Power battery health status prediction method and system based on user behavior

Abstract

The present invention relates to a method and system for predicting the health status of a power battery based on user behavior. The method comprises: sampling the battery parameters of a power battery to be tested and the driving parameters of a corresponding electric vehicle, obtaining the battery parameters and driving parameters within a preset sampling period, and dividing the sampling period into a charging period and a discharging period according to the battery parameters; obtaining the charging parameters of the power battery to be tested and the maximum available capacity in each charging period according to the battery parameters within each charging period; obtaining the health status of the power battery to be tested according to the maximum available capacity of the power battery to be tested in each charging period and the rated capacity of the power battery to be tested; obtaining the performance parameters of the electric vehicle according to the driving parameters within each discharging period; and inputting the performance parameters, charging parameters and health status into a battery health status prediction model to predict the expected health status of the power battery to be tested. The present invention improves the accuracy of power battery health status prediction.

Description

Technical Field

[0001] The present invention relates to the field of battery technology, and in particular to a method and system for predicting the health status of a power battery based on user behavior. Background Art

[0002] The state of health (SOH) of power battery systems (such as lithium-ion batteries and sodium-ion batteries) used in electrified vehicles such as electric vehicles, rail transit, electric ships, and electric aircraft will directly affect the performance and economy of the vehicles. Therefore, how to accurately estimate and predict the health state of power batteries has become one of the key technologies for promoting the electrification of transportation. At present, the research and development of power battery health state estimation and prediction mainly focuses on cells, modules, and systems in laboratory environments, while research and development for power battery systems under actual working conditions is relatively lacking. At the same time, such vehicles usually require human users to operate, and the behaviors of different users may vary greatly. If such differences are not taken into account, the health state estimation results may deviate greatly from the actual situation. Therefore, it is necessary to provide a power battery health state prediction method and system based on user behavior. Summary of the invention

[0003] The present invention provides a method for predicting the health status of a power battery based on user behavior, so as to solve the problem that the prior art cannot accurately predict the health status of a power battery.

[0004] The present invention provides a method for predicting the health status of a power battery based on user behavior, which is applied to an electric vehicle. The method includes: sampling battery parameters of a power battery to be tested and driving parameters of the corresponding electric vehicle, obtaining battery parameters and driving parameters within a preset sampling period, and dividing the sampling period into a charging period and a discharging period according to the battery parameters; obtaining the charging parameters of the power battery to be tested and the maximum available capacity in each charging period according to the battery parameters in each charging period; obtaining the health status of the power battery to be tested according to the maximum available capacity of the power battery to be tested in each charging period and the rated capacity of the power battery to be tested; obtaining performance parameters of the electric vehicle according to the driving parameters in each discharging period; and inputting the performance parameters, charging parameters and the health status into a battery health status prediction model to predict the expected health status of the power battery to be tested.

[0005] In one embodiment of the present invention, the battery parameter is the state of charge, and the process of dividing the sampling period into charging periods according to the battery parameter includes: arranging all battery parameters in the sampling period in time sequence to form a battery parameter sequence; determining the charging start time and the charging end time according to the change of each battery parameter in the battery parameter sequence; and taking the time interval between adjacent charging start times and charging end times as a charging period.

[0006] In one embodiment of the present invention, arranging all battery parameters in the sampling period in time sequence to form a battery parameter sequence includes: removing missing and abnormal battery parameters from all battery parameters; and arranging the remaining battery parameters in time sequence to form a battery parameter sequence.

[0007] In one embodiment of the present invention, the charging start time and the charging end time are determined according to the change of the battery parameters in the battery parameter sequence, including: for any battery parameter in the battery parameter sequence: if the current battery parameter is greater than the previous battery parameter and less than or equal to the next battery parameter, then the sampling time corresponding to the current battery parameter is the charging start time; if the current battery parameter is greater than the previous battery parameter and greater than or equal to the next battery parameter, then the sampling time corresponding to the current battery parameter is the charging end time.

[0008] In one embodiment of the present invention, the maximum available capacity of the power battery to be tested in each charging period is obtained, including: for any charging period: according to the difference in the charge state of the power battery to be tested at the beginning and end of the charging period and the current in the charging period, the maximum available capacity of the power battery to be tested in the charging period is obtained.

[0009] In one embodiment of the present invention, taking the time interval between adjacent charging start moments and charging end moments as a charging period includes: taking the time interval between adjacent charging start moments and charging end moments as an alternative charging period; and selecting, from all alternative charging periods, an alternative charging period in which the state of charge of the power battery to be tested is greater than or equal to a preset state of charge threshold as the charging period.

[0010] In one embodiment of the present invention, the battery parameters include state of charge and current, and obtaining the maximum available capacity of the power battery to be tested in each charging period includes: for each charging period: screening out battery parameters of the power battery to be tested whose state of charge is within a preset range from the charging period, and arranging each screened battery parameter according to a time sequence to form a first sequence; interpolating the first sequence, and screening out battery parameters whose current change between adjacent data points in the interpolated first sequence is less than a preset current threshold, to form a second sequence arranged according to a time sequence; and obtaining the maximum available capacity of the power battery to be tested in the charging period according to the difference in the state of charge at the start and end of the second sequence and the current in the second sequence.

[0011] In one embodiment of the present invention, the health status of the power battery to be tested is Where N is the total amount of charging time, Q i is the i-th maximum available capacity, Q rated is the rated capacity of the power battery to be tested.

[0012] In one embodiment of the present invention, the battery health state prediction model is obtained by training, and the training method of the battery health state prediction model includes: sampling the battery parameters of the power battery to be tested and the driving parameters of the corresponding electrified vehicle, obtaining the battery parameters and driving parameters within a plurality of preset sampling periods, and dividing the sampling period into a charging period and a discharging period according to the battery parameters, and obtaining the true health state of the power battery to be tested in each sampling period; for each sampling period: obtaining the charging parameters of the power battery to be tested and the maximum available capacity in each charging period according to the battery parameters in each charging period; obtaining the health state of the power battery to be tested in the current sampling period according to the maximum available capacity of the power battery to be tested in each charging period and the rated capacity of the power battery to be tested; obtaining the performance parameters of the electrified vehicle according to the driving parameters in each discharging period; inputting the performance parameters, charging parameters and health state of the power battery to be tested in the current sampling period into the battery health state prediction model to predict the expected health state of the power battery to be tested in the next sampling period; updating the battery health state prediction model based on the difference between the expected health state of the power battery to be tested in the next sampling period and the corresponding true health state, and obtaining a trained battery health state prediction model.

[0013] In one embodiment of the present invention, a power battery health status prediction system based on user behavior is also provided, and the system includes: a data acquisition module, which is used to sample the battery parameters of the power battery to be tested and the driving parameters of the corresponding electric vehicle, obtain the battery parameters and driving parameters within a preset sampling period, and divide the sampling period into a charging period and a discharging period according to the battery parameters; a maximum available capacity acquisition module, which is used to obtain the charging parameters of the power battery to be tested and the maximum available capacity in each charging period according to the battery parameters in each charging period; a health status acquisition module, which is used to obtain the health status of the power battery to be tested according to the maximum available capacity of the power battery to be tested in each charging period and the rated capacity of the power battery to be tested; a parameter acquisition module, which is used to obtain the performance parameters of the electric vehicle according to the driving parameters in each discharge period; a prediction module, which is used to input the performance parameters, charging parameters and the health status into a battery health status prediction model to predict the expected health status of the power battery to be tested.

[0014] The present invention proposes a method and system for predicting the health status of a power battery based on user behavior. By collecting and analyzing the battery parameters and driving parameters of the electrified vehicle, the performance of the power battery during the charging and discharging periods is accurately monitored, and the dynamic evaluation of the health status of the power battery in different sampling periods is realized. The battery health status prediction model is used to comprehensively analyze the health status, charging parameters and performance parameters of the power battery to accurately predict the health status of the power battery in the future. The present invention integrates different data sources by analyzing parameters related to user behavior such as charging parameters and performance parameters, and combining these parameters with the health status of the power battery. It can more accurately reflect the impact of user behavior on the health status of the power battery, and significantly improve the accuracy of the prediction of the health status of the power battery. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A schematic flow chart of a method for predicting the health status of a power battery based on user behavior provided by an embodiment of the present invention;

[0016] Figure 2 Shown is a structural block diagram of a power battery health status prediction system based on user behavior provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0017] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0018] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and thus the drawings only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0019] In the following description, numerous details are discussed to provide a more thorough explanation of the embodiments of the present invention. However, it is obvious to those skilled in the art that the embodiments of the present invention can be implemented without these specific details. In other embodiments, well-known structures and devices are shown in the form of block diagrams rather than in detail to avoid making the embodiments of the present invention difficult to understand.

[0020] The present invention provides a method for predicting the health status of a power battery based on user behavior. By collecting and analyzing the battery parameters and driving parameters of an electrified vehicle, the performance of the power battery during the charging and discharging periods is accurately monitored, and the dynamic evaluation of the health status of the power battery during different sampling periods is realized. The battery health status prediction model is used to comprehensively analyze the time series data such as the health status, charging parameters and performance parameters of the power battery to accurately predict the health status of the power battery in the future. The present invention can more accurately reflect the impact of user behavior on the health status of the power battery by analyzing parameters related to user behavior such as charging parameters and performance parameters, and combining these parameters with the health status of the power battery, and significantly improve the accuracy of the prediction of the health status of the power battery. The maintenance cost of the power battery is reduced, and higher operating safety and economic benefits are provided for electrified vehicles. Among them, the electrified vehicles include but are not limited to electric vehicles, rail transportation vehicles, electric ships, electric aircraft, electric bicycles, electric motorcycles, electric agricultural machinery or engineering machinery, etc. The power battery includes but is not limited to lithium-ion batteries, sodium-ion batteries, quasi-solid-state batteries, all-solid-state batteries, etc.

[0021] See also Figure 1 , the power battery health status prediction method based on user behavior includes the following steps:

[0022] S1. Sampling the battery parameters of the power battery to be tested and the driving parameters of the corresponding electric vehicle, obtaining the battery parameters and driving parameters within a preset sampling period, and dividing the sampling period into a charging period and a discharging period according to the battery parameters.

[0023] The cloud server samples the battery parameters of the power battery to be tested and the driving parameters of the electrified vehicle according to the preset sampling interval, wherein the sampling interval can be a fixed interval or a non-fixed interval. For example, the sampling interval is fixed to 10 seconds, that is, data is collected every 10 seconds. The sampling period is a continuous time period composed of multiple sampling intervals. Those skilled in the art can set the sampling period based on the adaptability of the power battery monitoring needs. For example, the sampling period is one month, that is, during the entire period from the beginning of the month to the end of the month, the cloud server continuously samples according to the sampling interval (such as 10 seconds), thereby ensuring the integrity and analyzability of the data. According to the battery parameters collected during the sampling period, the charging period and the discharging period are identified. Among them, the charging period is a time period in which the power battery to be tested is receiving energy. During this period, the power battery to be tested is charged by an external power supply, and the state of charge (SOC) of the power battery to be tested gradually increases. The discharging period is a time period in which the power battery to be tested releases energy for external purposes (such as driving an electrified vehicle or power supply equipment). During this period, the SOC of the power battery to be tested gradually decreases. It should be noted that the collected battery parameters and driving parameters are all timestamped with the corresponding sampling moments, so as to facilitate the subsequent determination of the charging period and the discharging period.

[0024] In one embodiment of the present invention, the battery parameter is a state of charge, and the process of dividing the sampling period into charging periods according to the battery parameter includes:

[0025] Arranging all battery parameters within the sampling period in time sequence to form a battery parameter sequence;

[0026] Determining the charging start time and the charging end time according to the change of each battery parameter in the battery parameter sequence;

[0027] The time interval between adjacent charging start times and charging end times is regarded as a charging period.

[0028] During the sampling period, the SOC data of the power battery to be tested is continuously acquired according to the sampling interval, and the collected SOC data is arranged according to the sampling time to form a continuous battery parameter sequence, which reflects the change of the SOC of the power battery to be tested over time during the sampling period. According to the change of SOC, all the charging start time and charging end time and all the discharging start time and discharging end time in the sampling period can be determined. The time interval between each pair of adjacent charging start time and charging end time is defined as a charging period, which reflects a complete time period from the start of receiving electric energy to the stop of receiving electric energy for the power battery to be tested. Correspondingly, the time interval between each pair of adjacent discharging start time and discharging end time is defined as a discharging period, which reflects a complete time period from the start of releasing electric energy to the stop of releasing electric energy for the power battery to be tested.

[0029] In one embodiment of the present invention, all battery parameters within the sampling period are arranged in time sequence to form a battery parameter sequence, including:

[0030] Remove missing and abnormal battery parameters from all battery parameters;

[0031] The remaining battery parameters are arranged in time sequence to form a battery parameter sequence.

[0032] In the present invention, battery parameters that are not within the preset battery parameter range are defined as abnormal battery parameters. In order to speed up data processing, it is necessary to identify and delete all missing battery parameters and abnormal battery parameters from all acquired battery parameters, thereby greatly improving the accuracy and speed of subsequent data processing. After these data are removed, all remaining battery parameters are arranged according to the corresponding sampling time to form a battery parameter sequence, thereby ensuring the continuity and integrity of the data sequence.

[0033] In one embodiment of the present invention, determining the charging start time and the charging end time according to the change of the battery parameters in the battery parameter sequence includes:

[0034] For any battery parameter in the battery parameter sequence:

[0035] If the current battery parameter is greater than the previous battery parameter and less than or equal to the next battery parameter, the sampling time corresponding to the current battery parameter is the charging start time;

[0036] If the current battery parameter is greater than the previous battery parameter and greater than or equal to the next battery parameter, the sampling time corresponding to the current battery parameter is the charging end time.

[0037] When analyzing the battery parameter sequence, the sampling moment corresponding to each battery parameter is regarded as the current sampling moment. Considering that when the battery parameter is SOC, the SOC usually continues to rise during the charging process. Therefore, if the current SOC value is greater than the SOC at the previous sampling moment and less than the SOC at the next sampling moment, it indicates that the power battery starts to charge, that is, the SOC starts to rise, so that the current sampling moment is determined as the charging start moment. Further, if the SOC at the next sampling moment is equal to the SOC at the current sampling moment, since the SOC growth rate may temporarily slow down, the current sampling moment can still be determined as the charging start moment. Conversely, if the current SOC is greater than the SOC at the previous sampling moment and is equal to or greater than the SOC at the next sampling moment, this indicates that the power battery is in a charging state from the previous sampling moment to the current sampling moment, and is in a discharging state or charging has stopped state from the current sampling moment to the next sampling moment. Therefore, the current sampling moment can be determined as the charging end moment.

[0038] In addition, the present invention also determines the discharge start time by identifying the sampling time when the SOC starts to decrease, and determines the discharge end time by identifying the sampling time when the SOC stops decreasing. The time interval between the discharge start time and the adjacent discharge end time is defined as a discharge period. Through this division method, all discharge periods and charging periods can be identified within a given sampling period.

[0039] S2. Obtain charging parameters of the power battery to be tested and the maximum available capacity in each charging period according to the battery parameters in each charging period.

[0040] Specifically, in one embodiment of the present invention, obtaining the maximum available capacity of the power battery to be tested in each charging period includes:

[0041] For any charging period:

[0042] The maximum available capacity of the power battery to be tested in the charging period is obtained according to the difference in SOC at the beginning and the end of the charging period and the current in the charging period.

[0043] After all charging and discharging periods are determined according to the changes in battery parameters, for the i-th charging period, the maximum available capacity Q of the i-th charging period is calculated according to formula (1) based on each current in the charging period and the corresponding sampling interval: i :

[0044]

[0045] Among them, I j(t) is the jth charging current in the charging period (in amperes, A), Δt is the sampling interval (in hours), ΔSOC i is the SOC difference of the power battery to be tested at the beginning and end of the i-th charging period.

[0046] For each charging period: the battery parameters corresponding to each sampling moment include but are not limited to the SOC and current of the power battery to be tested. The charging parameters of the power battery to be tested during the sampling period can be obtained according to the battery parameters, wherein the charging parameters include but are not limited to the number of charging times (times), average charging current (A), average maximum charging current (A), average charging time (seconds), average start charging SOC value (percentage), average end charging SOC value (percentage), average charging depth (percentage) and monthly fast charging rate (percentage), a total of 8 charging parameters.

[0047] Specifically, the number of charging times is the number of charging periods in the sampling period. The average charging current is the average value of the current calculated corresponding to each sampling moment in all charging periods. The average maximum charging current is the average value of the maximum current value in each charging period. The average charging time is the average value of the duration of each charging period. The average start charging SOC value and the average end charging SOC value are the average values ​​of the SOC at the beginning and end of all charging periods, respectively. The average charging depth is the average value of the SOC change from the beginning to the end of each charging period. The monthly fast charging rate is the ratio of the charging period whose maximum charging rate exceeds the preset charging rate threshold (such as 1C) to the total number of charging periods.

[0048] Further, in an embodiment of the present invention, taking the time interval between adjacent charging start times and charging end times as a charging period includes:

[0049] The time interval between adjacent charging start time and charging end time is used as a candidate charging period;

[0050] An alternative charging time period in which the state of charge of the power battery to be tested is greater than or equal to a preset state of charge threshold is selected from all the alternative charging time periods as the charging time period.

[0051] The time interval between each pair of adjacent charging start and end times is regarded as an alternative charging period, and these alternative charging periods represent the time intervals when the power battery to be tested is being charged. A SOC threshold (such as 40%) is preset based on the performance standards and safety requirements of the power battery. The cloud server detects all alternative charging periods. For any alternative charging period: as long as any SOC in the alternative charging period is greater than the SOC threshold, the alternative charging period is regarded as a charging period. In this way, the period with higher power battery charging efficiency can be selected for subsequent analysis, thereby improving the accuracy of subsequent power battery health status prediction.

[0052] Furthermore, in the calculation process of the above-mentioned maximum available capacity, due to factors such as the accuracy limitation of the sensor or measurement noise, the current value obtained has a measurement error, and this error will be particularly prominent in the long-term accumulation, resulting in the maximum available capacity finally obtained being of low accuracy. In order to reduce the impact of this cumulative error and improve the accuracy of the maximum available capacity, in one embodiment of the present invention, the battery parameters include the state of charge and the current, and obtaining the maximum available capacity of the power battery to be tested in each charging period includes:

[0053] For each charging period:

[0054] Screening out battery parameters of the power battery to be tested whose state of charge is within a preset range during the charging period, and arranging each screened battery parameter according to a time sequence to form a first sequence;

[0055] Performing interpolation processing on the first sequence, and selecting battery parameters whose current changes between adjacent data points in the interpolated first sequence are less than a preset current threshold, to form a second sequence arranged in time sequence;

[0056] The maximum available capacity of the power battery to be tested in the charging period is obtained according to the difference between the charge states at the start and the end of the second sequence and the current in the second sequence.

[0057] For each charging period: select all data points where the SOC is within a preset range (such as 40%-100%), and arrange these data points according to the timing of the corresponding sampling moments to form a first sequence. The first sequence is then interpolated to smooth the data and reduce the impact of outliers caused by measurement errors or temporary fluctuations. For example, if the sampling interval is 10 seconds, and the two adjacent SOCs in the first sequence are 10 and 20, then within this 10-second interval, a series of smooth intermediate values ​​are generated by interpolation, and these values ​​gradually increase from 10 to 20, and these values ​​may specifically include 11, 12...19.

[0058] For each data point in the interpolated first sequence, select the battery parameters whose current change between adjacent data points is less than a preset current threshold (such as 1A), and form a second sequence according to the corresponding sampling time. The second sequence obtained in this way can obtain constant current charging or relatively stable current data due to the small current change, thereby improving the accuracy of subsequent analysis. According to the difference in SOC at the start and end of the second sequence and the current values ​​and sampling intervals in the second sequence, the maximum available capacity of the power battery to be tested during the charging period is calculated according to the above formula (1).

[0059] S3. Obtaining a health status of the power battery to be tested according to the maximum available capacity of the power battery to be tested in each charging period and the rated capacity of the power battery to be tested.

[0060] The health status of a power battery is an important performance indicator used to describe the health level of a power battery relative to its new state. The health status of a power battery can be used to measure the degree of power battery capacity degradation. Specifically, in one embodiment of the present invention, the health status of the power battery to be tested is Where N is the total amount of charging time, Q i is the i-th maximum available capacity, Q rated is the rated capacity of the power battery to be tested. By comparing the ratio of the average maximum available capacity of the power battery to be tested in different effective charging periods to the rated capacity of the power battery to be tested, the health status of the power battery to be tested during the sampling period is determined. As time goes by, the health status of the power battery will gradually decrease.

[0061] S4. Obtain the performance parameters of the electrified vehicle according to the driving parameters in each discharge period.

[0062] For each discharge period: the driving parameters corresponding to each sampling moment include but are not limited to speed, acceleration pedal value (percentage) and insulation resistance (kilo ohms). According to these driving parameters, the performance parameters of the electrified vehicle in the sampling period can be obtained, wherein the performance parameters include but are not limited to the average driving speed (km / h), driving speed standard deviation, high-speed driving rate (percentage), average acceleration value (percentage), heavy acceleration value (percentage), average insulation resistance value (kΩ), insulation resistance standard deviation and insulation resistance "abnormal" rate (percentage), a total of 8 performance parameters. Specifically, the average driving speed is the average value calculated based on the speed corresponding to each sampling moment in all discharge periods. The high-speed driving rate is the proportion of sampling points whose speed exceeds the preset speed threshold (such as 80km / h) in all discharge periods to the total sampling points. The average acceleration value is the average value of the accelerator pedal values ​​at all sampling moments in all discharge periods. The heavy acceleration value is the proportion of sampling points whose accelerator pedal values ​​are greater than the preset pedal threshold (such as 60%) in all discharge periods to the total sampling points. The average insulation resistance value is the average value of all insulation resistances in all discharge periods. The "abnormal" rate of insulation resistance is the proportion of sampling points whose insulation resistance values ​​are not within a preset resistance value range (such as 4000 kΩ to 7500 kΩ) to the total sampling points during all discharge periods.

[0063] Furthermore, in addition to the above-mentioned 8 performance parameters and the 8 charging parameters obtained in step S2, the present invention also adds the number of alarms during the sampling period, mileage (kilometers), power battery capacity (Ah), cumulative mileage (kilometers) at the end of the sampling period, and sampling period, a total of 5 status parameters, which are sent to the battery health status prediction model together with the above-mentioned 16 parameters, a total of 21 behavioral parameters, to realize subsequent power battery health status prediction.

[0064] It is understandable that the division of sampling periods can be adaptively set by technicians skilled in the art based on actual needs. For example, when the sampling period is one month, the above 21 behavioral parameters are: monthly average driving speed, monthly driving speed standard deviation, monthly high-speed driving rate, monthly average acceleration value, monthly heavy acceleration value, monthly average insulation resistance value, monthly insulation resistance standard deviation, monthly insulation resistance "abnormal" rate, monthly charging times, monthly average charging current, monthly average maximum charging current, monthly average charging time, monthly average start charging SOC value, monthly average end charging SOC value, monthly average charging depth, monthly fast charging rate, monthly alarm times, monthly mileage, cumulative mileage at the end of the month, month and power battery capacity of that month.

[0065] S5. Input the performance parameter, the charging parameter and the health state into a battery health state prediction model to predict the expected health state of the power battery to be tested.

[0066] The 21 behavioral parameters of the above-mentioned current sampling period are input into the trained battery health status prediction model, and the model can predict the health status of the power battery to be tested in the next sampling period. Exemplarily, the 21 behavioral parameters of the current month are input into the battery health status prediction model, and the health status of the power battery to be tested in the next month can be predicted. It can be understood that the battery health status prediction model includes but is not limited to the Recurrent Neural Network (RNN) model, the Bidirectional Encoder Representations from Transformers (BERT) model, the Long Short Term Memory (LSTM) model, etc., and those skilled in the art can adaptively select based on actual needs, which is not limited here. Preferably, in the present invention, the battery health status prediction model is a long short-term memory network model.

[0067] In one embodiment of the present invention, the battery health state prediction model is obtained by training, and the training method of the battery health state prediction model includes:

[0068] Sampling the battery parameters of the power battery to be tested and the driving parameters of the corresponding electrified vehicle, obtaining the battery parameters and driving parameters in a plurality of preset sampling periods, dividing the sampling period into a charging period and a discharging period according to the battery parameters, and obtaining the real health status of the power battery to be tested in each sampling period;

[0069] For each sampling period:

[0070] Obtaining charging parameters of the power battery to be tested and the maximum available capacity in each charging period according to the battery parameters in each charging period;

[0071] Obtaining the health status of the power battery to be tested in the current sampling period according to the maximum available capacity of the power battery to be tested in each charging period and the rated capacity of the power battery to be tested;

[0072] Obtaining performance parameters of the electrified vehicle according to the driving parameters in each discharge period;

[0073] Inputting the performance parameters, charging parameters and health status of the power battery to be tested in the current sampling period into the battery health status prediction model to predict the expected health status of the power battery to be tested in the next sampling period;

[0074] Based on the difference between the expected health state of the power battery to be tested in the next sampling period and the corresponding actual health state, the battery health state prediction model is updated to obtain a trained battery health state prediction model.

[0075] The battery parameters of the power battery to be tested and the driving parameters of the corresponding electrified vehicle are continuously collected within a preset sampling period (such as one month), and the sampling period is divided into one or more charging periods and one or more discharging periods according to the changes in the collected battery parameters. For each charging period, the corresponding maximum available capacity is calculated using the above formula (1), and then the health status of the power battery to be tested in the current sampling period can be calculated based on the maximum available capacity and the rated capacity of the power battery to be tested. Then, the charging parameters of the current sampling period can be obtained by analyzing the battery parameters of each charging period within the sampling period, and the performance parameters of the current sampling period can be obtained by analyzing the driving parameters of each discharging period. The charging parameters, performance parameters and health status of the current sampling period are input into the battery health status prediction model, and the model will output the expected health status of the power battery to be tested in the next sampling period. The difference between the expected health status and the actual health status of the power battery to be tested in the next sampling period is calculated by the loss function, and the parameters of the model are updated according to the difference. The model is continuously iterated until the loss function converges to obtain a trained battery health status prediction model.

[0076] It should be noted that the training data in the present invention can be obtained by self-collection or through public data sources. In one example, the training data in the present invention comes from the Shanghai New Energy Vehicle Public Data Collection and Monitoring Research Center (EVData), which provides the operation data of 200 private cars from June 2022 to June 2023, with a sampling interval of 10 seconds. These vehicles include a total of 6 different models, and the rated capacity of the battery cell materials and batteries of each model is different. However, considering that some vehicle data are less, 99 vehicles with a time span of more than 10 months are selected from 200 vehicles in the present invention for constructing training sets and test sets. And 6 groups of training sets and test sets are constructed using the cross-validation method (each group of training sets and test sets have the same division ratio, the training set contains 93 vehicles, the test set contains 6 vehicles, and the 6 vehicles in each test set are not repeated). Six LSTM models are trained through these 6 groups of training sets and test sets, and the model with the smallest training error is selected as the final battery health state prediction model. In the test phase, taking the sampling period of one month as an example, the 21-dimensional feature vector of the starting month of each vehicle in the test set (including the 20 behavioral parameters of the current month and the health status of the power battery in the current month) is used as the input of the trained battery health status prediction model to predict the health status of the power battery in the next month. Then the predicted health status and the 20 behavioral parameters of the next month are combined to form a new 21-dimensional vector, which is input into the battery health prediction model again for single-step iterative prediction.

[0077] The trained battery health status prediction model has learned the impact of input features on output results. By adjusting the weight of input features on output, it can accurately estimate the influence of each input (health status and related parameters in the current sampling period) on the prediction result (i.e. health status in the next sampling period). The model performed well on the test set, with an average root mean square error of 0.0112 and a maximum error of no more than 1.8%, showing the high accuracy and reliability of the model.

[0078] See also Figure 2 , the power battery health status prediction system 100 based on user behavior includes: a data acquisition module 110, a maximum available capacity acquisition module 120, a health status acquisition module 130, a parameter acquisition module 140 and a prediction module 150. The data acquisition module 110 is used to sample the battery parameters of the power battery to be tested and the driving parameters of the corresponding electric vehicle, obtain the battery parameters and driving parameters within a preset sampling period, and divide the sampling period into a charging period and a discharging period according to the battery parameters. The maximum available capacity acquisition module 120 is used to obtain the charging parameters of the power battery to be tested and the maximum available capacity in each charging period according to the battery parameters in each charging period. The health status acquisition module 130 is used to obtain the health status of the power battery to be tested according to the maximum available capacity of the power battery to be tested in each charging period and the rated capacity of the power battery to be tested. The parameter acquisition module 140 is used to obtain the performance parameters of the electric vehicle according to the driving parameters in each discharging period. The prediction module 150 is used to input the performance parameters, charging parameters and the health status into the battery health status prediction model to predict the expected health status of the power battery to be tested.

[0079] For the specific limitations of the power battery health status prediction system based on user behavior, please refer to the limitations of the power battery health status prediction method based on user behavior above, which will not be repeated here. Each module in the above-mentioned power battery health status prediction system based on user behavior can be implemented in whole or in part through software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in hardware format, or can be stored in the memory of the computer device in software format, so that the processor can call the corresponding operations of each of the above modules.

[0080] It should be noted that, in order to highlight the innovative part of the present invention, the present embodiment does not introduce modules that are not closely related to solving the technical problem proposed by the present invention, but this does not mean that there are no other modules in the present embodiment.

[0081] In summary, the present invention discloses a method and system for predicting the health status of a power battery based on user behavior, which effectively reduces the cumulative error that may be introduced by traditional calculation methods by constraining the state of charge and current during the charging period. The constructed battery health status prediction model takes the user's charging and driving behavior parameters as input and the health status of the power battery as output, thereby accurately capturing the specific impact of these behavior parameters on the health status of the power battery. This method not only improves the accuracy of the prediction, but also provides a scientific basis for power battery maintenance and management, helps to extend the service life of the power battery and optimize the overall performance of the electrified vehicle. Therefore, the present invention effectively overcomes the various shortcomings of the prior art and has a high industrial utilization value.

[0082] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the art may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by a person of ordinary skill in the art without departing from the spirit and technical concept disclosed by the present invention shall still be covered by the claims of the present invention.

Claims

1. A method for predicting the health status of a power battery based on user behavior, characterized in that: Applied to an electric vehicle, the method comprises: Sampling the battery parameters of the power battery to be tested and the driving parameters of the corresponding electric vehicle, obtaining the battery parameters and the driving parameters within a preset sampling period, and dividing the sampling period into a charging period and a discharging period according to the battery parameters; Obtaining charging parameters of the power battery to be tested and the maximum available capacity in each charging period according to the battery parameters in each charging period; Obtaining a health status of the power battery to be tested according to the maximum available capacity of the power battery to be tested in each charging period and the rated capacity of the power battery to be tested; Obtaining performance parameters of the electrified vehicle according to the driving parameters in each discharge period; Inputting the performance parameter, the charging parameter and the health state into a battery health state prediction model to predict the expected health state of the power battery to be tested; The battery health state prediction model is obtained through training, and the training method of the battery health state prediction model includes: Sampling the battery parameters of the power battery to be tested and the driving parameters of the corresponding electrified vehicle, obtaining the battery parameters and driving parameters in a plurality of preset sampling periods, dividing the sampling period into a charging period and a discharging period according to the battery parameters, and obtaining the real health status of the power battery to be tested in each sampling period; For each sampling period: Obtaining charging parameters of the power battery to be tested and the maximum available capacity in each charging period according to the battery parameters in each charging period; Obtaining the health status of the power battery to be tested in the current sampling period according to the maximum available capacity of the power battery to be tested in each charging period and the rated capacity of the power battery to be tested; The performance parameters of the electric vehicle are obtained according to the driving parameters in each discharge period; the performance parameters, charging parameters and health status of the power battery to be tested in the current sampling period are input into the battery health status prediction model to predict the expected health status of the power battery to be tested in the next sampling period; Based on the difference between the expected health state of the power battery to be tested in the next sampling period and the corresponding actual health state, the battery health state prediction model is updated to obtain a trained battery health state prediction model.

2. The method for predicting the health status of a power battery based on user behavior according to claim 1, characterized in that: The battery parameter is the state of charge, and the process of dividing the sampling period into charging periods according to the battery parameter includes: Arranging all battery parameters within the sampling period in time sequence to form a battery parameter sequence; Determining the charging start time and the charging end time according to the change of each battery parameter in the battery parameter sequence; The time interval between adjacent charging start times and charging end times is regarded as a charging period.

3. The method for predicting the health status of a power battery based on user behavior according to claim 2, characterized in that: The step of arranging all battery parameters within the sampling period in time sequence to form a battery parameter sequence includes: Remove missing and abnormal battery parameters from all battery parameters; The remaining battery parameters are arranged in time sequence to form a battery parameter sequence.

4. The method for predicting the health status of a power battery based on user behavior according to claim 2, characterized in that: Determining the charging start time and the charging end time according to the change of the battery parameters in the battery parameter sequence includes: For any battery parameter in the battery parameter sequence: If the current battery parameter is greater than the previous battery parameter and less than or equal to the next battery parameter, the sampling time corresponding to the current battery parameter is the charging start time; If the current battery parameter is greater than the previous battery parameter and greater than or equal to the next battery parameter, the sampling time corresponding to the current battery parameter is the charging end time.

5. The method for predicting the health status of a power battery based on user behavior according to claim 2, characterized in that: Obtaining the maximum available capacity of the power battery to be tested in each charging period, including: For any charging period: The maximum available capacity of the power battery to be tested in the charging period is obtained according to the difference in the state of charge of the power battery to be tested at the beginning and the end of the charging period and the current in the charging period.

6. The method for predicting the health status of a power battery based on user behavior according to claim 2, characterized in that: The time interval between adjacent charging start times and charging end times is regarded as a charging period, including: The time interval between adjacent charging start time and charging end time is used as a candidate charging period; An alternative charging time period in which the state of charge of the power battery to be tested is greater than or equal to a preset state of charge threshold is selected from all the alternative charging time periods as the charging time period.

7. The method for predicting the health status of a power battery based on user behavior according to claim 5, characterized in that: The battery parameters include the state of charge and current, and obtaining the maximum available capacity of the power battery to be tested in each charging period includes: For each charging period: Screening out battery parameters of the power battery to be tested whose state of charge is within a preset range during the charging period, and arranging each screened battery parameter according to a time sequence to form a first sequence; Performing interpolation processing on the first sequence, and selecting battery parameters whose current changes between adjacent data points in the interpolated first sequence are less than a preset current threshold, to form a second sequence arranged in time sequence; The maximum available capacity of the power battery to be tested in the charging period is obtained according to the difference in the charge state at the beginning and the end of the second sequence and the current in the second sequence.

8. The method for predicting the health status of a power battery based on user behavior according to claim 1, characterized in that: The health status of the power battery to be tested Where N is the total amount of charging time, Q i is the i-th maximum available capacity, Q rated is the rated capacity of the power battery to be tested.

9. A power battery health status prediction system based on user behavior, characterized in that: The system comprises: A data acquisition module, used to sample the battery parameters of the power battery to be tested and the driving parameters of the corresponding electric vehicle, obtain the battery parameters and driving parameters within a preset sampling period, and divide the sampling period into a charging period and a discharging period according to the battery parameters; A maximum available capacity acquisition module, used to obtain the charging parameters of the power battery to be tested and the maximum available capacity in each charging period according to the battery parameters in each charging period; A health status acquisition module, used to obtain the health status of the power battery to be tested according to the maximum available capacity of the power battery to be tested in each charging period and the rated capacity of the power battery to be tested; A parameter acquisition module, used to obtain the performance parameters of the electrified vehicle according to the driving parameters in each discharge period; A prediction module, used for inputting the performance parameter, the charging parameter and the health state into a battery health state prediction model to predict the expected health state of the power battery to be tested; The battery health state prediction model is obtained through training, and the training method of the battery health state prediction model includes: Sampling the battery parameters of the power battery to be tested and the driving parameters of the corresponding electrified vehicle, obtaining the battery parameters and driving parameters in a plurality of preset sampling periods, dividing the sampling period into a charging period and a discharging period according to the battery parameters, and obtaining the real health status of the power battery to be tested in each sampling period; For each sampling period: Obtaining charging parameters of the power battery to be tested and the maximum available capacity in each charging period according to the battery parameters in each charging period; Obtaining the health status of the power battery to be tested in the current sampling period according to the maximum available capacity of the power battery to be tested in each charging period and the rated capacity of the power battery to be tested; The performance parameters of the electric vehicle are obtained according to the driving parameters in each discharge period; the performance parameters, charging parameters and health status of the power battery to be tested in the current sampling period are input into the battery health status prediction model to predict the expected health status of the power battery to be tested in the next sampling period; Based on the difference between the expected health state of the power battery to be tested in the next sampling period and the corresponding actual health state, the battery health state prediction model is updated to obtain a trained battery health state prediction model.

Images