Method and system for identifying fault battery in vehicle battery pack

Abstract

The invention discloses a method and a system for identifying a fault battery in a vehicle battery pack, which smartly monitors the battery pack connected to a vehicle for power supply to obtain battery unit monitoring data when the battery pack works, and identifies the fault battery in combination with vehicle monitoring data obtained according to the working state of the vehicle. According to the scheme, target feature data pointing to battery abnormity are screened out through a mathematical analysis method, then a detection neural network is used for judging the target feature data, and information of an abnormal battery is positioned according to a judgment result. Comprehensively evaluating the health condition of the battery pack by fusing multi-source data; according to the scheme, the faulty battery can be positioned according to the identification result, then power supply shielding is carried out on the faulty battery in time, meanwhile, power supply of the vehicle is intervened in time in combination with a redundant battery pack mode, and strategy guarantee with positive practical significance can be provided for reliable and stable work of the vehicle.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of battery fault detection, battery control and electric vehicle battery management, and particularly relates to a method and system for identifying a faulty battery in a vehicle battery pack. BACKGROUND

[0002] As a popular automobile subfield at present, electric vehicles are favored by many consumers due to their fast speed and no exhaust emission of polluted air during driving. The battery pack is the energy source of the electric vehicle, and the configuration capacity of the battery pack usually has high requirements because the electric vehicle is entirely powered by the battery pack. Under the current technology, the battery pack of an electric vehicle often needs to be configured with thousands of single batteries for cooperation. Since the battery pack used by the current electric vehicle is mostly a lithium battery, the working stability of the lithium battery is affected by the environmental temperature, and the lithium battery needs to be in a relatively stable environment to have stable working performance. At the same time, the internal structure of the lithium battery is relatively precise, which makes it possible to be damaged due to the destruction of the stable working environment or internal structure when subjected to external force. When the battery pack is working, the internal heat of the battery pack is generated due to the change of the current and voltage during charging and discharging, which to some extent also generates certain internal stress to cause fluctuations in the battery working parameters. In the case of a large number of batteries, how to identify the faulty battery is very critical. The existing disclosed documents mostly detect the battery pack as a whole, which lacks pertinence. Once an individual battery abnormally overheats, it is difficult to be discovered and shielded in time. Therefore, how to improve the monitoring accuracy and reliability of the vehicle battery during working is a very positive and practical subject. SUMMARY

[0003] In view of the above, the purpose of the present application is to provide a method and system for identifying a faulty battery in a vehicle battery pack, which is reliable in implementation, flexible in application, and good in reference of identification results.

[0004] In order to achieve the above technical purpose, the technical solution adopted by the present application is:

[0005] A method for identifying a faulty battery in a vehicle battery pack, comprising:

[0006] monitoring the working data of the batteries in the battery pack in response to the information that the battery pack is connected to the vehicle for power supply work, and generating battery monitoring data;

[0007] monitoring the working state of the vehicle in response to the working instruction signal of the vehicle, and generating vehicle monitoring data;

[0008] Obtaining vehicle monitoring data, marking the state of the vehicle at different times according to preset conditions, and generating vehicle state information marked in time series;

[0009] Obtaining battery monitoring data, combining vehicle state information in the same time sequence, and determining target feature data about the battery;

[0010] Based on the target feature data, it is determined whether the battery in the battery pack has a fault, and corresponding identification information is generated.

[0011] As a possible implementation, further, when the battery pack is connected to the vehicle for power supply, the battery sampling module installed on the battery pack power supply line senses the connection of the battery pack, and the battery sampling module is activated by the battery pack power supply;

[0012] Among them, the battery pack connected to the vehicle is more than one, each of the battery packs corresponds to a power supply line, the battery pack includes a plurality of battery units composed of one or more single batteries, and each of the battery units is assigned a unique ID.

[0013] As a preferred implementation option, preferably, the battery sampling module is a multi-channel module that samples temperature, state of charge, current and / or voltage data of the battery unit;

[0014] The data obtained by the battery sampling module for different battery units are associated with the ID of the corresponding battery unit and the sampling time to form a data set; the data sets of a plurality of battery units are collected to form battery monitoring data;

[0015] The vehicle monitoring data includes working state data of the power-consuming devices of the vehicle, and the power-consuming devices include one or more of the electric motor, air conditioner, lamp, screen, sensor, and electric brake controller loaded on the vehicle.

[0016] As a preferred implementation option, preferably, the battery sampling module is a multi-channel module that samples temperature, state of charge, current and / or voltage data of the battery unit;

[0017] Among them, the battery pack is packaged in a box-shaped shell, and is provided with an explosion-proof valve and a plug-in part, and the sensor group includes one or more of a MEMS pressure sensor, a nano humidity sensor, an optical fiber strain sensor, and a piezoelectric accelerometer.

[0018] Specifically, the MEMS pressure sensor is multiple, which is distributedly arranged at four corners inside the box of the battery pack, and is used for monitoring pressure change in the battery pack, and has a sampling frequency of 1 Hz and generates pressure monitoring data correspondingly;

[0019] The nano-humidity sensor is more than one, which is arranged near the box of the battery pack connected with the explosion-proof valve or the connector, and is used for monitoring relative humidity of the battery pack, and has a sampling frequency of 0.1 Hz and generates humidity monitoring data correspondingly;

[0020] The optical fiber strain sensor is arranged at the welding seam of the box of the battery pack, and is used for monitoring micro-strain of the box, and has a sampling frequency of 10 Hz and generates strain monitoring data correspondingly;

[0021] The piezoelectric accelerometer is arranged in the box of the battery pack, and is used for monitoring vibration of the battery pack, and has a sampling frequency of 100 Hz and generates vibration monitoring data correspondingly;

[0022] The pressure monitoring data, humidity monitoring data, strain monitoring data and vibration monitoring data are collected to form multi-source data, and then are classified into the battery pack monitoring information, and the battery pack monitoring information further includes preset ID information of the corresponding battery pack.

[0023] On the basis, the identification method further includes:

[0024] According to the battery pack monitoring information, comprehensive analysis is performed to generate a comprehensive analysis result, and then the comprehensive analysis result is judged or is jointly judged with data obtained by single sensor monitoring in the battery pack monitoring information, so as to realize abnormal identification of the battery pack, generate early warning level information, and guide response measures of the vehicle battery pack to start.

[0025] As a preferred implementation selection, preferably, according to the battery pack monitoring information, the comprehensive analysis to generate a comprehensive analysis result includes:

[0026] The pressure monitoring data, humidity monitoring data, strain monitoring data and vibration monitoring data in the battery pack monitoring information are extracted in time sequence to generate time sequence data x;

[0027] The time sequence data x is preprocessed to remove drift and synchronization alignment, and clean time sequence data x is obtained

[0028] The time sequence data x is preprocessed to remove drift and synchronization alignment, and clean time sequence data x is obtained Feature extraction is performed to obtain dimensionless abnormal quantities, including pressure abnormal quantity PA(t), humidity climb rate HCR(t), strain dispersion SD(t) and vibration root mean square VR(t), to generate a four-dimensional feature vector z(t) = [PA, HCR, SD, VR] about battery pack monitoring information;

[0029] The data in the four-dimensional feature vector is normalized and then substituted into a weighted fusion model to calculate a health index SHI, which is set as the comprehensive analysis result, and is defined as follows:

[0030] SHI(t) = αPA n (t) + βHCR n (t) + γSD n (t) + δVR n (t)

[0031] Wherein, PA n (t), HCR n (t), SD n (t), VR n (t) are respectively the normalized data of the pressure abnormal quantity, humidity climb rate, strain dispersion and vibration root mean square obtained at the sampling time t, and α, β, γ, δ are respectively weight coefficients, and α + β + γ + δ = 1.

[0032] As a preferred implementation option, preferably, the comprehensive analysis result is judged or combined with the data obtained by a single sensor in the battery pack monitoring information to realize abnormal recognition of the battery pack and generate early warning level information including one of the following:

[0033] (1) When the health index SHI is greater than the first threshold value for a duration exceeding a preset duration, record an abnormal event for directional inspection during maintenance;

[0034] (2) Combine the health index SHI with the humidity monitoring data, when the health index SHI is greater than the second threshold value, and the humidity monitoring data indicates that the humidity rises by more than 3% within 24 hours, limit the power of the battery pack, and push the repair information at the same time;

[0035] (3) Combine the health index SHI with the strain monitoring data, when the health index SHI is greater than the third threshold value, and the strain monitoring data indicates that the stress mutation is greater than 100 pa / s, start the safety mode, and push the suggestion to deactivate information at the same time.

[0036] As a preferred implementation option, preferably, the vehicle monitoring data is obtained, and the state of the vehicle at different times is information labeled according to a preset condition to generate vehicle state information labeled in time sequence, including:

[0037] Obtaining vehicle monitoring data, arranging it in time sequence, and then according to the on-off state of the power-consuming devices of the vehicle in the time sequence, combining the power data of the power-consuming devices when working, the total power demand of the power-consuming devices of the automobile at different time points is counted, and the formula is as follows:

[0038] W t = w 1t + w 2t +…+ w nt

[0039] Wherein, W t is the total power demand of the power-consuming devices of the automobile at time point t, w 1t , w 2t … w nt are the power consumption of different power devices in the preset range at time point t.

[0040] According to the total power demand of the power-consuming devices of the automobile at different time points, the total power demand is classified into 1-6 levels and information is labeled according to the preset power interval value, and the vehicle monitoring data with the same total power demand in the same time period is simplified to the relationship between the total power demand corresponding level and time interval, and the vehicle state information labeled in time sequence is generated.

[0041] As a preferred implementation option, preferably, the scheme obtains battery monitoring data, and determines the target feature data about the battery by combining the vehicle state information in the same time sequence, including:

[0042] Obtaining battery monitoring data, arranging it in time sequence;

[0043] According to the time sequence, the vehicle state information in the same time sequence as the battery monitoring data is called;

[0044] According to the vehicle state information in the same time sequence, the data groups in the battery monitoring data are screened for abnormal data by residual analysis, data clustering analysis or regression analysis method, and the target feature data about the battery and used for judging the battery fault is determined, the target feature data includes the data group with abnormal judgment result, and its corresponding time sequence and vehicle state information.

[0045] As a preferred implementation option, preferably, the scheme determines whether the battery in the battery pack has a fault based on the target feature data, and correspondingly generates identification information including:

[0046] The target feature data is imported into the trained detection neural network for detection, and the detection result is output by the detection neural network;

[0047] Obtaining the detection result,

[0048] When the detection result indicates that the battery corresponding to the target feature data is a faulty battery, fault information is generated, and then the ID of the battery cell corresponding to the target feature data is acquired according to the data group of the target feature data, and the fault information, the target feature data and the ID of the battery cell corresponding to the target feature data are output as identification information;

[0049] When the detection result indicates that the battery corresponding to the target feature data is a normal battery, the ID of the battery cell corresponding to the target feature data is acquired according to the data group of the target feature data, the number of times that the battery cell corresponding to the ID is judged to be abnormal is recorded, abnormal number information is generated, and the abnormal number information, the target feature data and the ID of the battery cell corresponding to the target feature data are output as identification information.

[0050] As a preferred implementation option, preferably, when the detection result indicates that the battery corresponding to the target feature data is a normal battery, the abnormal number information N corresponding to the ID of the target feature data is called according to the data group of the target feature data, and when the abnormal number information N is greater than a preset threshold M, artificial review information is generated and is associated with the identification information and is output together.

[0051] Based on the above, the application further provides an electric vehicle battery working management method, which comprises the identification method of a faulty battery in a vehicle battery pack as described above, and further comprises:

[0052] The identification information is acquired, when it indicates that there is a faulty battery in the battery pack, the faulty battery information is acquired, the power supply access of the faulty battery is cut off or shielded, and meanwhile, the output power range that can be provided when the battery pack accesses the vehicle for power supply work is updated;

[0053] The power supply power range required by the power-consuming devices of the vehicle is determined according to the vehicle monitoring data;

[0054] According to the power supply power range required by the power-consuming devices of the vehicle, in combination with the output power range that can be provided when the battery pack accesses the vehicle for power supply work, a redundant battery pack accesses the vehicle for power supply compensation under a preset condition.

[0055] Based on the above, the application further provides an identification system of a faulty battery in a vehicle battery pack, which comprises:

[0056] A battery monitoring module is configured to generate battery monitoring data by monitoring the working data of the batteries in the battery pack in response to the information that the battery pack accesses the vehicle for power supply work;

[0057] A vehicle monitoring module is configured to generate vehicle monitoring data by monitoring the working state of the vehicle in response to the working instruction signal of the vehicle;

[0058] An information marking module is configured to acquire vehicle monitoring data, mark the state of the vehicle at different times according to preset conditions, and generate vehicle state information marked in time sequence.

[0059] A data screening module is configured to acquire battery monitoring data, determine target feature data about the battery by combining the vehicle state information in the same time sequence.

[0060] A data processing module is configured to determine whether a fault exists in the battery in the battery pack based on the target feature data, and generate identification information accordingly.

[0061] Based on the above, the application further provides a computer-readable storage medium, which stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by a processor to realize the identification method of the fault battery in the vehicle battery pack.

[0062] Compared with the prior art, the application has the beneficial effects that: the application ingeniously monitors the battery pack that supplies power to the vehicle to obtain monitoring data of the battery cells included in the battery pack when the battery pack works, combines the vehicle monitoring data obtained by monitoring the working state of the vehicle, screens out target feature data pointing to the battery anomaly by a mathematical analysis method, judges the target feature data by using a detection neural network, and locates the information of the abnormal battery according to the judgment result. In addition, the application can monitor the physical parameters such as the air tightness and structure of the battery pack, and then comprehensively evaluate the health status of the battery pack by fusing multi-source data, thereby providing multiple safeguards for the safe work of the battery pack. The application is not only reliable in implementation, but also can further locate the fault battery through the identification result, shield the power supply of the fault battery in time, and timely intervene in the power supply of the vehicle in combination with the redundant battery pack, thereby providing a positive and practical strategy guarantee for the reliable and stable work of the vehicle. BRIEF DESCRIPTION OF DRAWINGS

[0063] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only constitute some embodiments of the application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0064] Figure 1 is a schematic diagram of the brief implementation process steps of the identification method of the application;

[0065] Figure 2is a brief unit module connection diagram of the identification system of the present solution. DETAILED DESCRIPTION

[0066] The application will be further described below in conjunction with the drawings and examples. It is particularly pointed out that the following examples are only used to illustrate the present application, but do not limit the scope of the present application. Similarly, the following examples are only part of the embodiments of the present application, not all embodiments, and all other embodiments obtained by those skilled in the art without creative labor are within the scope of the present application.

[0067] As shown in Figure 1 the present embodiment is a method for identifying a faulty battery in a vehicle battery pack, which comprises:

[0068] S01, in response to the information that the battery pack is connected to the vehicle for power supply work, the working data of the battery in the battery pack is monitored to generate battery monitoring data;

[0069] S02, in response to the working instruction signal of the vehicle, the working state is monitored to generate vehicle monitoring data;

[0070] S03, obtaining vehicle monitoring data, labeling information of the state of the vehicle at different times according to preset conditions, and generating vehicle state information labeled in time sequence;

[0071] S04, obtaining battery monitoring data, and determining target feature data about the battery by combining vehicle state information in the same time sequence;

[0072] S05, based on the target feature data, determining whether the battery in the battery pack is faulty, and generating identification information accordingly.

[0073] In order to facilitate the response to the access signal of the battery pack, an inductive device module can be arranged on the circuit of the battery pack connected to the vehicle to feedback that the power supply port has the battery pack access in time. Specifically, as a possible implementation, further, when the battery pack is connected to the vehicle for power supply work, the battery sampling module installed on the battery pack power supply circuit senses the access of the battery pack, and the battery sampling module is activated by the battery pack power supply.

[0074] Since the battery pack of the electric vehicle can be connected to the vehicle according to the actual design needs or scheduling arrangement needs, the battery pack connected to the vehicle can be more than one. In order to facilitate the positioning and information recording of the battery in the battery pack, in the present solution, each battery pack corresponds to a power supply line, the battery pack comprises a plurality of battery units composed of one or more single batteries, and each battery unit is assigned a unique ID.

[0075] In order to improve the convenience and integrity of the data sampling of the battery pack, as a preferred embodiment, the battery sampling module is a multi-channel module, which samples the temperature, state of charge, current and / or voltage data of the battery unit.

[0076] On this basis, the data obtained by the battery sampling module for different battery units are associated with the ID and sampling time of the corresponding battery unit to form a data group, and the data groups of several battery units are collected to form battery monitoring data.

[0077] Since many power-consuming devices of the electric vehicle are powered by the battery pack, in order to improve the close type of battery pack monitoring and vehicle operation, the vehicle monitoring data includes the working state data of the power-consuming devices of the vehicle, and the power-consuming devices include one or more of the electric motor, air conditioner, lamp, screen, sensor and electric brake.

[0078] Among them, the main purpose of monitoring the working state in response to the working instruction of the vehicle is to record the power consumption demand of the vehicle in real time, because the power supply demand of the vehicle is different in different states, therefore, it is of positive significance to convert the vehicle monitoring data into data corresponding to the battery monitoring data, and it can also be judged that the vehicle is in a high or low power supply state, especially when the battery pack has a higher power supply output, the heat is more obvious, and the stability monitoring at this time is more important.

[0079] In the present application, as a preferred embodiment, the vehicle monitoring data is obtained, the state of the vehicle at different times is labeled according to the preset condition, and the vehicle state information labeled in time sequence includes:

[0080] The vehicle monitoring data is obtained, arranged in time sequence, and then according to the working start-stop condition of the power-consuming devices of the vehicle in the time sequence, combined with the power data of the power-consuming devices when working, the total power demand of the power-consuming devices of the vehicle at different time points is counted, and the formula is as follows:

[0081] W t =w 1t +w 2t +…+w nt

[0082] Among them, W t is the total power demand of the power-consuming devices of the vehicle at time point t, w 1t , w 2t …w nt are the power consumptions of different power devices in the preset range at time point t.

[0083] The present scheme, by the total power demand of the power consuming device of the automobile at different time points, the total power demand is classified into 1-6 levels according to the preset power interval value, and the information is marked, and the vehicle monitoring data of the same time period is simplified as the relationship between the total power demand and the corresponding level and time interval, and the vehicle state information marked with time sequence is generated.

[0084] As an example, the total power demand is classified into 1-6 levels according to the preset power interval value, which can be classified as shown in the following table:

[0085] Table 1 Total power classification interval

[0086] Serial number Power consumption interval (unit / kW) Level 1 [0，1] 1 2 (1，5] 2 3 (5，10] 3 4 (10，20] 4 5 (20，40] 5 6 ＞40 6

[0087] In the present scheme, by statistically analyzing the total power demand of the power consuming device of the automobile at different time points, and then classifying and marking the vehicle state information according to the time sequence, the power demand intensity is intuitively reflected, and the power supply output intensity of the battery pack is reflected. Based on this, data that does not meet the power supply output intensity is selected to determine whether there is a faulty battery in the battery pack.

[0088] Specifically, as a preferred implementation option, preferably, the present scheme obtains battery monitoring data, and determines the target feature data about the battery by combining the vehicle state information in the same time sequence, including:

[0089] Obtain battery monitoring data and arrange it according to the time sequence;

[0090] According to the time sequence, the vehicle state information in the same time sequence as the battery monitoring data is retrieved;

[0091] According to the vehicle state information in the same time sequence, the data groups in the battery monitoring data are screened for abnormal data by residual analysis, data clustering analysis or regression analysis method, and the target feature data about the battery for judging battery failure is determined, including the data groups with abnormal judgment results, and the corresponding time sequence and vehicle state information.

[0092] In the present scheme, the battery working parameter information of the battery pack at different power supply output levels (i.e. the power supply output under different vehicle state information, recording the battery working parameter information, including at least one of temperature, current, SOC, voltage) can be constructed in advance, then combined with the working time to form a reference data, and compared with the corresponding data groups of each battery unit in the battery monitoring data, thereby realizing the preliminary fault screening of abnormal battery units. For some parameters that are not obvious, further screening is required through statistical methods and associated vehicle state information.

[0093] Among them, the embodiment scheme is analyzed with the analysis parameters of the voltage and temperature of the battery of the battery pack, and is substituted into the residual analysis, data clustering analysis or regression analysis to obtain the process of the example target feature data.

[0094] (1) Residual analysis

[0095] As one of the examples of residual analysis, when the data group in the battery monitoring data is screened by residual analysis, the method includes the following steps:

[0096] In the residual analysis, the residual of the real-time data of the calculated voltage and temperature of each battery unit of the battery pack and the voltage and temperature in the reference data can be calculated, and the formula is as follows:

[0097]

[0098] Wherein, r t is the residual value, y t is the temperature or voltage of the battery unit of the battery pack obtained by measurement, is the temperature or voltage of the battery unit of the battery pack in the reference data.

[0099] Through the above calculation, the measured temperature and voltage of the battery unit of the battery pack and the corresponding residual of the reference data can be obtained, and then the residual data obtained by principal component analysis (PCA) is processed and calculated to reduce the data dimension and highlight the important statistical characteristics.

[0100] The results of the PCA analysis are judged by using the CUSUM technology, and the data exceeding the threshold value is defined as the data group with possible failure, and then the corresponding time sequence, vehicle state information and its collection are associated to generate target feature data.

[0101] (2) Data clustering analysis

[0102] As one of the examples of data clustering analysis, when the data group in the battery monitoring data is screened by data clustering analysis, the parameters processed by the battery unit are temperature and voltage, which generally include the following steps:

[0103] The voltage and temperature data of the data group in the battery monitoring data are standardized to make each feature dimension in the same order of magnitude for subsequent processing.

[0104] The voltage and temperature of the data group in the battery monitoring data are segmented in time sequence to make them become data containing a certain time window, and each segment of data includes data points of fixed time length.

[0105] K-shape clustering algorithm is applied to segmented data, and clustering is performed based on the similarity of data shapes to compare the similarity between each segment of data, then select a segment of data as the center of clustering, and then optimize the center point by iteration, maximize the shape similarity in the group, and finally assign each data segment to the cluster corresponding to the center with the most similar shape.

[0106] The clustering results are analyzed to detect whether there is a significant change in the distance between the cluster members (i.e. data) or the cluster center, and for the anomaly that gradually develops over time, the change of the cluster members over time is analyzed to see if there is a data segment transferred from one cluster to another.

[0107] In the K-shape clustering algorithm, the data preprocessing, distance calculation, center point updating and anomaly detection can be performed as follows:

[0108] 1. Data preprocessing

[0109] Standardize each time series x t to ensure that the data is on the same scale, the formula is as follows:

[0110]

[0111] Where x t is the data to be preprocessed, μ is the mean of the sequence data, and σ is the standard deviation of the sequence data.

[0112] 2. Distance calculation

[0113] Shape similarity, use cross-correlation to measure the similarity between two different time series data X, Y, the cross-correlation function is defined as follows:

[0114]

[0115] Where t is the time lag parameter, are the mean values of the time series data respectively.

[0116] 3. Center point update

[0117] Iterative update: select a sequence data as the initial center, then update the center point by iteration, so that the total cross-correlation distance of the sequence data is maximized, the formula is defined as follows:

[0118]

[0119] Where S is the set of all data segments, C new is the updated center.

[0120] 4. Anomaly detection

[0121] Outlier analysis: detecting data segments with relatively low similarity to any cluster center, for example, a data segment X has a maximum span correlation value with all centers lower than a threshold θ, then it is marked as abnormal data, defined as follows:

[0122]

[0123] (3) Regression analysis

[0124] As an example of regression analysis, when the battery monitoring data is analyzed by regression analysis, the parameter processed is the state of charge SOC of the battery unit, which generally includes the following steps:

[0125] (1) Establish a ridge regression model to predict the SOC value of the battery unit when the battery pack is working; wherein the ridge regression model is defined as follows:

[0126]

[0127] Where X is the input feature, y is the target variable, and λ is the regularization parameter.

[0128] (2) By comparing the predicted value (state of charge SOC) of the ridge regression model with the actual collected state of charge SOC value, the data difference absolute value and change trend between the actual value and the predicted value are determined;

[0129] (3) When the difference absolute value and change trend deviate from the predicted value by a preset amplitude, the data group is marked as abnormal.

[0130] After screening the target feature data with the above method, the present scheme further determines whether the target feature data is abnormal through a more detailed method.

[0131] As a preferred implementation option, preferably, the present scheme determines whether the battery in the battery pack has a fault based on the target feature data, and generates identification information accordingly, including:

[0132] Importing the target feature data into the trained detection neural network for detection, and outputting the detection result from the detection neural network;

[0133] Obtaining the detection result,

[0134] When the detection result indicates that the battery corresponding to the target feature data is a fault battery, generate fault information, then according to the data group of the target feature data, obtain the ID of the corresponding battery unit, and output the fault information, the target feature data and the ID of the corresponding battery unit as identification information;

[0135] When the detection result indicates that the battery corresponding to the target feature data is a normal battery, the ID of the corresponding battery unit is obtained according to the data group of the target feature data, the number of times that the battery unit corresponding to the ID is judged to be abnormal is recorded, abnormal times information is generated, and the abnormal times information, the target feature data, and the ID of the corresponding battery unit are output as identification information.

[0136] As a preferred implementation option, preferably, when the detection result indicates that the battery corresponding to the target feature data is a normal battery, the ID corresponding to the data group of the target feature data is called to obtain the abnormal times information N, and when the abnormal times information N is greater than a preset threshold M, artificial review information is generated and is associated with the identification information and is output together.

[0137] In the present scheme, the detection neural network can be trained and verified by constructing a data set during training. The vehicle monitoring data and the corresponding battery monitoring data during operation are associated in the form of artificial or collected abnormal cases, and then labeled whether they are abnormal to form a reference data set. A preset amount of data is selected from the reference data set as a training set and a verification set. After being imported into the neural network for training, the trained neural network is verified by the verification set until the model converges, and the detection neural network for detecting whether the battery in the battery pack is abnormal is obtained.

[0138] In the present scheme, the target feature data includes data groups whose judgment results by the aforementioned analysis method are abnormal, and their corresponding time series and vehicle state information. By combining the current, voltage or SOC data in the data group with the vehicle state information as input items and importing them into the detection neural network, the detection result is output by the detection neural network, thereby realizing secondary judgment and improving the reliability and reference value of the identification result.

[0139] Since the battery pack is used in the vehicle, it will also be physically stressed due to the operation of the vehicle. For example, the battery pack is packaged in a box-shaped shell. When the vehicle is driving, it may collide with external obstacles or drive in some poor road conditions, causing the battery pack to vibrate and affecting the internal physical environment. In order to improve the monitoring diversity of the battery pack and provide protection for its reliable operation, as a preferred implementation option, preferably, the present scheme responds to the working instruction signal of the vehicle and monitors the working state of the battery pack on the vehicle. The gas tightness and structural safety of the battery pack are monitored by the sensor group to generate battery pack monitoring information.

[0140] The battery pack is packaged in a box-shaped shell and is provided with an explosion-proof valve and a plug-in piece. The sensor group includes one or more of a MEMS pressure sensor, a nano-humidity sensor, an optical fiber strain sensor, and a piezoelectric accelerometer.

[0141] Specifically, the MEMS pressure sensor is multiple, which is distributedly arranged at four corners inside the battery pack box, and is used for monitoring the pressure change in the battery pack, the sampling frequency is 1 Hz, the accuracy is ±0.1 pa, and the corresponding pressure monitoring data is generated;

[0142] The nano-humidity sensor is more than one, which is arranged near the battery pack box connected with the explosion-proof valve or the connector, and is used for monitoring the relative humidity of the battery pack, the sampling frequency is 0.1 Hz, the accuracy is ±0.1% RH, and the corresponding humidity monitoring data is generated;

[0143] The optical fiber strain sensor is arranged at the welding seam of the battery pack box, and is used for monitoring the micro-strain of the box, the sampling frequency is 10 Hz, the accuracy is ±0.1με, and the corresponding strain monitoring data is generated;

[0144] The piezoelectric accelerometer is arranged in the battery pack box, and is used for monitoring the vibration of the battery pack, the sampling frequency is 100 Hz, the vibration frequency spectrum range of the sampling is 0-500 Hz, and the corresponding vibration monitoring data is generated;

[0145] The pressure monitoring data, humidity monitoring data, strain monitoring data and vibration monitoring data are collected to form multi-source data, and then are classified into the battery pack monitoring information, and the battery pack monitoring information further includes preset ID information of the corresponding battery pack.

[0146] On the basis, the identification method further includes:

[0147] According to the battery pack monitoring information, comprehensive analysis is performed to generate a comprehensive analysis result, and then the comprehensive analysis result is judged or is combined with the data monitored by the single sensor in the battery pack monitoring information to realize abnormal identification of the battery pack, generate early warning level information, and guide the response measure of the vehicle battery pack to start.

[0148] As a preferred implementation selection, preferably, according to the battery pack monitoring information, the comprehensive analysis includes:

[0149] The pressure monitoring data, humidity monitoring data, strain monitoring data and vibration monitoring data in the battery pack monitoring information are extracted in time sequence to generate time sequence data x;

[0150] The time sequence data x is preprocessed to remove drift and synchronization alignment to obtain cleaned time sequence data

[0151] The time sequence data Feature extraction is performed to obtain dimensionless abnormality quantities, including pressure abnormality quantity PA(t), humidity climb rate HCR(t), strain dispersion SD(t) and vibration root mean square VR(t), to generate a four-dimensional feature vector z(t) = [PA, HCR, SD, VR] about the battery pack monitoring information;

[0152] The data in the four-dimensional feature vector is normalized and then substituted into a weighted fusion model to calculate a health index SHI, which is set as the comprehensive analysis result, and is defined as follows:

[0153] SHI(t) = αPA n (t) + βHCR n (t) + γSD n (t) + δVR n (t)

[0154] wherein PA n (t), HCR n (t), SD n (t) and VR n (t) are respectively the normalized data of the pressure abnormality quantity, humidity climb rate, strain dispersion and vibration root mean square obtained at the sampling time t, and α, β, γ and δ are respectively weight coefficients, and α + β + γ + δ = 1.

[0155] As a preferred implementation option, preferably, the comprehensive analysis result is judged or is jointly judged with the data obtained by a single sensor in the battery pack monitoring information to realize abnormal recognition of the battery pack and to generate early warning level information including one of the following:

[0156] (1) when the duration that the health index SHI is greater than a first threshold value exceeds a preset duration, an abnormal event is recorded for directional inspection during maintenance;

[0157] (2) the health index SHI is combined with humidity monitoring data, when the health index SHI is greater than a second threshold value and the humidity monitoring data indicates that the humidity rises by more than 3% within 24 hours, the power of the battery pack is limited, and repair information is pushed;

[0158] (3) the health index SHI is combined with strain monitoring data, when the health index SHI is greater than a third threshold value and the strain monitoring data indicates that the stress mutation is > 100 pa / s, a safety mode is started, and a suggestion to deactivate information is pushed.

[0159] The present scheme about battery pack monitoring information collection and using battery pack monitoring information to judge the state of the battery pack is described below through an example.

[0160] Among them, the MEMS pressure sensor is four, which is distributedly arranged at four corners inside the box of the battery pack, and the other sensors are one.

[0161] When the working state of the vehicle is monitored in response to the working instruction signal of the vehicle, the airtightness and structural safety of the battery pack on the vehicle are also monitored by the sensor group to generate battery pack monitoring information; then the pressure monitoring data, humidity monitoring data, strain monitoring data and vibration monitoring data in the battery pack monitoring information are extracted in time sequence to generate time sequence data x;

[0162] The time sequence data x is expressed as:

[0163]

[0164] Among them, t is the sampling time point of the sensor, P1(t), P2(t), P3(t), P4(t) are the sampling values of the four MEMS pressure sensors at t, and the unit is pa; RH(t) is the sampling value of the nano-humidity sensor at t, and the unit is %RH; ε(t) is the sampling value of the optical fiber strain sensor at t, and the unit is με; a(t) is the sampling value of the piezoelectric accelerometer at t, and the unit is m / s 2 .

[0165] Among them, the sampling beat of the sensor is constrained as follows:

[0166] T sync = lcm(1, 10, 0.01, 0.001) = 1s

[0167] T sync The minimum common multiple sampling period is 1s, and the sampling values of all sensors are aligned at 1s node by linear interpolation or zero-order hold to form a multi-dimensional vector, so as to generate an equidistant time interval sequence, which is expressed as: {x(0), x(1), x(2), … x(t)}.

[0168] In order to improve the data quality, the data in the multi-dimensional vector is preprocessed to remove drift and synchronize alignment.

[0169] The scheme removes drift by sliding mean method, which is defined as follows:

[0170]

[0171] Among them, k = 1, 2, … 7; The data after removing drift is x k(t) is the kth data channel in the time series data x, which corresponds to P1(t), P2(t), P3(t), P4(t), RH(t), ε(t), a(t) in turn; w is the window width (which can be 3600s, corresponding to 1h data).

[0172] On the basis of the above, the scheme also suppresses outliers through median filtering, which is defined as follows:

[0173]

[0174] Wherein, L is the window length, which can take the value of 9-11 points.

[0175] After the above processing, the cleaned data sequence Each component is used for subsequent feature calculation processing.

[0176] Based on the data sequence Feature extraction is performed to obtain dimensionless anomaly quantity, which is as follows:

[0177] 1. Pressure anomaly quantity

[0178]

[0179] Wherein, P ref is the steady-state average pressure of the battery pack after 10min; PA(t) is the relative pressure offset; is the average value of the data obtained by the four MEMS pressure sensors after preprocessing; is the data value obtained by one of the four MEMS pressure sensors after preprocessing;

[0180] 2. Humidity climb rate

[0181]

[0182] Wherein, Δt=24h, RHmax(t, Δt), RHmin(t, Δt) are the maximum and minimum values of humidity in the past 24h; the unit of HCR(t) is %RH / h;

[0183] 3. Strain dispersion

[0184]

[0185] Wherein, N=10×f ε , f ε =10Hz, SD(t) is the standard deviation of strain in a 1s time window, whose dimension is με, is the average value of the data obtained by the strain sensor in a time window after preprocessing; The data value of the data collected at t-i after pre-processing.

[0186] 4. The root mean square of vibration

[0187]

[0188] Wherein, M=10xf a , f a =100Hz, VR(t) is the root mean square of vibration in 1s.

[0189] By collecting the characteristic data, a four-dimensional feature vector z(t)=[PA, HCR, SD, VR] about time series is formed.

[0190] The characteristic data in the four-dimensional feature vector z(t) is subjected to Min-Max normalization processing, which is defined as follows:

[0191]

[0192] Wherein, z i，max , z i，min are the maximum and minimum values in the characteristic data respectively, z i (t) is the characteristic data, which corresponds to PA, HCR, SD, VR in the four-dimensional feature vector z(t).

[0193] After the data in the four-dimensional feature vector is subjected to normalization processing, the data PA n (t), HCR n (t), SD n (t), VR n (t) of the pressure monitoring data, the humidity monitoring data, the strain monitoring data and the vibration monitoring data under different sampling time points after normalization processing of the pressure anomaly, the humidity climbing rate, the strain dispersion and the root mean square of vibration are obtained, and then substituted into the weighted fusion model to calculate the health index SHI, which is set as the comprehensive analysis result, which is defined as follows:

[0194] SHI(t)=αPA n (t)+βHCR n (t)+γSD n (t)+δVR n (t)

[0195] Wherein, PA n (t), HCR n (t), SD n (t), VR n(t) are the normalized data of pressure anomaly, humidity climbing rate, strain dispersion and vibration root mean square obtained at sampling time t, α, β, γ and δ are weight coefficients, which are preset values, and α + β + γ + δ = 1.

[0196] The above contents of this scheme are explained below in combination with three sets of data, which are as follows:

[0197]

[0198] Note: Strain dispersion and vibration root mean square are directly processed data.

[0199] Taking 36h data as an example, the calculation results are as follows:

[0200] 1. Average pressure

[0201]

[0202] 2. Abnormal pressure

[0203]

[0204] Among them, P ref is the steady-state average pressure 10 minutes after the battery pack is powered on. In this example, it is assumed to be 100 kPa. PA(t) is the relative pressure offset.

[0205] 3. Humidity rise rate

[0206]

[0207] Where Δt = 24h, RHmax(t, Δt) and RHmin(t, Δt) are the maximum and minimum humidity values ​​in the past 24h, respectively; the unit of HCR(t) is %RH / h.

[0208] 4. Strain discreteness

[0209] SD(t)=20

[0210] 5. Vibration root mean square

[0211] VR(t)=0.80

[0212] By gathering the characteristic data, a four-dimensional characteristic vector z(t)=[PA, HCR, SD, VR] of the time series is formed.

[0213] In this scheme, in order to simplify the statistical process, the range of pressure anomaly is assumed to be [0, 0.1], that is, PA max =0.1, PA min =0; the range of humidity climb rate is assumed to be [0, 2], that is, HCRmax = 2, HCR min = 0; the range of strain dispersion is assumed to be [0, 100], i.e. SD max = 100, SD min = 0; the range of vibration root mean square is assumed to be [0, 2], i.e. VR max = 2, VR min = 0. The Min-Max normalization processing is performed on the feature data in the four-dimensional feature vector z(t), which is defined as follows:

[0214]

[0215] After the normalization processing is performed on the data in the four-dimensional feature vector, the data is substituted into the weighted fusion model to calculate the health index SHI, which is set as the comprehensive analysis result, and the definition is as follows:

[0216] SHI(t) = aPA n (t) + bHCR n (t) + gSD n (t) + dVR n (t)

[0217] = 0.3 x 0.5 + 0.25 x 0.42 + 0.25 x 0.2 + 0.2 x 0.4 = 0.385

[0218] Wherein, a, b, g, d are respectively set as 0.3, 0.25, 0.25, 0.2.

[0219] In the comprehensive analysis structure judgment, the comprehensive analysis result is judged or combined with the data obtained by the single sensor monitoring in the battery pack monitoring information to realize the abnormal recognition of the battery pack.

[0220] Wherein, taking the first threshold value as 0.3, the second threshold value as 0.5, and the third threshold value as 0.8 as an example, the generated early warning level information includes one of the following:

[0221] (1) When the health index SHI is greater than the first threshold value (0.3) for more than the preset time length 6h, record the abnormal event for directional check when maintaining;

[0222] (2) Combine the health index SHI with the humidity monitoring data, when the health index SHI is greater than the second threshold value (0.5), and the humidity monitoring data indicates that the humidity rises more than 3% within 24h, limit the power of the battery pack, and push the repair information at the same time;

[0223] (3) The health index SHI is combined with the strain monitoring data. When the health index SHI is greater than a third threshold (0.8) and the strain monitoring data indicates a stress mutation > 100 pa / s, the safety mode is started, and the suggestion of stopping working is pushed.

[0224] And combined with the data in the foregoing table, it can be known that the health index SHI is 0.106, 0.385, and 0.753 in turn over time. It can be known from this that the "health" condition of the battery pack presents signs of deterioration. Therefore, the battery management BMS system can be combined to take measures such as power-off or power limitation on the abnormal battery pack, so as to improve the safety and reliability of the battery pack in use.

[0225] Based on the above, the embodiment further provides a battery working management method for an electric vehicle, which comprises the identification method for a fault battery in a vehicle battery pack as described above, and further comprises:

[0226] When the identification information indicates that there is a fault battery in the battery pack, the fault battery information is acquired, the power supply access of the fault battery is cut off or shielded, and meanwhile, the output power range that can be provided when the battery pack accesses the vehicle for power supply work is updated;

[0227] According to the vehicle monitoring data, the power supply power range required by the power-consuming devices of the vehicle is determined;

[0228] According to the power supply power range required by the power-consuming devices of the vehicle, in combination with the output power range that can be provided when the battery pack accesses the vehicle for power supply work, the redundant battery pack accesses the vehicle for power supply compensation according to a preset condition.

[0229] In the present scheme, the fault battery is located through the identification result, and then its power supply is shielded in time, and meanwhile, the redundant battery pack is timely intervened in the power supply of the vehicle, which can provide a positive and practical strategy guarantee for the reliable and stable working of the vehicle.

[0230] In combination with Figure 2 the above, the embodiment further provides an identification system for a fault battery in a vehicle battery pack, which comprises:

[0231] A battery monitoring module is configured to monitor the working data of the batteries in the battery pack in response to the information that the battery pack accesses the vehicle for power supply work, and generate battery monitoring data;

[0232] A vehicle monitoring module is configured to monitor the working state of the vehicle in response to the working instruction signal of the vehicle, and generate vehicle monitoring data;

[0233] An information labeling module is configured to acquire vehicle monitoring data, label the state of the vehicle at different times according to preset conditions, and generate vehicle state information labeled in time sequence.

[0234] A data screening module is configured to acquire battery monitoring data, and determine target feature data about the battery by combining the vehicle state information at the same time sequence.

[0235] A data processing module is configured to determine whether the battery in the battery pack has a fault based on the target feature data, and generate identification information accordingly.

[0236] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0237] If the integrated unit is realized in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application, essentially or the part that contributes to the prior art, or all or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to execute all or part of the steps of the embodiments of the method of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0238] The above only describes some embodiments of the present application, and does not limit the protection scope of the present application, and any equivalent device or equivalent process transformation, or direct or indirect application in other related technical fields, is also included in the patent protection scope of the present application.

Claims

1. A method for identifying a faulty battery in a vehicle battery pack, characterized in that: It includes: In response to information that the battery pack is connected to the vehicle to provide power, monitoring operating data of batteries in the battery pack to generate battery monitoring data; Responding to the working instruction signal of the vehicle, monitoring its working status and generating vehicle monitoring data; Obtain vehicle monitoring data, annotate the vehicle status at different times according to preset conditions, and generate vehicle status information annotated in time series; Obtain battery monitoring data and combine it with vehicle status information in the same time series to determine target characteristic data about the battery; Based on the target characteristic data, it is determined whether a battery in the battery pack has a fault, and identification information is generated accordingly.

2. The method for identifying a faulty battery in a vehicle battery pack according to claim 1, wherein: When the battery pack is connected to the vehicle for power supply, a battery sampling module installed on the power supply line of the battery pack senses the connection of the battery pack, and the battery sampling module is activated by power supplied by the battery pack; There are more than one battery pack connected to the vehicle, each of which corresponds to a power supply line, and the battery pack includes a plurality of battery cells composed of one or more single batteries, and each of the battery cells is assigned a unique ID; The battery sampling module is a multi-channel module that samples the temperature, state of charge, current and / or voltage data of the battery cell; The battery sampling module collects data from different battery cells and associates them with the ID and sampling time of the corresponding battery cell to form a data group; the data groups of several battery cells are aggregated to form battery monitoring data; The vehicle monitoring data includes operating status data of power-consuming devices of the vehicle, and the power-consuming devices include one or more of an electric motor, an air conditioner, a lamp, a screen, a sensor, and an electronically controlled brake installed on the vehicle.

3. The method for identifying a faulty battery in a vehicle battery pack according to claim 1 or 2, wherein: In response to the vehicle's work command signal and monitoring its working status, the sensor group also monitors the air tightness and structural safety of the battery pack on the vehicle and generates battery pack monitoring information; The battery pack is encapsulated in a box-shaped housing, which is provided with an explosion-proof valve and a connector, and the sensor group includes one or more of a MEMS pressure sensor, a nanometer humidity sensor, an optical fiber strain sensor, and a piezoelectric accelerometer; There are multiple MEMS pressure sensors, which are distributed at the four corners of the battery pack body and are used to monitor the pressure changes in the battery pack. The sampling frequency is 1 Hz, and corresponding pressure monitoring data is generated; There is one or more nano humidity sensors, which are arranged near the battery pack box connected to the explosion-proof valve or connector, and are used to monitor the relative humidity of the battery pack. The sampling frequency is 0.1 Hz, and the humidity monitoring data is generated accordingly; The optical fiber strain sensor is set at the welding seam of the battery pack box and is used to monitor the micro-strain of the box. Its sampling frequency is 10Hz, and corresponding strain monitoring data is generated; The piezoelectric accelerometer is arranged in the box of the battery pack and is used to monitor the vibration of the battery pack. The sampling frequency is 100 Hz and the corresponding vibration monitoring data is generated; The pressure monitoring data, humidity monitoring data, strain monitoring data and vibration monitoring data are collected to form multi-source data, which is then included in the battery pack monitoring information, which also includes the preset ID information of the corresponding battery pack; The identification method further includes: Based on the battery pack monitoring information, a comprehensive analysis is performed to generate a comprehensive analysis result, which is then judged or jointly judged with the data obtained from single sensor monitoring in the battery pack monitoring information to realize abnormal identification of the battery pack and generate early warning level information to guide the initiation of response measures for the vehicle battery pack.

4. The method for identifying a faulty battery in a vehicle battery pack according to claim 3, wherein: Based on the battery pack monitoring information, a comprehensive analysis is performed to generate comprehensive analysis results including: Extract pressure monitoring data, humidity monitoring data, strain monitoring data, and vibration monitoring data from the battery pack monitoring information in time series to generate time series data x; Preprocess the time series data x to remove drift and synchronize alignment to obtain cleaned time series data For time series data Perform feature extraction to obtain dimensionless anomalies, including pressure anomaly PA(t), humidity climb rate HCR(t), strain dispersion SD(t), and vibration root mean square VR(t), and generate a four-dimensional feature vector z(t) = [PA, HCR, SD, VR] for battery pack monitoring information; The data in the four-dimensional feature vector is normalized and then substituted into the weighted fusion model to calculate the health index SHI, which is set as the comprehensive analysis result and is defined as follows: SHI(t)=αPA n (t)+βHCR n (t)+γSD n (t)+δVR n (t) Among them, PA n (t), HCR n (t), SD n (t), VR n (t) are the normalized data of pressure anomaly, humidity climbing rate, strain dispersion and vibration root mean square obtained at sampling time t, α, β, γ and δ are weight coefficients, and α+β+γ+δ=1; The comprehensive analysis results are judged or combined with the data obtained from the single sensor monitoring in the battery pack monitoring information to realize the abnormal identification of the battery pack and generate warning level information including one of the following: (1) When the health index SHI is greater than the first threshold for a period exceeding a preset time, the abnormal event is recorded for targeted inspection during maintenance; (2) The health index SHI is combined with the humidity monitoring data. When the health index SHI is greater than the second threshold and the humidity monitoring data indicates that the humidity has risen by more than 3% within 24 hours, the power of the battery pack is limited and maintenance information is pushed; (3) The health index SHI is combined with the strain monitoring data. When the health index SHI is greater than the third threshold and the strain monitoring data indicates a stress mutation greater than 100 Pa / s, the safety mode is activated and a deactivation recommendation message is pushed.

5. The method for identifying a faulty battery in a vehicle battery pack according to claim 2, wherein: Obtain vehicle monitoring data, annotate the vehicle status at different times according to preset conditions, and generate vehicle status information annotated in time series, including: Obtain vehicle monitoring data and arrange it in time series. Then, based on the on / off status of the vehicle's power-consuming devices in the time series and the power data of the power-consuming devices during operation, calculate the total power demand of the vehicle's power-consuming devices at different time points. The formula is as follows: IN t =in 1t +in 2t +…+in nt Among them, W t is the total power demand of the car's power-consuming devices at time t, w 1t 、w 2t …w nt are the power consumption of different power devices within a preset range at time point t; Based on the total power demand of the vehicle's power-consuming devices at different time points, they are classified into levels 1 to 6 and labeled with information according to preset power interval values. Then, the vehicle monitoring data with the same level of total power demand in the same time period is simplified into the relationship between the corresponding level of total power demand and the time interval, generating vehicle status information labeled in a time series.

6. The method for identifying a faulty battery in a vehicle battery pack according to claim 5, wherein: Obtain battery monitoring data and combine it with vehicle status information in the same time series to determine the target characteristic data about the battery, including: Obtain battery monitoring data and arrange it in time series; According to the time series, retrieve the vehicle status information in the same time series as the battery monitoring data; Based on the vehicle status information in the same time series, the data groups in the battery monitoring data are screened for abnormal data through residual analysis, data cluster analysis or regression analysis to determine target feature data about the battery for judging battery failure. The target feature data includes the data group whose judgment result is abnormal, and its corresponding time series and vehicle status information.

7. The method for identifying a faulty battery in a vehicle battery pack according to claim 6, wherein: Based on the target characteristic data, determining whether a battery in the battery pack is faulty and generating corresponding identification information includes: Importing target feature data into the trained detection neural network for detection, and having the detection neural network output the detection results; Get the test results, When the detection result indicates that the battery corresponding to the target characteristic data is a faulty battery, fault information is generated. Then, based on the data group of the target characteristic data, the ID of the corresponding battery cell is obtained. The fault information, the target characteristic data and the ID of the corresponding battery cell are output as identification information. When the detection result indicates that the battery corresponding to the target characteristic data is a normal battery, the ID of the corresponding battery cell is obtained according to the data group of the target characteristic data, the number of times the battery cell corresponding to the ID is judged to be abnormal is recorded, and the number of abnormal times information is generated. Then, the number of abnormal times information, the target characteristic data and the ID of its corresponding battery cell are output as identification information.

8. The method for identifying a faulty battery in a vehicle battery pack according to claim 7, wherein: When the detection result indicates that the battery corresponding to the target characteristic data is a normal battery, the corresponding abnormal number information N is retrieved according to the ID corresponding to the data group of the target characteristic data. When it is greater than the preset threshold M, manual re-inspection information is generated and associated with the identification information and output together.

9. A method for managing the operation of an electric vehicle battery, characterized in that: It includes the method for identifying a faulty battery in a vehicle battery pack according to any one of claims 1 to 8, and further includes: Acquire identification information, and when it indicates that there is a faulty battery in the battery pack, obtain information about the faulty battery, cut off or block its power supply access, and at the same time, update the output power range that can be provided by the battery pack when connected to the vehicle for power supply; Determining the power supply range required by the power-consuming devices of the vehicle based on the vehicle monitoring data; Based on the power supply range required by the power-consuming devices of the vehicle and the output power range that can be provided by the battery pack when connected to the vehicle for power supply, the redundant battery pack is started according to preset conditions to connect to the vehicle for power supply compensation.

10. A system for identifying faulty batteries in a vehicle battery pack, characterized in that: It includes: A battery monitoring module, configured to monitor the working data of the batteries in the battery pack in response to information that the battery pack is connected to the vehicle for power supply, and generate battery monitoring data; The vehicle monitoring module is used to respond to the working instruction signal of the vehicle, monitor its working status and generate vehicle monitoring data; The information labeling module is used to obtain vehicle monitoring data, label the vehicle status at different times according to preset conditions, and generate vehicle status information labeled in time series; A data screening module is used to obtain battery monitoring data and combine it with vehicle status information in the same time series to determine target characteristic data about the battery; The data processing module is used to determine whether there is a fault in the battery in the battery pack based on the target characteristic data and generate identification information accordingly.

Images