnormal feature vectors and normal feature vectors within each parameter group within the set; calculating the distance between abnormal clusters and normal clusters to obtain a distance parameter, where the abnormal cluster includes the abnormal feature vectors and the normal cluster includes the normal feature vectors; calculating the radius of the normal cluster to obtain a radius parameter; calculating the ratio of the distance parameter to the radius parameter to obtain a distance factor; and determining the abnormal feature vector corresponding to the maximum distance factor as the outlier in the set.

6. An electronic device, characterized in that, include: Processor, memory; The memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1-4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method described in any one of claims 1-4.

Images