Training method for capacity retention ratio determination model, application method for capacity retention ratio determination model, and apparatus and storage medium

Abstract

The present disclosure relates to the technical field of vehicles. Provided are a training method for a capacity retention ratio determination model, an application method for a capacity retention ratio determination model, and an apparatus and a storage medium. The training method comprises: acquiring a plurality of raw data sets of a target battery, wherein each raw data set comprises the following parameters of the target battery when a first standing duration condition is met: the usage time of the target battery, usage mileage, a standing time, parameter information of a standing temperature, and a capacity retention ratio of the target battery; constructing a training sample set on the basis of the plurality of raw data sets; and substituting the training sample set into a preset neural network model, and performing training to acquire a battery capacity retention ratio determination model, wherein the battery capacity retention ratio determination model is used for determining a battery capacity retention ratio on the basis of battery data. The battery capacity retention ratio determination model obtained by using the model training method in the present disclosure can automatically determine a battery capacity retention ratio, thereby improving the efficiency of acquiring a battery capacity retention ratio.

Description

Training, application method, device and storage medium of capacity retention rate judgment model

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This disclosure claims priority to Chinese patent application number 202410064130.5 filed with the State Intellectual Property Office of China on January 17, 2024, entitled “Training, application method, device and storage medium for capacity retention rate judgment model,” the entire contents of which are incorporated by reference into this disclosure. Technical Field

[0003] The present disclosure relates to the field of vehicle technology, and in particular to a training and application method, device, and storage medium for a capacity retention rate judgment model. Background Art

[0004] The capacity retention rate of new energy vehicle power batteries refers to the capacity retention ratio obtained after the power battery has been stationary for a long time compared with the capacity before the stationary state. The capacity retention rate is an important indicator to measure the performance of power batteries.

[0005] With the increase in new energy vehicles, the functional performance verification of power batteries is becoming increasingly important. However, at present, the method of obtaining the capacity retention rate of power batteries is mainly through laboratory testing and confirmation. Due to the limitations of test equipment and long test cycles, there is a problem of low efficiency.

[0006] Application Contents

[0007] In response to the above-mentioned deficiencies in the prior art, the present disclosure provides a method, an application method, an apparatus and a storage medium for training and applying a capacity retention rate judgment model, so as to solve the problems existing in the prior art.

[0008] The technical solutions adopted in the embodiments of the present disclosure are as follows:

[0009] In a first aspect, an optional embodiment of the present disclosure provides a method for training a capacity retention rate judgment model, comprising:

[0010] Acquire multiple raw data sets of a target battery; wherein each of the raw data sets includes the following parameters of the target battery when the first resting time condition is satisfied: parameter information of the target battery's usage time, usage mileage, resting time, and resting temperature, and a capacity retention rate of the target battery;

[0011] Constructing a training sample set based on the multiple original data sets;

[0012] The training sample set is substituted into a preset neural network model to train and obtain a battery capacity retention rate judgment model, wherein the battery capacity retention rate judgment model is used to determine the battery capacity retention rate according to the battery data.

[0013] In an optional embodiment, the acquiring of multiple original data sets of the target battery includes:

[0014] Collect multiple frames of raw data from multiple vehicles equipped with the target battery in corresponding preset time sequences; wherein each frame of raw data includes: a vehicle identification corresponding to the target battery, a time when the data was collected, a voltage of the target battery, a vehicle mileage at the time of data collection, and temperature information;

[0015] Acquire multiple long-term static events based on multiple frames of raw data corresponding to each of the vehicles; the long-term static events are used to indicate that the target battery meets the first static time;

[0016] Calculate a first remaining capacity and a second remaining capacity corresponding to the target battery at the beginning and end of a plurality of first rest periods, respectively;

[0017] A plurality of raw data sets of the target battery are constructed according to a plurality of the first remaining capacities, a plurality of the second remaining capacities, and a plurality of frames of the raw data.

[0018] In an optional embodiment, acquiring multiple long-term stationary events based on multiple frames of raw data corresponding to each of the vehicles includes:

[0019] Segmenting the multiple frames of raw data to obtain multiple events corresponding to the target battery; wherein each event includes at least: event start time, event end time, event type, event start mileage, event end mileage, and the voltage and temperature of the target battery when the event occurs;

[0020] According to the time difference between each two adjacent events and the mileage difference between each two adjacent events, it is determined that the events whose time difference and the mileage difference meet preset thresholds meet the first static duration.

[0021] In an optional embodiment, segmenting the multiple frames of raw data to obtain multiple events corresponding to the target battery includes:

[0022] According to the connection information of the vehicle main relay and the connection information of the vehicle charging gun corresponding to the multiple frames of the original data, the multiple frames of the original data are segmented according to multiple preset events to obtain multiple events corresponding to the target battery.

[0023] In an optional embodiment, before constructing multiple raw data sets of the target battery based on the multiple first remaining capacities, the multiple second remaining capacities, and the multiple frames of raw data, the method further includes:

[0024] Calculating and obtaining a cumulative time error parameter according to a sampling frequency of the original data of the plurality of frames;

[0025] Filtering multiple frames of raw data according to the accumulated time error parameter to obtain multiple frames of target data;

[0026] The step of constructing a plurality of raw data sets of the target battery according to the plurality of first remaining capacities, the plurality of second remaining capacities, and the plurality of frames of raw data includes:

[0027] A plurality of original data sets of the target battery are constructed according to a plurality of the first remaining capacities, a plurality of the second remaining capacities, and a plurality of frames of the target data.

[0028] In an optional embodiment, after collecting multiple frames of raw data of multiple vehicles equipped with the target battery in corresponding preset time sequences, the method further includes:

[0029] Cleaning the multiple frames of raw data to obtain cleaned raw data;

[0030] The segmenting of the multiple frames of raw data to obtain multiple events corresponding to the target battery includes:

[0031] The cleaned raw data is segmented to obtain multiple events corresponding to the target battery.

[0032] In an optional embodiment, the calculating of the first remaining capacity and the second remaining capacity corresponding to the target battery at the beginning and end of the plurality of first rest periods, respectively, includes:

[0033] Based on information about a target standby event before the start of the first standby period and multiple events corresponding to the target battery, a first remaining capacity corresponding to the target battery is determined, and the battery capacity at the end of the first standby period is used as the second remaining capacity, wherein the target standby event is an event that satisfies the second standby period, and the second standby period is less than the first standby period.

[0034] In an optional embodiment, the step of using the battery capacity at the end of the first rest period as the second remaining capacity includes:

[0035] An open-circuit voltage interpolation method is used to calculate and obtain a second remaining capacity corresponding to the target battery according to the voltage and temperature of the target battery at the end of the first rest period.

[0036] In an optional embodiment, substituting the training sample set into a preset neural network model to train and obtain a battery capacity retention rate judgment model includes:

[0037] performing sample feature normalization on the training sample set;

[0038] The preset neural network model is trained according to the normalized training sample set, activation function, and optimization function to obtain the battery capacity retention rate judgment model.

[0039] In a second aspect, another optional embodiment of the present disclosure provides an application method of a capacity retention rate judgment model, comprising:

[0040] Acquiring battery data of the battery to be tested, wherein the battery data of the battery to be tested includes: usage time, usage mileage, long-term static time, and static temperature;

[0041] Substitute the battery data of the battery to be tested into the battery capacity retention rate judgment model described in any of the above embodiments to obtain the battery capacity retention rate of the battery to be tested.

[0042] In a third aspect, another optional embodiment of the present disclosure provides a training device for a capacity retention rate judgment model, comprising:

[0043] an acquisition module, configured to acquire multiple raw data sets of a target battery; wherein each of the raw data sets includes the following parameters of the target battery when the first resting time condition is satisfied: parameter information of the target battery's usage time, usage mileage, resting time, and resting temperature, and a capacity retention rate of the target battery;

[0044] A construction module, configured to construct a training sample set based on the plurality of original data sets;

[0045] The training module is used to substitute the training sample set into a preset neural network model to train and obtain a battery capacity retention rate judgment model, wherein the battery capacity retention rate judgment model is used to determine the battery capacity retention rate based on battery data.

[0046] In a fourth aspect, another optional embodiment of the present disclosure provides an electronic device, comprising: a processor, a storage medium and a bus, wherein the storage medium stores program instructions executable by the processor. When the electronic device is running, the processor and the storage medium communicate through the bus, and the processor executes the program instructions to implement the training method of the capacity retention rate judgment model described in any of the above embodiments.

[0047] In a fifth aspect, another optional embodiment of the present disclosure provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the training method of the capacity retention rate judgment model described in any of the above embodiments is implemented.

[0048] The beneficial effects of the present disclosure are: the present disclosure provides a training, application method, device and storage medium for a capacity retention rate judgment model, the training method of the capacity retention rate judgment model includes obtaining multiple original data sets of a target battery; wherein each original data set includes the following parameters of the target battery under a first rest time condition: parameter information of the target battery's usage time, usage mileage, rest time, and rest temperature, and the capacity retention rate of the target battery; based on the multiple original data sets, a training sample set is constructed; the training sample set is substituted into a preset neural network model to train and obtain a battery capacity retention rate judgment model, and the battery capacity retention rate judgment model is used to determine the battery capacity retention rate based on the battery data.

[0049] Among them, the battery capacity retention rate is determined according to the battery capacity retention rate judgment model obtained by training, which realizes the automatic acquisition of the battery capacity retention rate and improves the efficiency of obtaining the battery capacity retention rate. Secondly, compared with the prior art in which staff conduct tests and confirmations on the battery capacity retention rate through laboratories, the present disclosure automatically obtains the battery capacity retention rate, reducing manpower and material costs. Finally, since the capacity retention rate refers to the capacity ratio of the battery after long-term static state to the capacity ratio before long-term static state, the present disclosure uses the original data set that meets the long-term static time to train the preset neural network model, so that the obtained battery capacity retention rate judgment model can be used to determine the capacity retention rate of the battery to be tested. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present disclosure 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.

[0051] FIG1 is a flow chart showing a method for training a capacity retention rate determination model according to an optional embodiment of the present disclosure;

[0052] FIG2 is a second flow chart of a method for training a capacity retention rate determination model according to an optional embodiment of the present disclosure;

[0053] FIG3 is a third flow chart of a method for training a capacity retention rate judgment model according to an optional embodiment of the present disclosure;

[0054] FIG4 is a fourth flow chart of a method for training a capacity retention rate judgment model according to an optional embodiment of the present disclosure;

[0055] FIG5 is a fifth flow chart of a method for training a capacity retention rate determination model according to an optional embodiment of the present disclosure;

[0056] FIG6 is one of the schematic diagrams of multiple events in a time series provided by the present disclosure;

[0057] FIG7 is a second schematic diagram of multiple events in a time series provided by the present disclosure;

[0058] FIG8 is a sixth flow chart of a method for training a capacity retention rate determination model according to an optional embodiment of the present disclosure;

[0059] FIG9 is a flow chart of a method for applying a capacity retention rate judgment model according to an optional embodiment of the present disclosure;

[0060] FIG10 is a schematic structural diagram of a training device for a capacity retention rate judgment model according to an optional embodiment of the present disclosure;

[0061] FIG11 is a schematic structural diagram of an electronic device provided in an optional embodiment of the present disclosure. DETAILED DESCRIPTION

[0062] In order to make the purpose, technical solutions and advantages of the embodiments of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in combination with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, not all of the embodiments.

[0063] Therefore, the following detailed description of the embodiments of the present disclosure provided in the accompanying drawings is not intended to limit the scope of the present disclosure as claimed, but merely represents selected embodiments of the present disclosure. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present disclosure without creative effort shall fall within the scope of protection of the present disclosure.

[0064] In addition, the terms "first," "second," and the like in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or precedence. It should be understood that the numbers used in this way are interchangeable where appropriate so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having," as well as any variations thereof, are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or apparatus comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products, or apparatus.

[0065] It should be noted that, in the absence of conflict, the features in the embodiments of the present disclosure may be combined with each other.

[0066] The embodiments of the present disclosure provide a method for training a capacity retention rate judgment model. The method can be generated by any electronic device with processing and computing capabilities. The electronic device can be, for example, a terminal-oriented electronic device or a back-end server.

[0067] The following specifically illustrates the method for training the capacity retention rate judgment model provided by the present disclosure through multiple examples in conjunction with the accompanying drawings.

[0068] FIG1 is a flow chart of a method for training a capacity retention rate judgment model according to an optional embodiment of the present disclosure. As shown in FIG1 , the method includes:

[0069] S101. Acquire multiple original data sets of a target battery.

[0070] Before training the battery capacity retention rate judgment model, it is necessary to obtain multiple raw data sets of the target battery. These data sets are used to construct the training sample set. The data types contained in the raw data sets are shown in Table 1.

[0071] Table 1 Fields included in each original dataset

[0072] As shown in Table 1, among the multiple original data sets, each original data set includes the following parameters of the target battery when the first rest time condition is met: parameter information of the target battery's usage time, usage mileage, rest time, and rest temperature, and the capacity retention rate of the target battery.

[0073] In an optional embodiment, since the capacity retention rate refers to the capacity retention ratio obtained after the power battery has been stationary for a long time compared with the capacity before the stationary state, correspondingly, if the trained battery capacity retention rate judgment model is to be used to determine the battery capacity retention rate, the value of the first stationary time is used to indicate the long-term stationary time. Usually, the long-term stationary time refers to the battery stationary time being greater than 10 days. The first stationary time of this embodiment can be set to be greater than 10 days, that is, the data in the original data set is the data collected when the target battery has been stationary for more than 10 days, and each original data set corresponds to a long-term stationary event of the target battery.

[0074] Among them, the usage time of the target battery refers to the interval between the time of data collection and the time when the target battery was shipped from the factory, the usage mileage refers to the total mileage of the vehicle corresponding to the target battery when the data was collected, the static time refers to the time difference before and after the long-term static event, the parameter information of the static temperature refers to the temperature of the target battery during the long-term static event, and the capacity retention rate of the target battery refers to the capacity retention ratio of the target battery after the long-term static event compared with the capacity before the long-term static event.

[0075] S102: Construct a training sample set based on multiple original data sets.

[0076] After obtaining multiple original data sets, a training sample set can be obtained based on the multiple original data sets.

[0077] In an optional embodiment, a portion of the original data set can be used as a training sample set and another portion as a test set, for example, 75% of the original data set can be used as a training sample set and 25% can be used as a test set. The training sample set can be, for example, a text file, and the number of original data sets in the training sample set is at least 1,000. A sufficiently large number of original data sets can ensure the training effect, so that the resulting battery capacity retention rate judgment model has a higher accuracy in judging the battery capacity retention rate.

[0078] In the training sample set, the structure of each original data set can be, for example, {(X1, Y1), (X2, Y2), (X3, Y3), …, (X n ,Y n )}, and any one of the samples is recorded as (X i ,Y i ), X i is characterized by the form (X i 1 ,X i 2 ,X i 3 ,X i 4 ), the four features are the target battery usage time, target battery usage mileage, target battery long-term static time, and target battery static temperature, corresponding to which, Y i is the label of the sample, that is, the capacity retention rate of the target battery.

[0079] S103: Substitute the training sample set into a preset neural network model to obtain a battery capacity retention rate judgment model through training.

[0080] The preset neural network model can be trained by substituting the training sample set into the preset neural network model. Since the training sample set contains parameter information of the target battery's usage time, usage mileage, rest time, and rest temperature, and the target battery's capacity retention rate, the trained preset neural network model is correspondingly a battery capacity retention rate judgment model, which can be used to determine the battery's capacity retention rate based on the corresponding data during the training process, such as the battery's usage time, usage mileage, rest time, and rest temperature.

[0081] In summary, this embodiment provides a training method for a capacity retention rate judgment model. The battery capacity retention rate judgment model trained according to this method can be used to determine the battery capacity retention rate, thereby realizing automatic acquisition of the battery capacity retention rate and improving the efficiency of obtaining the battery capacity retention rate. Secondly, compared with the prior art in which staff conduct tests and confirmation of the battery capacity retention rate through laboratories, this embodiment automatically obtains the battery capacity retention rate, reducing manpower and material costs. Finally, since the capacity retention rate refers to the capacity ratio of the battery after long-term static storage to the capacity ratio before long-term static storage, this embodiment uses an original data set that meets the long-term static time to train a preset neural network model, so that the obtained battery capacity retention rate judgment model can be used to determine the capacity retention rate of the battery to be tested.

[0082] The following describes a specific implementation method for obtaining the original data set in conjunction with Figures 2 to 6. Figure 2 is a second flow chart of a method for training a capacity retention rate judgment model provided in an optional embodiment of the present disclosure. As shown in Figure 2, the step of obtaining multiple original data sets of a target battery in S101 may include:

[0083] S201 : Collect multiple frames of raw data of multiple vehicles equipped with target batteries in corresponding preset time sequences.

[0084] In the disclosed embodiment, this can be achieved based on high-frequency data from new energy vehicles. The data acquisition frequency must be at least greater than 1 Hz to ensure accurate calculation of important parameters. In an optional embodiment, multiple frames of raw data are first collected from multiple vehicles equipped with target batteries in corresponding preset time series. The specific types of raw data collected are shown in Table 2.

[0085] Table 2 Fields included in each frame of raw data

[0086] As shown in Table 2, each frame of raw data includes at least: the vehicle identification (Vehicle Serial Number) corresponding to the target battery, the time when the data was collected (Current Time), the voltage of the target battery (Voltage Sampling List), the vehicle mileage (Mileage) when the data was collected, and the temperature information (Temperature Sampling List).

[0087] S202 : Acquire multiple long-term stationary events based on multiple frames of original data corresponding to each vehicle.

[0088] To improve computational efficiency, multiple frames of raw data are divided into events so that only corresponding segments are selected for subsequent processing, saving computational resources and improving computational efficiency. It is understood that each raw data set described in the above embodiment corresponds to a long-term static event, and each raw data set includes more than one raw data frame.

[0089] After collecting multiple frames of raw data, this embodiment divides the multiple frames of raw data into multiple events, and obtains multiple long-term static events therefrom. The long-term static events are used to indicate that the target battery meets the first static time length.

[0090] S203 : Calculate the first remaining capacity and the second remaining capacity corresponding to the target battery at the beginning and end of the plurality of first rest time periods.

[0091] After obtaining multiple long-term rest events, it is necessary to calculate the first remaining capacities of the target battery at the beginning of multiple first rest periods and the second remaining capacities of the target battery at the end of multiple first rest periods.

[0092] S204 : Constructing multiple original data sets of the target battery according to the multiple first remaining capacities, the multiple second remaining capacities, and the multiple frames of original data.

[0093] Based on the multiple frames of raw data collected in S201 and the multiple first remaining capacities and second remaining capacities obtained in S203 , multiple raw data sets of the target battery can be constructed.

[0094] Among them, in the multiple frames of raw data, the time when the data was collected is used to obtain the original data set, the usage time and static time of the target battery, the vehicle mileage when the data was collected is used to obtain the usage mileage of the target battery, and the temperature information is used to obtain the static temperature of the target battery. The capacity retention rate of the target battery in the original data set is calculated from the voltage parameters and static temperature in the original data.

[0095] Another optional embodiment of the present disclosure provides a specific implementation method for obtaining multiple long-term static events. FIG3 is a third flow diagram of a method for training a capacity retention rate judgment model provided in an optional embodiment of the present disclosure. As shown in FIG3 , obtaining multiple long-term static events based on multiple frames of raw data corresponding to each vehicle in S202 may include:

[0096] S301 , segmenting multiple frames of raw data to obtain multiple events corresponding to a target battery.

[0097] After collecting multiple frames of raw data, in order to improve computing efficiency, the multiple frames of raw data are segmented. That is, the multiple frames of raw data can be divided into multiple events so that only corresponding segments are selected during subsequent processing, saving computing resources and improving computing efficiency.

[0098] The multiple frames of raw data are segmented to obtain multiple events corresponding to the target battery, specifically including: segmenting the multiple frames of raw data according to multiple preset events based on the connection information of the vehicle main relay and the connection information of the vehicle charging gun corresponding to the multiple frames of raw data, and obtaining multiple events corresponding to the target battery.

[0099] For example, Table 2 includes the main relay status and the charging gun connection status. In this embodiment, in multiple frames of data, the data in which the main relay is energized and the charging gun connection status is disconnected can be marked as driving data (recorded as segment_type=10); the data frame in which the charging gun connection status is fast charging is marked as fast charging data (recorded as segment_type=30); the data frame in which the charging gun connection status is non-fast charging and the connection status is connected is marked as slow charging data (recorded as segment_type=40); and the data frames other than driving, fast charging, and slow charging data can be marked as static data (recorded as segment_type=20).

[0100] Then, the data is arranged in ascending order by time, and the time difference between the previous and next frames of data is greater than or equal to 3 minutes and the segment_type field changes as the basis for event segmentation. If either of the above two conditions is met, an event segmentation is performed, and the corresponding features of the segmented events are extracted to form an event table. The obtained event table is shown in Table 3.

[0101] Table 3 Fields included in each event

[0102] As shown in Table 3, each event includes at least: event start time, event end time, event type, event start mileage, event end mileage, and the voltage and temperature of the target battery when the event occurs.

[0103] S302: Determine, based on the time difference between each two adjacent events and the mileage difference between each two adjacent events, that an event whose time difference and mileage difference meet a preset threshold meets a first static duration.

[0104] After obtaining multiple events corresponding to the target battery, all events are sorted in ascending order by event start time, and the time difference between each two adjacent events is calculated according to formula (1), and the mileage difference between each two adjacent events is calculated according to formula (2). time_dif i =start_time i -end_time i-1 (1) mileage_dif i =start_mileage i -end_mileage i-1 (2)

[0105] Then, based on the time difference between each two adjacent events and the mileage difference between each two adjacent events, it is determined that the events whose time difference and mileage difference both meet the preset thresholds meet the first static period. In an optional embodiment, the preset threshold for the time difference is set to be greater than 10 days, and the preset threshold for the mileage difference is set to 0. That is, it is considered that the events whose time difference is greater than 10 days and the mileage difference is 0 meet the first static period, and the event is a long-term static event.

[0106] In summary, dividing multiple frames of raw data into segments to obtain multiple events is beneficial for selecting only corresponding segments in subsequent data processing, saving computing resources and improving computing efficiency. Moreover, long-term static events can be determined based on the information of multiple events, which improves the speed and accuracy of obtaining long-term static events, speeds up the training of the battery capacity retention rate judgment model, and is conducive to further improving the efficiency of obtaining the battery capacity retention rate.

[0107] To further ensure the accuracy of the trained battery capacity retention rate judgment model, FIG4 is a fourth flow chart of a method for training a capacity retention rate judgment model provided in an optional embodiment of the present disclosure. As shown in FIG4 , before constructing multiple raw data sets of the target battery based on the multiple first remaining capacities, the multiple second remaining capacities, and the multiple frames of raw data in S204, the method further includes:

[0108] S401: Calculate and obtain cumulative time error parameters according to the sampling frequency of multiple frames of original data.

[0109] As described in S204, the capacity retention rate of the target battery in the original data set is calculated based on the voltage parameters and the rest temperature in the original data. In the following embodiments, it will be mentioned that when calculating the capacity retention rate of the target battery based on the voltage parameters and the rest temperature in the original data, the ampere-hour integral is applied to calculate the capacity change value. Inevitably, it is necessary to consider the estimation error caused by missing data. At the same time, due to the inconsistent confidence levels of the capacity estimation of each event, the cumulative time error parameter TE is introduced.

[0110] Assuming that the sampling frequency of each frame of raw data in this disclosure is x (hz), the sampling interval of the raw data can be calculated to be 1 / x (s). If the actual interval of each frame of data is delta_time i , calculate the time error TL of each frame of original data according to formula (3) i : TL i =delta_time i -1 / x (3)

[0111] Normally, delta_time i is equal to 1 / x, so TL i The result should be zero when TLi When it is greater than zero, it means that the time of the i-th frame is lost. According to formula (3), each frame of raw data is screened to obtain TL i A data frame with values ​​greater than zero.

[0112] Then, according to the event type segment_type of the original data, the corresponding event error coefficient μ is set for the time error in each event type type The design principle of the event error coefficient is to formulate it according to the fluctuation and size of the current. For example, the current fluctuation in the driving event is large. When there is the same data loss, the estimation accuracy is small. Then the μ of the driving event is type Set to a larger value. Similarly, due to the small fluctuation of the current in the slow charging event, the current is small. Under the same data loss situation, the estimation accuracy is greater, so the μ of the slow charging event is type Set it to a smaller value, and based on this principle, set the error coefficient μ for each event type type Parameters and parameter values ​​are not described in detail in this disclosure.

[0113] For events that require battery capacity retention rate estimation, the estimation error TL of event i is calculated according to formula (4): ei , where n is the total number of data frames contained in event i.

[0114] Aggregate all events that require battery capacity retention rate estimation, and obtain the cumulative time error parameter TE according to formula (5):

[0115] Where n is the number of events.

[0116] S402: Filter multiple frames of original data according to the accumulated time error parameter to obtain multiple frames of target data.

[0117] After S401 calculates and obtains the cumulative time error parameter, multiple frames of original data are screened according to the cumulative time error parameter to obtain multiple frames of target data. The obtained target data eliminates the estimation error caused by missing data, which can make the trained battery capacity retention rate judgment model more accurate.

[0118] Optionally, multiple frames of raw data are filtered, for example, each frame of data is filtered based on a preset cumulative time error threshold TBD: after S401 calculates and obtains the cumulative time error parameter, if TE>TBD, the state of charge (SOC) of the frame data before standing still is considered unreliable, and the frame data is removed. Then, the TBD threshold is adjusted according to factors such as the actual operation of the model and the amount of data to ensure that the amount of raw data obtained can meet the needs of subsequent calculations.

[0119] The step S204 constructs multiple original data sets of the target battery based on the multiple first remaining capacities, the multiple second remaining capacities, and the multiple frames of original data, including:

[0120] S403 : Constructing multiple original data sets of the target battery according to the multiple first remaining capacities, the multiple second remaining capacities, and the multiple frames of target data.

[0121] According to the cumulative time error parameter, multiple frames of original data are screened to obtain multiple frames of target data. Then, multiple original data sets of the target battery are constructed based on multiple first remaining capacities, multiple second remaining capacities, and multiple frames of target data. This eliminates the influence of estimation errors caused by missing data, ensures that the data frames in the multiple original data sets obtained are more accurate, and further ensures the accuracy of the battery capacity retention rate judgment model obtained by training.

[0122] In order to ensure that the original data is highly consistent with the actual data under the actual operating conditions of the target battery, an optional embodiment of the present disclosure provides a specific implementation method for preprocessing the original data.

[0123] FIG5 is a fifth flow chart of a method for training a capacity retention rate judgment model according to an optional embodiment of the present disclosure. As shown in FIG5 , after collecting multiple frames of raw data of multiple vehicles equipped with target batteries in corresponding preset time series in S201, the method further includes:

[0124] S501 , cleaning multiple frames of original data to obtain cleaned original data.

[0125] Data preprocessing plays an important role in building a neural network model and often determines the training effect. This embodiment provides a specific preprocessing strategy, as shown in Table 4.

[0126] Table 4: Strategies for cleaning raw data

[0127] Among them, TBD is the threshold of the relevant parameters, TBD1-TBD2 indicates that the voltage threshold is 0V-5V, TBD3-TBD4 indicates that the temperature threshold is -40℃-150℃, and TBD5-TBD6 indicates that the current threshold is -900A-1200A.

[0128] According to the cleaning strategy shown in Table 4, the original data shown in Table 2 is cleaned, and then S202 and subsequent operations are performed based on the cleaned data.

[0129] Then, based on S501, the segmentation of the multiple frames of raw data in S301 to obtain multiple events corresponding to the target battery may include:

[0130] S502: Segment the cleaned raw data to obtain multiple events corresponding to the target battery.

[0131] S301 divides the multiple frames of original data into segments to obtain multiple events corresponding to the target battery. Specifically, it can be: divide the cleaned original data into segments to obtain multiple events corresponding to the target battery. This ensures that the data frames contained in the multiple events obtained are consistent with the actual working conditions, making the output results of the trained battery capacity retention rate judgment model more consistent with the actual situation.

[0132] In this embodiment, multiple frames of raw data are cleaned to obtain cleaned raw data, thereby realizing preprocessing of the raw data. Multiple events corresponding to the target battery are obtained based on the cleaned raw data, ensuring that the data contained in the multiple events are highly consistent with the actual data under actual working conditions, making the battery capacity retention rate judgment model obtained through training more accurate, and the battery capacity retention rate obtained using the battery capacity retention rate judgment model more accurate.

[0133] Based on the embodiment shown in Figure 5, the calculation of the first remaining capacity and the second remaining capacity corresponding to the target battery at the start and end of multiple first standby periods in S203 may include: determining the first remaining capacity corresponding to the target battery based on information of the target standby event before the start of the first standby period and multiple events corresponding to the target battery, and taking the battery capacity at the end of the first standby period as the second remaining capacity, wherein the target standby event is an event that satisfies the second standby period, and the second standby period is less than the first standby period.

[0134] Specifically, during battery use, the state of charge (SOC) is the ratio of the actual amount of power that can be provided in the current state to the amount of power that can be provided in a fully charged state. For pure electric vehicles, accurate SOC estimation is the main basis for ensuring that the power battery is charged and discharged within the operating range. The open-circuit voltage (OCV) is the voltage across the battery after a long-term static event. At a certain temperature, the state of charge of the battery and the open-circuit voltage are in a one-to-one correspondence. Therefore, the open-circuit voltage interpolation method (OCV interpolation method) can be used according to the battery voltage and temperature to calculate the remaining capacity of the target battery when the target static event occurs, as well as the second remaining capacity corresponding to the target battery at the end of the first static time.

[0135] Furthermore, in the present disclosure, the first remaining capacity corresponding to the target battery can be determined based on the information of the target standby event before the start of the first standby period and multiple events corresponding to the target battery, including: according to the voltage and temperature of the target battery when the target standby event occurs, the open circuit voltage interpolation method (OCV interpolation method) is used to calculate and obtain the remaining capacity of the target battery when the target standby event occurs, and then, according to the data frames of multiple events between the target standby event and the long-term standby event, the ampere-hour integral is used to calculate and obtain the first remaining capacity corresponding to the target battery when the first standby period starts.

[0136] Table 5 is an example of the OCV interpolation method. As shown in Table 5, the first column is the temperature parameter, the first row is the battery remaining capacity, and the remaining values ​​are voltage parameters. After obtaining the voltage and temperature of the target battery, first find the value corresponding to its temperature parameter in the first column, then find the value corresponding to the voltage parameter in the row corresponding to the temperature parameter, and finally, determine the battery remaining capacity corresponding to the voltage parameter in the first row, and use this value as the remaining capacity of the target battery (that is, the state of charge SOC).

[0137] Table 5 OCV interpolation method example

[0138] Figure 6 is a schematic diagram of multiple events in a time series provided by the present disclosure, and Figure 7 is a schematic diagram of multiple events in a time series provided by the present disclosure. As shown in Figures 6 and 7, the last frame before the long-term static event does not meet the static condition and cannot be calculated using the OCV interpolation method. To ensure accurate calculation, for each long-term static event event i The last frame before the long-term static event event i As the starting point, trace back n events until the most recent target static event that meets the second static duration. i-n , event i-nThe following conditions must be met: time_dif i-n >2(h) (6) mileage_dif i-n =0(km) (7)

[0139] That is, if the second static time of the target static event is greater than 2 hours and the mileage difference is 0, and the above conditions are met at the same time, the event is considered i-n Satisfy the static state and no data loss, event event i-n The first frame is credible, so the event event i-n As the target static event, event event i-n The first frame of is used as the estimation starting point of the first remaining capacity.

[0140] In addition, in order to ensure the accuracy of the first remaining capacity estimation, the event i-n To event i The mileage difference of n events is required, as shown in formula (8):

[0141] Among them, sum_mileage_dif i For event i-n To event i The total mileage difference within is only sum_mileage_dif i =0, it is considered that event i-n To event i The data are continuous and there is no data loss, which satisfies the condition of estimating the first remaining capacity of the target battery before the long-term standstill event.

[0142] Then, due to the calculated event event i-n , meeting the static condition, so the first frame can be trusted. According to the voltage list and temperature of the first frame, the event can be calculated according to the OCV interpolation method. i-n Initial cloud SOC could_start , that is, the remaining capacity of the target battery when the target static event occurs.

[0143] The next step is to i-n To event i-1 Perform cloud SOC estimation:

[0144] Event i-n Taking events as an example, we intercept the original data based on the event start time and event end time, and then sort the data by time.x =timestamp x -timestamp x-1 (9)

[0145] Among them, delta_time x For event i-n The time difference of the x-th frame data in the event.

[0146] Furthermore, we have: delta_Q x =(packcurrent x ×delta_time x ) / 3600 (10)

[0147] Among them, delta_Q x is the capacity change value of the x-th frame data. 3600 means there are 3600 seconds per hour.

[0148] Further, there are:

[0149] Among them, sum_delta_Q is the event event i-n The content volume change of the entire segment data, where n is the total number of data frames in the segment.

[0150] Furthermore, there is: Q = Q0 × (SOH / 100) (12) where Q is the current full charge capacity of the battery, Q0 is the rated capacity, and SOH is the mode of the SOH in the current segment. could_end =SOC could_start -sum_delta_Q / Q (13)

[0151] Among them, SOC cpuld_end For event i-n The cloud ends SOC, so we have completed the event i-n Cloud-based SOC estimation of the event.

[0152] Similarly, event i-n SOC could_end , for event event i-n+1 SOC could_start , repeat the entire process of formula (9)-(13) above to get the event event i-n+1 SOC could_end , and so on, you can get the previous event event of the long-term static event i-1 SOC could_end , as the SOC value at the beginning of the long-term static event of the present disclosure.

[0153] FIG8 is a sixth flow chart of a method for training a capacity retention rate judgment model according to an optional embodiment of the present disclosure. As shown in FIG8 , a training sample set is substituted into a preset neural network model to train and obtain a battery capacity retention rate judgment model, including:

[0154] S601: Normalize sample features of the training sample set.

[0155] Since the value ranges of different features of the sample are inconsistent, it will affect the convergence speed of the subsequent data algorithm. Therefore, the features are normalized. For any feature X J (J=1,2,3,4), find its mathematical expectation as Calculate its standard deviation std(X J ), further, record the transformed features for The conversion process is calculated as follows:

[0156] S602: Train the preset neural network model according to the normalized training sample set, activation function, and optimization function to obtain a battery capacity retention rate judgment model.

[0157] The activation function is the Sigmoid function, which has a value range of (0,1) and has the advantages of monotonicity and being differentiable everywhere. It can increase the prediction accuracy and convergence speed. The formula of the activation function S(x) is as follows:

[0158] The optimization function selects the steepest descent method, and the idea of ​​iterating in the direction of the maximum gradient is as follows:

[0159] In formula (16), ω represents the weight of any neuron in the preset neural network, θ represents the threshold of any neuron, η is the learning rate, that is, the step size of a single iteration, Δω k is the neuron weight adjustment after the kth training, Δθ k is the amount of neuron threshold adjustment after the kth training. By taking the partial derivative of the weights of each neuron and following the set learning rate, we can get the change in the current weight and threshold, and then complete the parameter modification, and finally perform training again.

[0160] In the present disclosure, when training to obtain a battery capacity retention rate judgment model, the sample features of the training sample set are first normalized. Then, based on the normalized training sample set, the activation function shown in formula (15), and the optimization function shown in formula (16), the preset maximum training times of 3000, the set loss function threshold of 0.03, and the default values ​​of the initial thresholds and weights of the neurons in each layer are used to train the preset neural network model. When the preset neural network model reaches the maximum training times or the loss function value is lower than the set threshold, it means that the model has completed the training and the battery capacity retention rate judgment model is obtained.

[0161] Optionally, as described in S102, the embodiment of the present disclosure uses 75% of the original data set as a training sample set and 25% as a test set. Therefore, after training to obtain the battery capacity retention rate judgment model, the test set can also be input into the trained model. When the test set error is less than the set threshold, it is considered that the model training effect is good. Otherwise, the parameters need to be adjusted and the model needs to be retrained. This cycle is repeated until the model accuracy meets the standard.

[0162] In summary, the present disclosure provides a method for training a capacity retention rate judgment model, which has the following advantages:

[0163] 1. The trained battery capacity retention rate judgment model can automatically obtain the battery capacity retention rate based on the battery data of the battery to be tested. Compared with laboratory test confirmation, it reduces manpower and material costs. In addition, the capacity retention rate is calculated based on the OCV interpolation method, which improves the accuracy of the calculation. Moreover, using this method, the battery capacity retention rate data is automatically calculated and obtained by the electronic device, which improves the universality of the method's usage scenarios and allows users to obtain the battery capacity retention rate performance indicators more conveniently and accurately.

[0164] 2. The cumulative time error parameter is introduced. By adjusting the threshold of the cumulative time error parameter, the amount of data input into the preset neural network model can be adjusted. It can also help avoid abnormal scenarios where a large amount of data is lost, thereby improving calculation accuracy.

[0165] 3. The trained battery capacity retention rate judgment model can fit the battery performance data with the capacity retention rate under standard operating conditions, thereby calculating the actual capacity retention rate performance index of the battery. This can not only provide a more accurate battery performance evaluation, but also provide battery manufacturers and users with key information about the battery health status.

[0166] After training and acquiring the battery capacity retention rate judgment model according to the above embodiment, the standard battery operating time, battery mileage, long-term battery rest time, and battery rest temperature in the design specification of the battery to be tested can be combined into a feature sequence X_normal. This is input into the trained battery capacity retention rate judgment model to obtain the predicted battery capacity retention rate value Y_normal for the battery to be tested. By comparing the predicted capacity retention rate value under standard operating conditions with the standard capacity retention rate performance index, it can be determined whether the performance index of the battery to be tested meets the performance index.

[0167] FIG9 is a flow chart of a method for applying a capacity retention rate judgment model according to an optional embodiment of the present disclosure. As shown in FIG9 , an optional embodiment of the present disclosure provides a method for applying a capacity retention rate judgment model, including:

[0168] S100: Obtain battery data of a battery to be tested.

[0169] When it is necessary to obtain the capacity retention rate of the battery to be tested, it is only necessary to obtain the battery data of the battery to be tested and substitute the battery data of the battery to be tested into the battery capacity retention rate judgment model. The battery data of the battery to be tested includes: usage time, usage mileage, long-term static time, and static temperature.

[0170] S200 , substituting battery data of the battery to be tested into a battery capacity retention rate judgment model to obtain the battery capacity retention rate of the battery to be tested.

[0171] In the above embodiment, the battery capacity retention rate judgment model is trained by the usage time of the target battery, the usage mileage of the target battery, the long-term static time of the target battery, the static temperature of the target battery, and the capacity retention rate of the target battery. Therefore, by inputting the usage time, usage mileage, long-term static time, and static temperature into the battery capacity retention rate judgment model, the battery capacity retention rate can be obtained.

[0172] By adopting the application method provided in this embodiment, after training and obtaining the battery capacity retention rate judgment model, if one wants to judge the capacity retention rate of the battery to be tested, one only needs to substitute the battery data of the battery to be tested into the battery capacity retention rate judgment model, and the battery capacity retention rate of the battery to be tested can be automatically judged, thereby improving the judgment accuracy and efficiency while also saving a lot of manpower and material resources.

[0173] The following continues to explain the apparatus, device and storage medium for executing the training method of the capacity retention rate judgment model provided by any of the above optional embodiments of the present disclosure. The specific implementation process and the technical effects produced are the same as those of the corresponding method embodiments mentioned above. For the sake of brief description, for the parts not mentioned in the following embodiments, reference can be made to the corresponding content in the method embodiments.

[0174] An optional embodiment of the present disclosure provides a training device for a capacity retention rate judgment model. FIG10 is a schematic structural diagram of the training device for a capacity retention rate judgment model provided by an optional embodiment of the present disclosure. As shown in FIG10 , the training device for a capacity retention rate judgment model includes:

[0175] The acquisition module 10 is used to obtain multiple original data sets of the target battery; wherein each original data set includes the following parameters of the target battery when the first rest time condition is met: parameter information of the target battery's usage time, usage mileage, rest time, and rest temperature, and the capacity retention rate of the target battery.

[0176] The construction module 20 is used to construct a training sample set based on multiple original data sets.

[0177] The training module 30 is used to substitute the training sample set into the preset neural network model to obtain the battery capacity retention rate judgment model through training. The battery capacity retention rate judgment model is used to determine the battery capacity retention rate according to the battery data.

[0178] Optionally, the acquisition module 10 is used to collect multiple frames of raw data from multiple vehicles equipped with target batteries in corresponding preset time sequences; wherein each frame of raw data includes: the vehicle identification corresponding to the target battery, the time when the data is collected, the voltage of the target battery, the vehicle mileage when the data is collected, and the temperature information; based on the multiple frames of raw data corresponding to each vehicle, multiple long-term static events are obtained; the long-term static events are used to indicate that the target battery meets a first static time; the first remaining capacity and the second remaining capacity corresponding to the target battery at the beginning and end of the multiple first static time periods are calculated; based on the multiple first remaining capacities, the multiple second remaining capacities, and the multiple frames of raw data, multiple raw data sets of the target battery are constructed.

[0179] Optionally, the acquisition module 10 is used to segment multiple frames of raw data to obtain multiple events corresponding to the target battery; wherein each event includes at least: event start time, event end time, event type, event start mileage, event end mileage, and the voltage and temperature of the target battery when the event occurs; based on the time difference between each two adjacent events and the mileage difference between each two adjacent events, it is determined that the events whose time difference and mileage difference meet the preset thresholds meet the first static time length.

[0180] Optionally, the acquisition module 10 is used to segment the multiple frames of original data according to multiple preset events based on the connection information of the vehicle main relay and the connection information of the vehicle charging gun corresponding to the multiple frames of original data, and obtain multiple events corresponding to the target battery.

[0181] Optionally, the acquisition module 10 is used to calculate and obtain a cumulative time error parameter based on the sampling frequency of multiple frames of original data; filter the multiple frames of original data based on the cumulative time error parameter to obtain multiple frames of target data; and construct multiple original data sets of the target battery based on multiple first remaining capacities, multiple second remaining capacities, and multiple frames of target data.

[0182] Optionally, the acquisition module 10 is configured to clean multiple frames of raw data to obtain cleaned raw data; and segment the cleaned raw data to obtain multiple events corresponding to the target battery.

[0183] Optionally, the acquisition module 10 is used to determine the first remaining capacity corresponding to the target battery based on information of the target standby event before the start of the first standby period and multiple events corresponding to the target battery, and use the battery capacity at the end of the first standby period as the second remaining capacity, wherein the target standby event is an event that satisfies the second standby period, and the second standby period is less than the first standby period.

[0184] Optionally, the acquisition module 10 is configured to calculate and acquire a second remaining capacity corresponding to the target battery by adopting an open circuit voltage interpolation method according to the voltage and temperature of the target battery at the end of the first rest period.

[0185] Optionally, the training module 30 is configured to normalize sample features of the training sample set;

[0186] According to the normalized training sample set, activation function, and optimization function, the preset neural network model is trained to obtain a battery capacity retention rate judgment model.

[0187] The above-mentioned device is used to execute the method provided in the above-mentioned embodiment. Its implementation principle and technical effect are similar and will not be repeated here.

[0188] The above modules can be one or more integrated circuits configured to implement the above methods, such as one or more application-specific integrated circuits (ASICs), one or more microprocessors, or one or more field programmable gate arrays (FPGAs). For another example, when a module is implemented by scheduling program code through a processing element, the processing element can be a general-purpose processor, such as a central processing unit (CPU) or other processor that can call program code. For another example, these modules can be integrated together and implemented in the form of a system-on-a-chip (SOC).

[0189] Optionally, the present disclosure further provides an application device for a capacity retention rate judgment model, comprising:

[0190] The data acquisition module is used to acquire battery data of the battery to be tested, wherein the battery data of the battery to be tested includes: usage time, usage mileage, long-term static time, and static temperature.

[0191] The substitution module is used to substitute the battery data of the battery to be tested into the battery capacity retention rate judgment model of any of the above embodiments to obtain the battery capacity retention rate of the battery to be tested.

[0192] An optional embodiment of the present disclosure further provides an electronic device. Figure 11 is a structural schematic diagram of the electronic device provided by an optional embodiment of the present disclosure. As shown in Figure 11, the electronic device provided by the present disclosure includes: a processor 100, a storage medium 200 and a bus 300. The storage medium stores program instructions executable by the processor. When the electronic device is running, the processor and the storage medium communicate through the bus, and the processor executes the program instructions to implement the training method of the capacity retention rate judgment model of any of the above embodiments.

[0193] An optional embodiment of the present disclosure further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the training method of the capacity retention rate judgment model of any of the above embodiments is implemented.

[0194] In the several embodiments provided in the present disclosure, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0195] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0196] In addition, the functional units in the various embodiments of the present disclosure may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0197] The above-mentioned integrated unit implemented in the form of a software functional unit can be stored in a computer-readable storage medium. The above-mentioned software functional unit is stored in a storage medium, including a number of instructions for enabling an electronic device (which can be a personal computer, a server, or a network device, etc.) or a processor (English: processor) to perform some steps of the method described in each embodiment of the present disclosure. The aforementioned storage medium includes: a USB flash drive, a mobile hard disk, a read-only memory (English: Read-Only Memory, abbreviated: ROM), a random access memory (English: Random Access Memory, abbreviated: RAM), a disk or an optical disk, and other media that can store program code.

[0198] The above are only specific embodiments of the present disclosure, but the scope of protection of the present disclosure is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this disclosure should be included in the scope of protection of the present disclosure. Therefore, the scope of protection of the present disclosure should be based on the scope of protection of the claims.

Claims

1. A training method for a capacity retention rate judgment model, comprising: Obtaining a plurality of original data sets of a target battery; wherein each of the original data sets includes the following parameters of the target battery under the condition of satisfying the first static time duration condition: the usage time of the target battery, the driving mileage, the static time, and the parameter information of the static temperature, and the capacity retention rate of the target battery; Constructing a training sample set according to the plurality of original data sets; Substituting the training sample set into a preset neural network model to train and obtain a battery capacity retention rate judgment model, which is used to determine the battery capacity retention rate according to battery data.

2. The method according to claim 1, wherein, The obtaining of the plurality of original data sets of the target battery includes: Collecting multiple frames of original data of a plurality of vehicles equipped with the target battery respectively under corresponding preset time series; wherein each frame of original data includes: the vehicle identifier corresponding to the target battery, the time when the data is collected, the voltage of the target battery, the vehicle mileage when the data is collected, and the temperature information; Obtaining a plurality of long-term static events according to the multiple frames of original data corresponding to each vehicle; the long-term static events are used to indicate that the target battery satisfies the first static time duration; Calculating the first remaining capacity and the second remaining capacity respectively corresponding to the target battery at the start and end of the plurality of first static time durations; Constructing a plurality of original data sets of the target battery according to the plurality of first remaining capacities, the plurality of second remaining capacities, and the multiple frames of original data.

3. The method according to claim 2, wherein The obtaining of the plurality of long-term static events according to the multiple frames of original data corresponding to each vehicle includes: Performing segment division on the multiple frames of original data to obtain a plurality of events corresponding to the target battery; wherein each event at least includes: the start time of the event, the end time of the event, the event type, the start mileage of the event, the end mileage of the event, and the voltage and temperature of the target battery when the event occurs; Determining that the events whose time difference and mileage difference satisfy a preset threshold satisfy the first static time duration according to the time difference between every two adjacent events and the mileage difference between every two adjacent events.

4. The method according to claim 3, wherein, The performing of segment division on the multiple frames of original data to obtain a plurality of events corresponding to the target battery includes: Performing segment division on the multiple frames of original data according to the connection information of the vehicle main relay and the connection information of the vehicle charging gun corresponding to the multiple frames of original data to obtain a plurality of events corresponding to the target battery.

5. The method according to claim 2, wherein, Before constructing the plurality of original data sets of the target battery according to the plurality of first remaining capacities, the plurality of second remaining capacities, and the multiple frames of original data, it further includes: Calculating and obtaining an accumulated time error parameter according to the sampling frequency of the multiple frames of original data; Filtering the multiple frames of original data according to the accumulated time error parameter to obtain multiple frames of target data; The constructing of the plurality of original data sets of the target battery according to the plurality of first remaining capacities, the plurality of second remaining capacities, and the multiple frames of original data includes: Construct multiple original data sets of the target battery according to the multiple first remaining capacities, the multiple second remaining capacities, and multiple frames of the target data.

6. The method according to claim 3, wherein, After collecting multiple frames of original data of multiple vehicles equipped with the target battery respectively at corresponding preset time series, it further includes: cleaning the multiple frames of original data to obtain the cleaned original data; The segmenting the multiple frames of original data to obtain multiple events corresponding to the target battery includes: Segment the cleaned original data to obtain multiple events corresponding to the target battery.

7. The method according to claim 6, wherein, The calculating the first remaining capacity and the second remaining capacity corresponding to the target battery respectively at the start and end of the multiple first standing durations includes: Determine the first remaining capacity corresponding to the target battery according to the information of the target standing event before the start of the first standing duration and the multiple events corresponding to the target battery, and use the battery capacity at the end of the first standing duration as the second remaining capacity, where the target standing event is an event that satisfies the second standing duration, and the second standing duration is less than the first standing duration.

8. The method according to claim 7, wherein The using the battery capacity at the end of the first standing duration as the second remaining capacity includes: According to the voltage and temperature of the target battery at the end of the first standing duration, use the open-circuit voltage interpolation method to calculate and obtain the second remaining capacity corresponding to the target battery.

9. The method according to claim 1, wherein The substituting the training sample set into a preset neural network model to train and obtain a battery capacity retention rate judgment model includes: Normalize the sample features of the training sample set; Train the preset neural network model according to the normalized training sample set, activation function, and optimization function to obtain the battery capacity retention rate judgment model.

10. A method for applying a capacity retention rate judgment model, including: Obtain the battery data of the battery to be tested, where the battery data of the battery to be tested includes: usage time, mileage, long-term standing time, and standing temperature; Substitute the battery data of the battery to be tested into the battery capacity retention rate judgment model according to any one of claims 1-9 to obtain the battery capacity retention rate of the battery to be tested.

11. A training device for a capacity retention rate judgment model, including: An acquisition module, configured to acquire multiple original data sets of the target battery; where each original data set includes the following parameters of the target battery under the condition of satisfying the first standing duration: parameter information of the usage time, mileage, standing time, and standing temperature of the target battery, and the capacity retention rate of the target battery; A construction module, configured to construct a training sample set according to the multiple original data sets; A training module, configured to substitute the training sample set into a preset neural network model to train and obtain a battery capacity retention rate judgment model, and the battery capacity retention rate judgment model is used to determine the battery capacity retention rate according to the battery data.

12. An electronic device, comprising: A processor, a storage medium, and a bus, wherein the storage medium stores program instructions executable by the processor. When the electronic device is running, the processor communicates with the storage medium through the bus, and the processor executes the program instructions to implement the training method of the capacity retention rate judgment model according to any one of claims 1 to 9.

13. A computer-readable storage medium, on which a computer program is stored. When the computer program is run by a processor, the training method of the capacity retention rate judgment model according to any one of claims 1 to 9 is implemented.

Images