Method for constructing battery fault detection model and battery fault detection method

Abstract

The invention discloses a method for constructing a battery fault detection model and a battery fault detection method. The method for constructing the battery fault detection model comprises the following steps: acquiring historical operation data of a battery; calculating basic index data of the battery according to the historical operation data; performing signal processing based on the basic index data to obtain a low-frequency approximate signal for representing the intrinsic state change of the battery; based on the low-frequency approximate signal, generating reference feature data through an automatic feature generation method; and determining model parameters based on the reference feature data to obtain a battery fault detection model. According to the method for constructing the battery fault detection model and the battery fault detection method provided by the invention, whether the state of the battery has a fault can be detected on the basis of the historical operation data of the battery, and the dangerous condition of thermal runaway of the battery can be avoided.

Description

Technical Field

[0001] The present application relates to the technical field of battery fault diagnosis, and in particular to a method for constructing a battery fault detection model and a battery fault detection method. Background Art

[0002] As a core component of electric vehicles, power batteries are typically composed of multiple single cells connected in series and parallel, providing sufficient power for the vehicle. Therefore, the battery's condition directly determines the overall performance of the electric vehicle. However, during actual operation, battery performance and safety gradually deteriorate with age and changes in operating conditions. This can lead to various faults, including reduced capacity, reduced power, slower charging rates, and poor consistency. In extreme cases, thermal runaway can occur, resulting in serious safety incidents. Therefore, accurate detection of potential battery faults is crucial, enabling customers to implement timely countermeasures and avoid dangerous situations. Summary of the Invention

[0003] The purpose of this application is to provide a method for constructing a battery fault detection model and a battery fault detection method, which can detect whether there is a fault in the battery state based on the historical operating data of the battery, and can avoid the dangerous situation of battery thermal runaway.

[0004] To achieve the above objectives: In a first aspect, an embodiment of the present application provides a method for constructing a battery fault detection model, comprising: Obtain historical operating data of the battery; Calculating basic indicator data of the battery based on the historical operating data; Performing signal processing based on the basic indicator data to obtain a low-frequency approximate signal for characterizing a change in an intrinsic state of the battery; Based on the low-frequency approximate signal, generating reference feature data by an automatic feature generation method; Based on the reference characteristic data, model parameters are determined to obtain a battery fault detection model.

[0005] In one embodiment, obtaining historical operating data of the battery includes: Obtaining historical operating data of the battery collected by the battery management system through the vehicle-mounted TBOX; and / or Obtaining the historical operating data of the battery sent by the charger through the station control terminal of the battery swap station; and / or When a faulty vehicle is repaired through after-sales service, the historical operating data of the battery is obtained.

[0006] In one embodiment, the calculating basic indicator data of the battery based on the historical operating data includes: Performing data cleaning on the historical operation data; Based on the historical operating data after cleaning, basic indicator data of the battery is calculated.

[0007] In one embodiment, the generating of reference feature data based on the low-frequency approximation signal by an automatic feature generation method includes: Performing labeling processing on the historical operation data to associate fault data in the historical operation data with preset fault type labels; Based on the fault type label associated with the fault data, obtaining the fault type label of the fault data corresponding to the low-frequency approximation signal; Based on the low-frequency approximation signal and the corresponding fault type label, reference feature data associated with the corresponding fault type label is generated through an automatic feature generation method.

[0008] In one embodiment, determining model parameters based on the reference characteristic data to obtain a battery fault detection model includes: Performing feature dimensionality reduction on the reference feature data; Obtaining a battery fault detection result based on the reference feature data after feature dimensionality reduction processing; adjusting model parameters according to the battery fault detection result and the fault type label associated with the reference feature data; When the battery fault detection result matches the fault type corresponding to the fault type label, the corresponding model parameters are saved to obtain the battery fault detection model.

[0009] In one embodiment, the basic indicator data of the battery to be tested includes at least one of internal resistance, capacity, static pressure difference, maximum temperature difference and charged capacity.

[0010] In one embodiment, the signal processing method based on the basic indicator data includes at least one of variational mode decomposition, empirical mode decomposition, wavelet decomposition and wavelet packet decomposition.

[0011] In a second aspect, an embodiment of the present application provides a battery fault detection method, which uses a battery fault detection model constructed by any of the methods described above, and the method includes: Obtain historical operating data of the battery under test; Inputting the historical operating data into the battery fault detection model to perform fault detection to obtain a fault detection result of the battery to be tested; The fault detection result is output.

[0012] In a third aspect, an embodiment of the present application provides a battery fault detection system, comprising a processor and a memory storing a computer program, wherein when the processor runs the computer program, the steps of any of the above methods are implemented.

[0013] In a fourth aspect, an embodiment of the present application provides a readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described in any one of the above items are implemented.

[0014] The method for constructing a battery fault detection model provided in an embodiment of the present application includes: obtaining historical operating data of the battery; calculating basic indicator data of the battery based on the historical operating data; performing signal processing based on the basic indicator data to obtain a low-frequency approximate signal used to characterize the change in the intrinsic state of the battery; generating reference feature data based on the low-frequency approximate signal through an automatic feature generation method; and determining model parameters based on the reference feature data to obtain a battery fault detection model. In this way, by analyzing the low-frequency approximate signal, the battery's usage status can be more realistically reflected, reducing interference from other factors. In addition, combined with the automatic feature generation method, the battery's gradual change features can be automatically generated, increasing the coverage of feature recognition and reducing the model construction cycle.

[0015] The battery fault detection method provided in an embodiment of the present application includes: obtaining historical operating data of a battery under test; inputting the historical operating data into a battery fault detection model to perform fault detection to obtain a fault detection result for the battery under test; and outputting the fault detection result. In this way, based on the battery's historical operating data, it is possible to quickly detect whether the battery is faulty, thereby avoiding the dangerous situation of battery thermal runaway. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without creative work.

[0017] Figure 1 A flowchart of a method for constructing a battery fault detection model provided in an embodiment of the present application.

[0018] Figure 2 Source diagram of historical operation data provided in this application embodiment Figure 1 .

[0019] Figure 3 Source diagram of historical operation data provided in this application embodiment Figure 2 .

[0020] Figure 4 A flowchart of a battery fault detection method provided in an embodiment of the present application.

[0021] Figure 5 A schematic diagram of the trend of battery internal resistance changing over time provided in an embodiment of the present application.

[0022] Figure 6 A schematic diagram of a low-frequency approximate signal of a battery internal resistance provided in an embodiment of the present application.

[0023] Figure 7 A characteristic diagram showing the internal resistance of a battery provided in an embodiment of the present application. DETAILED DESCRIPTION

[0024] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

[0025] It should be noted that, in this document, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprising a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element. In addition, components, features, and elements with the same name in different embodiments of the present application may have the same meaning or different meanings, and their specific meanings need to be determined by their explanation in the specific embodiment or further combined with the context of the specific embodiment.

[0026] It should be understood that although the terms first, second, third, etc. may be used herein to describe various information, such information should not be limited to these terms. These terms are merely used to distinguish information of the same type from one another. For example, without departing from the scope of this document, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the term "if" as used herein may be interpreted as "at the time of," "when," or "in response to a determination." Furthermore, as used herein, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context indicates otherwise. It should be further understood that the terms "comprising" and "including" indicate the presence of the described features, steps, operations, elements, components, items, types, and / or groups, but do not exclude the presence, occurrence, or addition of one or more other features, steps, operations, elements, components, items, types, and / or groups. The terms "or" and "and / or" as used herein are to be interpreted as inclusive, meaning any one or any combination. Thus, “A, B, or C” or “A, B, and / or C” means “any of: A; B; C; A and B; A and C; B and C; A, B, and C.” An exception to this definition occurs only when a combination of elements, functions, steps, or operations are inherently mutually exclusive in some manner.

[0027] It should be understood that, although the various steps in the flowchart in the embodiment of the present application are shown in sequence according to the indication of the arrows, these steps are not necessarily performed in sequence in the order indicated by the arrows. Unless clearly stated herein, the execution of these steps is not strictly limited in order, and they can be performed in other orders. Moreover, at least a portion of the steps in the figure may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and their execution order is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0028] It should be noted that in this article, step codes such as S101 and S102 are used for the purpose of expressing the corresponding content more clearly and concisely, and do not constitute a substantial limitation on the order. When implementing the step, those skilled in the art may execute S102 first and then S101, etc., but these should all be within the scope of protection of this application.

[0029] It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application.

[0030] In the subsequent description, the use of suffixes such as "module", "component" or "unit" to represent elements is only for the purpose of facilitating the description of the present application and has no specific meaning. Therefore, "module", "component" or "unit" can be used interchangeably.

[0031] See Figure 1 , is a method for constructing a battery fault detection model provided in an embodiment of the present application, the method specifically comprising the following steps: Step S101: Acquire historical operating data of the battery.

[0032] Specifically, in the embodiments of the present application, historical battery operation data within a preset time period is acquired to provide training data for the battery fault detection model. The acquired historical battery operation data includes both normal data and fault data during battery operation, including but not limited to parameters such as voltage, current, temperature, charge cycle, discharge cycle, and state of charge during battery operation.

[0033] In one embodiment, obtaining the historical operating data of the battery includes: obtaining the historical operating data of the battery collected by the battery management system through the vehicle-mounted TBOX. Figure 2 As shown in the figure, when the battery is installed on the vehicle, the battery operation data is generally collected through the battery management system (BMS) to monitor the battery health status. When it is necessary to collect the historical operation data of the battery, the battery operation data collected by the BMS can be uploaded to the cloud through the on-board TBOX using the corresponding communication protocol (such as GB / T 32960.3-2016 protocol) to complete the collection of the battery's historical operation data.

[0034] In another embodiment, obtaining the historical operating data of the battery further includes: obtaining the historical operating data of the battery sent by the charger through the station control terminal of the battery swap station. Figure 3 As shown, when the vehicle is a battery-swapping vehicle and the battery is being charged in a battery-swapping station, data can be exchanged between the charger in the battery-swapping station and the BMS corresponding to the rechargeable battery. The battery operation data collected by the BMS is aggregated to the station control end through the charger, and then the station control end uploads the battery's historical operation data to the cloud, completing the collection of the battery's historical operation data.

[0035] In some other embodiments, obtaining the historical operating data of the battery further includes obtaining the historical operating data of the battery during after-sales repair of a faulty vehicle. For example, when a vehicle breaks down and requires repair during after-sales repair, the historical operating data of the faulty battery can be collected during the repair to serve as training data for the battery fault detection model.

[0036] Step S102: Calculate basic indicator data of the battery based on historical operating data.

[0037] Specifically, the basic indicator data of the battery refers to the relevant parameters required to evaluate the battery performance and status, such as: the pressure difference, battery cell internal resistance and other indicator parameters required to judge the capacity consistency, and the temperature difference indicator parameters required to judge the degree of battery aging.

[0038] In one embodiment, the basic indicator data of the battery includes at least one of internal resistance, capacity, static pressure difference, maximum temperature difference and charged capacity. Among them, the internal resistance of the battery cell will gradually increase with the use of the battery, the aging of the electrolyte, the structural change of the electrode material and other factors, thereby affecting the charging and discharging efficiency and energy loss of the battery. A higher internal resistance may cause the battery to overheat, accelerate the battery aging process, and shorten the battery life. The internal resistance of the battery cell can be calculated by the change in voltage and current in the constant current stage when the battery is charged, such as R=ΔU / ΔI. The capacity of the battery reflects the ability of the battery to store electrical energy. As the use time increases, the capacity of the battery will gradually decrease. It can be calculated based on the integral of the current and time of the charging process, or the differential voltage and differential capacity method. The static pressure difference is the maximum pressure difference of the battery in the full static stage after charging. This indicator can reflect the consistency inside the battery. If the pressure difference is too large, it may indicate that there is a problem inside the battery, such as the performance of some battery cells is degraded, which will cause the performance of the entire battery pack to decline. The maximum temperature difference refers to the difference in temperature inside the battery that may occur due to chemical reactions during the charging process. The maximum temperature difference can be obtained through temperature change data monitored by temperature sensors arranged inside or outside the battery. The charge input refers to the amount of electricity received by the battery during the charging process. The change in the charge input can reflect the charging efficiency and self-discharge of the battery. If the self-discharge rate of the battery is high, it means that the battery will also lose more energy when it is not working, which will affect the actual use time of the battery. It can be calculated based on the open circuit voltage (OCV) at rest before charging and the OCV at rest after charging, as well as the change in the battery state of charge (SOC) obtained through the OCV-SOC curve. In some other embodiments, the basic indicator data may also include parameters such as energy density and charge and discharge rate. This application does not limit the specific parameter types contained in the basic indicator data.

[0039] In one embodiment, calculating basic indicator data of the battery according to historical operating data includes: performing data cleaning on the historical operating data; and calculating the basic indicator data of the battery based on the cleaned historical operating data.

[0040] Specifically, the acquired historical operating data undergoes data cleaning, including denoising, missing value processing, data conversion, and normalization for abnormal data collected during the collection process. Since various data quality issues may arise during the collection of historical operating data, including but not limited to null data, data transmission errors, time misalignment, precision errors, and other data quality issues, it is necessary to process these abnormal data to ensure the accuracy of the basic indicator data calculation. For example, if the collected historical operating data contains null data, missing values ​​can be removed or filled using methods such as deletion, interpolation, K-nearest neighbor algorithm, and random forest filling. When time misalignment occurs, time alignment can be achieved based on the sampled time series data to convert the time-misaligned data into normal time series data. When data with precision errors occurs, data precision conversion can be performed based on the precision of other data in the same data list. After the historical operating data is collected, it is normalized to bring it to the same scale to facilitate the processing of the battery fault detection model.

[0041] Step S103: performing signal processing based on the basic indicator data to obtain a low-frequency approximate signal for characterizing the intrinsic state change of the battery.

[0042] Specifically, based on the historical operating data within a preset time period, the basic indicator data corresponding to different time nodes can be calculated to form a basic indicator data time series. The calculated basic indicator data time series can be processed by signal processing methods and decomposed into high-frequency detail signals and low-frequency approximate signals. Among them, the high-frequency detail signal is used to characterize the detailed changes of the battery, which refers to the short-term fluctuations and rapid changes of the battery during use, such as Figure 5 The low-frequency approximate signal is mainly used to characterize the intrinsic state change of the battery. The intrinsic state change usually refers to the long-term change of the internal chemical state of the battery, reflecting the long-term trend of the battery performance, such as Figure 6 As the battery life increases, the internal resistance of the battery cell gradually increases, or as the battery life increases, the battery capacity gradually decreases, the electrode material gradually decays, and the electrolyte gradually decomposes. In the embodiments of the present application, by analyzing the low-frequency approximate signal and monitoring the changes in the battery's intrinsic state, it is possible to identify battery anomalies at an early stage of failure, understand the battery's true state in a timely manner, reduce interference from other factors during the detection process, and enable customers to formulate maintenance strategies as early as possible to avoid battery thermal runaway.

[0043] In one embodiment, the processing method used for signal processing based on basic indicator data includes at least one of variational mode decomposition, empirical mode decomposition, wavelet decomposition and wavelet packet decomposition. Specifically, variational mode decomposition (VMD) is an adaptive non-recursive signal decomposition technology that can decompose complex battery data into a series of intrinsic mode functions (IMFs), that is, multiple modal components. Usually, the low-frequency modal components represent the low-frequency approximate signals of the signal and can characterize the intrinsic state changes of the battery. Empirical mode decomposition (EMD) is an adaptive signal decomposition method that can decompose the signal according to the time scale characteristics of the data itself. It can decompose the data into several IMFs, each IMF characterizing the local characteristics of the signal. The intrinsic state change trend and periodicity information can be extracted through the low-frequency IMF components to identify the operating status and degradation mode of the battery. Wavelet decomposition can decompose a signal into a series of wavelet functions, each with a specific frequency and time range. After wavelet decomposition, multiple detail coefficients and approximate coefficients are usually obtained. These coefficients correspond to signal components of different frequencies. Through signal reconstruction, the low-frequency approximate coefficients can be retained to obtain the low-frequency trend of the signal. Wavelet packet decomposition is an extension of wavelet decomposition, which can provide more detailed time-frequency analysis and better capture the detailed characteristics of the signal. Similarly, by extracting the coefficients of the low-frequency part from the results of wavelet packet decomposition, a low-frequency approximate signal representing the change of the battery's intrinsic state can be obtained.

[0044] Step S104: Generate reference feature data based on the low-frequency approximate signal by an automatic feature generation method.

[0045] Specifically, automatic feature generation methods can be used to extract features useful for battery health assessment and fault diagnosis based on low-frequency approximate signals. The automatic feature generation methods used to generate reference feature data include, but are not limited to, deep feature synthesis and genetic feature generation.

[0046] In one embodiment, reference feature data is generated based on a low-frequency approximate signal through an automatic feature generation method, including: labeling historical operation data to associate fault data in the historical operation data with a preset fault type label; obtaining a fault type label of the fault data corresponding to the low-frequency approximate signal based on the fault type label associated with the fault data; and generating reference feature data associated with the corresponding fault type label through an automatic feature generation method based on the low-frequency approximate signal and the corresponding fault type label.

[0047] Specifically, through labeling, the cleaned data is associated with the corresponding battery status (normal or faulty), providing training labels for the battery fault detection model. In one embodiment, fault type labels can be defined based on the battery fault type and fault severity. For example, based on the fault type, labels such as overheating, sensor damage, and capacity decay can be defined, and based on the severity, labels such as primary fault, secondary fault, and tertiary fault can be defined. Each type of fault data can also be associated with multiple labels simultaneously, so that the battery fault detection model can identify and analyze the fault data.

[0048] During the training process of the battery fault detection model, since the fault data is associated with the fault type label, when calculating the basic indicator data, the basic indicator data calculated based on the fault data is also associated with the fault type label, which in turn causes the low-frequency approximate signal generated based on the basic indicator data to be associated with the corresponding fault type label. Therefore, when automatic feature generation is performed based on the low-frequency approximate signal and the corresponding fault type label, reference feature data associated with the corresponding fault type label can be generated. During the automatic feature generation process, the battery fault detection model can identify the fault type label, analyze the feature types that are highly correlated with the corresponding fault type, and use the automatic feature generation method to purposefully generate relevant features corresponding to the fault type label, thereby enhancing the correlation between the reference feature data and the fault type and reducing the computational complexity of the model training process.

[0049] Step S105: Determine model parameters based on the reference characteristic data to obtain a battery fault detection model.

[0050] Specifically, after generating reference feature data that can be used to evaluate the battery health status through the battery fault detection model, one or more feature data with reference significance can be selected to analyze the relevant fault types. Based on the battery fault detection results obtained from the analysis, the parameters of the model can be continuously adjusted to build a model that can accurately detect battery faults, ensuring the effectiveness and reliability of the model in practical applications.

[0051] In one embodiment, model parameters are determined based on reference feature data to obtain a battery fault detection model, including: performing feature dimensionality reduction on the reference feature data; obtaining a battery fault detection result based on the reference feature data after the feature dimensionality reduction processing; adjusting the model parameters according to the battery fault detection result and a fault type label associated with the reference feature data; when the battery fault detection result matches the fault type corresponding to the fault type label, saving the corresponding model parameters to obtain a battery fault detection model.

[0052] Specifically, since automatic feature generation can generate a large amount of feature data based on low-frequency approximate signals, there may be redundant features in these feature data, and high-dimensional feature data may increase the computational cost of model training and detection. Therefore, it is necessary to screen and reconstruct the features through feature dimensionality reduction to remove redundant features and obtain features with more reference significance. Among them, the methods used for feature dimensionality reduction of reference feature data include but are not limited to principal component analysis, linear discriminant analysis, matrix singular value decomposition, least squares method, etc. By reducing the number of features and the dimension of feature data through feature dimensionality reduction, the time and computing resources required for model training and detection are reduced, and the space required for data storage is also reduced.

[0053] After feature dimensionality reduction, the battery health status is analyzed based on the reduced feature data to obtain the battery fault detection results. When the battery fault detection result matches the fault type corresponding to the fault type label, the battery fault detection model's detection results are accurate. During the battery fault detection model training process, different algorithms, different feature combinations, and different algorithm parameters can be adjusted to achieve the optimal recall rate and false alarm rate of the battery fault detection model, thereby building a model that can accurately detect battery faults.

[0054] The method for constructing a battery fault detection model provided in an embodiment of the present application obtains historical operating data of the battery; calculates basic indicator data of the battery based on the historical operating data; performs signal processing based on the basic indicator data to obtain a low-frequency approximate signal used to characterize the change in the intrinsic state of the battery; generates reference feature data based on the low-frequency approximate signal through an automatic feature generation method; and determines model parameters based on the reference feature data to obtain a battery fault detection model. In this way, by analyzing the low-frequency approximate signal, the battery's usage status is more realistically reflected, reducing interference from other factors. In addition, combined with the automatic feature generation method, the battery's gradual change features are automatically generated, which can reduce the model's computational complexity, increase feature recognition coverage, and shorten the model construction cycle.

[0055] See Figure 4 , is a battery fault detection method provided in an embodiment of the present application, which uses a battery fault detection model constructed by the method described in the above embodiment. The battery fault detection method includes: Step S201: Acquire historical operating data of the battery to be tested.

[0056] Step S202: inputting historical operation data into a battery fault detection model to perform fault detection to obtain a fault detection result of the battery to be tested.

[0057] Step S203: Output the fault detection result.

[0058] Specifically, when performing fault detection on a battery to be tested, the historical operating data of the battery to be tested is input into a trained battery fault detection model. The battery fault detection model can quickly analyze the health status of the battery, detect whether there is a fault in the current state of the battery to be tested, and detect the corresponding fault type. In a specific embodiment, when the trained battery fault detection model is deployed in the cloud to detect the battery to be tested, the voltage, current and other related parameters of the battery to be tested are collected within a preset time period. After calculation by the battery fault detection model, the change trend of the internal resistance of each battery cell in the battery over time can be obtained. Figure 5 As shown, from Figure 5 It can be found that the internal resistance of cell 1 increases faster than that of other cells as the usage time increases, indicating that the internal resistance of this cell is abnormal. After the battery fault detection model processes the signal using the variational mode decomposition method, the low-frequency approximate signal is obtained as follows: Figure 6 As shown in the figure, it can be more clearly found that the growth trend of the internal resistance of cell 1 is significantly different from that of the internal resistance of other cells. The automatic feature generation method generates relevant features such as maximum value, minimum value, mean, median, number of peaks, standard deviation, etc., and the feature data is reduced in dimension by principal component analysis. The large amount of feature data generated is reconstructed and screened, and the obtained features are more meaningful for reference. In this embodiment, one of the features obtained after feature dimensionality reduction is as follows Figure 7 As shown in , the clustering algorithm is used to detect outliers on this feature, and the cells corresponding to the detected outliers are abnormal cells, such as Figure 7 The characteristic value of the cell numbered 1 is the detected outlier. It can be clearly seen from the figure that the characteristic value of the cell numbered 1 is significantly smaller than the characteristic values ​​of other cells. At this time, the battery fault detection model will identify the cell as an abnormal cell and determine that the cell has an abnormal internal resistance fault.

[0059] The battery fault detection method provided in the embodiment of the present application can quickly detect whether the battery status is faulty based on the historical operating data of the battery, and can avoid the dangerous situation of battery thermal runaway.

[0060] Based on the same inventive concept as the aforementioned embodiments, an embodiment of the present application further provides a battery fault detection system, which may include a processor and a memory storing a computer program. When the processor runs the computer program, the method for constructing a battery fault detection model and the battery fault detection method described in the aforementioned embodiments are implemented.

[0061] Based on the same inventive concept as the aforementioned embodiments, embodiments of the present application further provide a readable storage medium storing a computer program. The readable storage medium may be a ferromagnetic random access memory (FRAM), a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a flash memory, a magnetic surface mount device, an optical disc, or a compact disc read-only memory (CD-ROM). When the computer program stored in the readable storage medium is executed by a processor, the method for constructing a battery fault detection model and the battery fault detection method described above are implemented. The specific steps implemented when the computer program is executed by the processor are described in the aforementioned embodiments and are not further elaborated here.

[0062] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0063] As used herein, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion of elements other than the listed elements and may also include additional elements not specifically listed.

[0064] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A method for constructing a battery fault detection model, characterized in that: include: Get historical operating data of the battery; Calculate basic indicator data of the battery according to the historical operation data; Performing signal processing based on the basic indicator data to obtain a low-frequency approximate signal for characterizing a change in an intrinsic state of the battery; Based on the low-frequency approximate signal, generating reference feature data by an automatic feature generation method; Based on the reference characteristic data, model parameters are determined to obtain a battery fault detection model.

2. The method for constructing a battery fault detection model according to claim 1, characterized in that: The obtaining of historical operating data of the battery includes: Obtaining historical operating data of the battery collected by the battery management system through the vehicle-mounted TBOX; and / or Obtaining the historical operating data of the battery sent by the charger through the station control terminal of the battery swap station; and / or When a faulty vehicle is repaired through after-sales service, the historical operating data of the battery is obtained.

3. The method for constructing a battery fault detection model according to claim 1, characterized in that: The calculating of basic indicator data of the battery according to the historical operation data includes: Performing data cleaning processing on the historical operation data; Based on the historical operation data after cleaning, basic indicator data of the battery is calculated.

4. The method for constructing a battery fault detection model according to claim 1, characterized in that: The method of generating reference feature data based on the low-frequency approximate signal by an automatic feature generation method includes: Labeling the historical operation data to associate the fault data in the historical operation data with a preset fault type label; Based on the fault type label associated with the fault data, obtaining the fault type label of the fault data corresponding to the low-frequency approximate signal; Based on the low-frequency approximation signal and the corresponding fault type label, reference feature data associated with the corresponding fault type label is generated through an automatic feature generation method.

5. The method for constructing a battery fault detection model according to claim 4, characterized in that: Based on the reference characteristic data, determining model parameters to obtain a battery fault detection model includes: Performing feature dimension reduction on the reference feature data; Obtaining a battery fault detection result based on the reference feature data after feature dimension reduction processing; adjusting model parameters according to the battery fault detection result and the fault type label associated with the reference feature data; When the battery fault detection result matches the fault type corresponding to the fault type label, the corresponding model parameters are saved to obtain the battery fault detection model.

6. The method for constructing a battery fault detection model according to claim 1, characterized in that: The basic indicator data of the battery includes at least one of internal resistance, capacity, static pressure difference, maximum temperature difference and charged power.

7. The method for constructing a battery fault detection model according to claim 1, characterized in that: The processing method used for signal processing based on the basic indicator data includes at least one of variational mode decomposition, empirical mode decomposition, wavelet decomposition and wavelet packet decomposition.

8. A battery failure detection method, characterized in that: A battery fault detection model constructed using the method according to any one of claims 1 to 7, the method comprising: Obtain historical operating data of the battery to be tested; Inputting the historical operation data into the battery fault detection model to perform fault detection to obtain a fault detection result of the battery to be tested; The fault detection result is outputted.

9. A battery fault detection system, characterized in that: The method comprises a processor and a memory storing a computer program, and when the processor runs the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

10. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images