Battery SOH vehicle cloud collaborative estimation method and system suitable for multiple charging modes

Abstract

The invention relates to the technical field of battery management, and particularly discloses a battery SOH vehicle cloud collaborative estimation method and system suitable for multiple charging modes, and the method comprises the steps: obtaining the current charging data of a battery, and calculating the current charging capacity of the battery in an optimal charging SOC interval according to the current charging data of the battery; performing normalization processing on the current charging capacity of the battery in the optimal charging SOC interval to obtain the normalized current charging capacity of the battery in the optimal charging SOC interval; and inputting the normalized current charging capacity of the battery in the optimal charging SOC interval into the trained charging aging model so as to output a current health state estimation value of the battery. According to the battery SOH vehicle cloud collaborative estimation method adaptive to multiple charging modes provided by the invention, online estimation of the SOH of the lithium battery is realized by utilizing computing resources of the cloud server, the method does not depend on complete full charge data, the accuracy of online estimation of the SOH is improved, and the method is adaptive to multiple charging modes and has relatively high applicability and practicability.

Description

Technical Field

[0001] The present invention relates to the technical field of battery management, and in particular to a battery SOH vehicle-cloud collaborative estimation method adapting to multiple charging modes and a battery SOH vehicle-cloud collaborative estimation system adapting to multiple charging modes. Background Art

[0002] The performance of lithium-ion batteries will decay over time. Usually, the battery capacity declines to 80% of the initial capacity or the battery internal resistance doubles, which means the battery reaches the end of life (EOL). As the battery ages, the battery's usable driving range decreases, and the battery safety risk increases. Therefore, accurate assessment of the battery health status is of great significance in the energy management and safety warning of the entire life cycle of automotive power batteries.

[0003] The current battery health status SOH algorithms can be divided into three categories: experimental-based methods, battery model-based methods, and data-driven methods. The experimental-based method needs to be carried out under strictly controlled test conditions, and the battery is fully charged and discharged under constant temperature and constant current conditions to obtain the battery capacity. This method is limited to laboratory research and cannot be used for online SOH evaluation. The battery model-based method simulates the dynamic characteristics of the battery during charging and discharging by establishing a battery model to evaluate the SOH. Battery models include empirical models or semi-empirical models, equivalent circuit models, and electrochemical models. The empirical model or semi-empirical model establishes a formula for factors such as temperature, charge and discharge rate, and depth of discharge and battery aging based on aging test data. The difference between the test conditions and the actual vehicle operating conditions leads to poor adaptability of this model. The equivalent circuit model can only be targeted at a certain battery and is not universal. In addition, the optimization model requires a large amount of calculation and takes a long time. The electrochemical model is limited in its application in actual battery management systems due to its high model calculation complexity. The data-driven method does not require in-depth research on the electrochemical mechanism of the battery, nor does it require the construction of a battery model. Instead, it extracts characteristic parameters related to the health status of the battery from a large amount of test data or actual vehicle operation data, and builds a big data model of characteristic parameters and SOH through machine learning, intelligent optimization algorithms and other methods. The data-driven method usually includes four aspects: data acquisition and preprocessing, characteristic parameter extraction, establishment of estimation model and training, and model deployment. Characteristic parameter extraction refers to extracting features that can reflect the changes in its health status from easily accessible parameters such as current, voltage, and temperature to characterize the attenuation of battery capacity during the aging process. It includes (1) differential analysis method: extracting the position, height, area, and distance between different peaks of the incremental capacity (IC) curve as characteristic parameters; (2) direct parameter analysis method: extracting the constant current and constant voltage period, the area under the current curve, the area under the temperature curve, the maximum slope of the voltage and current curve, the charging time used in the fixed voltage interval, the voltage change in the same time period, the open circuit voltage in the fully charged static state, and the probability density function of the voltage curve as characteristic parameters. Characteristic parameter extraction is the key step of the data-driven method.

[0004] The data-driven method has high flexibility, but it has high requirements for the quantity and quality of training data. Model training and big data deduction consume a lot of CPU resources and need to be implemented on hardware with strong computing resources to meet actual performance requirements. It is difficult for automotive embedded chips to meet the resource requirements of data-driven models.

[0005] The current SOH estimation methods have the following problems:

[0006] (1) The SOH estimation methods based on experiments and battery models have low accuracy and poor adaptability in practical applications: Most of the existing SOH models are empirical models or semi-empirical models based on laboratory test data. The experiments are conducted at specific temperatures, charge and discharge rates, and depths of discharge. However, actual vehicles rarely operate under fixed operating conditions. It is difficult to track the aging trajectory of actual vehicle batteries by building SOH models based on experimental data. The experimental method cannot be used for online SOH evaluation. The prediction accuracy of equivalent circuit models and empirical models is low and their adaptability is poor in practical applications. The electrochemical model method is too complex to be applied.

[0007] (2) Implementing and deploying data-driven estimation of SOH on vehicle-mounted BMS embedded chips has the problem of insufficient computing resources: the life cycle of automotive lithium-ion batteries is as long as 10-15 years, and the health status of the battery is estimated based on a long time scale. However, the storage resources and CPU computing resources of vehicle-mounted BMS embedded chips are limited, and cannot cope with the increasingly complex SOH estimation algorithms and the processing of large data sets.

[0008] (3) Actual application conditions are changeable, and the characteristic parameters extracted by current technologies have poor adaptability in actual applications: Existing data-driven SOH models often extract characteristic parameters and train models through complete charging curves. However, actual applications usually use partial charging data, so it is difficult to extract characteristic parameters consistent with the pre-trained model in actual applications, which leads to low estimation accuracy. The data-driven method requires a large amount of training data, and model training and deduction require a lot of computing resources. The on-board BMS cannot meet this requirement. The characteristic parameter extraction process of the data-driven method usually requires a complete charging of the battery, but it is difficult to ensure a complete charging process in actual applications.

[0009] (4) The current SOH estimation method is difficult to apply to various charging modes: The current estimation strategy either requires constant voltage charging information or constant current charging information, which is difficult to apply to various charging modes of electric vehicles. Summary of the invention

[0010] In view of the defects and shortcomings in the prior art, the present invention provides a battery SOH vehicle-cloud collaborative estimation method that is adaptable to multiple charging modes. It uses the computing resources of the cloud server to realize the online estimation of the SOH of the lithium battery. It does not rely on complete full charge data, improves the accuracy of the online SOH estimation, and is adaptable to multiple charging modes. It has strong applicability and practicality.

[0011] As a first aspect of the present invention, a battery SOH vehicle-cloud collaborative estimation method adapting to multiple charging modes is provided, and the battery SOH vehicle-cloud collaborative estimation method adapting to multiple charging modes includes:

[0012] Step S1: obtaining current charging data of the battery, and calculating the current charging capacity of the battery in the optimal charging SOC interval according to the current charging data of the battery;

[0013] Step S2: normalizing the current charging capacity of the battery in the optimal charging SOC interval to obtain the normalized current charging capacity of the battery in the optimal charging SOC interval;

[0014] Step S3: inputting the normalized current charging capacity of the battery in the optimal charging SOC interval into the trained charging aging model to output an estimated value of the current health state of the battery.

[0015] Furthermore, before step S1, the method further includes:

[0016] Obtaining battery historical aging data; wherein the battery historical aging data includes battery historical charging data and its corresponding battery historical discharging data, and the battery historical charging data is divided into battery historical slow charging data and battery historical fast charging data according to the charging mode;

[0017] Calculate the battery's historical slow-charge capacity in the optimal charging SOC interval based on the battery's historical slow-charge data, and perform normalization processing; and calculate the first battery's historical health state based on the first battery's historical discharge data corresponding to the battery's historical slow-charge data;

[0018] The historical fast-charge capacity of the battery in the optimal charging SOC interval is calculated according to the historical fast-charge data of the battery, and normalized; meanwhile, the historical health state of the second battery is calculated according to the historical discharge data of the second battery corresponding to the historical fast-charge data of the battery;

[0019] Analyze the correlation between the historical slow charging capacity of the battery in the normalized optimal charging SOC interval and the historical health state of the first battery, and construct a slow charging data set with the historical slow charging capacity of the battery in the normalized optimal charging SOC interval and the historical health state of the first battery; at the same time, analyze the correlation between the historical fast charging capacity of the battery in the normalized optimal charging SOC interval and the historical health state of the second battery, and construct a fast charging data set with the historical fast charging capacity of the battery in the normalized optimal charging SOC interval and the historical health state of the second battery;

[0020] The slow charging aging model is trained according to the slow charging data set to obtain a trained slow charging aging model; meanwhile, the fast charging aging model is trained according to the fast charging data set to obtain a trained fast charging aging model; wherein the trained slow charging aging model and the trained fast charging aging model are both stored in a cloud server.

[0021] Further, the training of the slow charging aging model according to the slow charging data set to obtain the trained slow charging aging model; and the training of the fast charging aging model according to the fast charging data set to obtain the trained fast charging aging model also include:

[0022] The slow charging data set is divided into a slow charging training set, a slow charging verification set and a slow charging test set; and the fast charging data set is divided into a fast charging training set, a fast charging verification set and a fast charging test set;

[0023] The slow charging aging model is trained using the slow charging training set. During the training process, the slow charging validation set is used to validate the slow charging aging model to prevent the slow charging aging model from overfitting. After the training is completed, the slow charging test set is used to test and evaluate the performance of the trained slow charging aging model.

[0024] The fast charging training set is used to train the fast charging aging model. During the training process, the fast charging verification set is used to verify the fast charging aging model to prevent the fast charging aging model from overfitting. After the training is completed, the fast charging test set is used to test and evaluate the performance of the trained fast charging aging model.

[0025] Furthermore, it also includes:

[0026] Determine whether the vehicle is in slow charging mode or fast charging mode each time it is charged;

[0027] If the vehicle is charged in slow charging mode, the current slow charging data of the battery is obtained; when the vehicle passes through the optimal charging SOC interval during slow charging, the current slow charging capacity of the battery in the optimal charging SOC interval is calculated according to the current slow charging data of the battery; then the current slow charging capacity of the battery in the optimal charging SOC interval is normalized to obtain the normalized current slow charging capacity of the battery in the optimal charging SOC interval; the normalized current slow charging capacity of the battery in the optimal charging SOC interval is input into the trained slow charging aging model to output a first battery current health status estimation value;

[0028] If the vehicle is charged in fast charging mode, the current fast charging data of the battery is obtained; when the vehicle passes through the optimal charging SOC interval during fast charging, the current fast charging capacity of the battery in the optimal charging SOC interval is calculated according to the current fast charging data of the battery; then the current fast charging capacity of the battery in the optimal charging SOC interval is normalized to obtain the normalized current fast charging capacity of the battery in the optimal charging SOC interval; the normalized current fast charging capacity of the battery in the optimal charging SOC interval is input into the trained fast charging aging model to output a second battery current health state estimation value;

[0029] Arrange the first battery current health state estimation value or the second battery current health state estimation value output each time in chronological order to obtain an arranged battery current health state estimation value;

[0030] The arranged battery current health state estimation value is filtered by a sliding average filtering method to obtain a filtered battery current health state estimation value.

[0031] Furthermore, it also includes:

[0032] Extracting battery historical charging data from the battery historical aging data every N cycles, wherein the battery historical charging data includes charging time, charging current and charging voltage; assuming that the battery is charged L times in total during its complete life cycle, and the battery historical charging data is extracted every 10 cycles, then a total of M times of battery historical charging data is extracted, that is, there are a total of M groups of battery historical charging data, M=L / 10;

[0033] Preprocessing the battery historical charging data extracted each time to obtain a plurality of preprocessed battery historical charging data;

[0034] The battery charging capacity Q of each historical charging process is calculated according to the plurality of pre-processed battery historical charging data. fullchrg and battery charge SOC, where the battery charge capacity Q fullchrg And the calculation formula of battery charging SOC is as follows:

[0035]

[0036] Where I is the charging current, t0 is the time when charging starts, t1 is the time when charging ends, SOC0 is the initial state of charge of the battery, and Qmax is the maximum available capacity of the battery;

[0037] Determine the optimal charging SOC interval, and calculate the battery historical charging capacity Q of the optimal charging SOC interval according to the battery charging SOC in each historical charging process opt ; Among them, the optimal charging SOC interval is set to [50%, 80%], then the battery historical charging capacity Q in this interval opt The calculation formula is as follows:

[0038]

[0039] Among them, t_SOC0 is the charging time when the battery charging SOC is 50%, and t_SOC1 is the charging time when the battery charging SOC is 80%;

[0040] The Min-Max normalization method is used to calculate the battery historical charging capacity Q in the optimal charging SOC interval. opt Normalization is performed to obtain the battery historical charging capacity Q of the normalized optimal charging SOC interval opt_norm ; The normalization formula is as follows:

[0041] Q opt_norm =(Q opt -min(Q opt )) / (max(Q opt )-min(Q opt ))

[0042] Among them, min(Q opt ) represents the historical charging capacity Q of group M batteries opt The minimum value in the opt Represents the historical charging capacity Q of group M batteries opt The maximum value in .

[0043] Furthermore, it also includes:

[0044] Extracting battery historical discharge data every N cycles from the battery historical aging data, wherein the extracted battery historical discharge data and the extracted battery historical charging data are discharge data and charging data of the same aging cycle, respectively;

[0045] The battery historical health status SOH during each historical discharge process is calculated based on the battery historical discharge data extracted each time. The calculation formula is as follows:

[0046]

[0047] Among them, Qnow is the current maximum available capacity of the battery, I is the discharge current, t0_dis and t1_dis are the start time and end time of discharge respectively, and Qnew is the rated capacity of the battery when it leaves the factory.

[0048] As a second aspect of the present invention, a battery SOH vehicle-cloud collaborative estimation system adapting to multiple charging modes is provided, which is used to implement the battery SOH vehicle-cloud collaborative estimation method adapting to multiple charging modes as described above, and the battery SOH vehicle-cloud collaborative estimation system adapting to multiple charging modes includes a vehicle, a vehicle-side BMS and a cloud server; wherein the vehicle includes a battery, and the battery can be charged and discharged to provide power for the vehicle;

[0049] The vehicle-side BMS is used to obtain the current charging data of the battery, and calculate the current charging capacity of the battery in the optimal charging SOC interval according to the current charging data of the battery; then normalize the current charging capacity of the battery in the optimal charging SOC interval to obtain the normalized current charging capacity of the battery in the optimal charging SOC interval;

[0050] The cloud server is used to input the current charging capacity of the battery in the normalized optimal charging SOC interval into the trained charging aging model to output an estimated value of the current health state of the battery.

[0051] Furthermore, the vehicle-side BMS includes a slave controller and a main controller. The slave controller collects voltage data and temperature data of the battery through the front-end sampling chip AFE, and transmits the collected voltage data and temperature data to the main controller through CAN communication or daisy chain communication; the main controller collects current data of the battery through a current sensor, and calculates the current charging capacity of the battery in the optimal charging SOC range through the MCU and normalizes it, and then uploads the normalized current charging capacity of the battery in the optimal charging SOC range to the cloud server through the 4G communication module carried by the vehicle-side BMS.

[0052] Furthermore, the cloud server calls the trained charging aging model stored in itself according to the received normalized current charging capacity of the battery in the optimal charging SOC interval to output an estimated value of the current health state SOH of the battery.

[0053] Furthermore, the cloud server will store the battery's current state of health SOH estimate outputted during each charging, and obtain a filtered and stable battery's current state of health SOH estimate through data fusion and filtering algorithms, and then transmit the stable battery's current state of health SOH estimate back to the vehicle-side BMS through the 4G communication module.

[0054] The battery SOH vehicle-cloud collaborative estimation method adapted to multiple charging modes provided by the present invention has the following advantages:

[0055] (1) The present invention proposes a SOH estimation algorithm based on big data drive, which can effectively avoid the problem of large estimation errors of empirical models or semi-empirical models in practical applications and has better applicability compared with traditional empirical models or semi-empirical models.

[0056] (2) The present invention proposes a vehicle-cloud collaborative estimation method, which can solve the contradiction between the complexity of the SOH estimation algorithm and the insufficient computing resources of the vehicle-side BMS;

[0057] (3) The present invention proposes a method for extracting characteristic parameters based on partial charging curves, which can improve the adaptability of the SOH model in practical applications;

[0058] (4) The present invention proposes a SOH estimation method that can adapt to multiple charging modes, which can improve the adaptability of the SOH model in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the present invention, but do not constitute a limitation of the present invention.

[0060] Figure 1 A flow chart of the battery SOH vehicle-cloud collaborative estimation method adapted to multiple charging modes provided by the present invention.

[0061] Figure 2 A flowchart of a specific implementation of the battery SOH vehicle-cloud collaborative estimation method that is adaptable to multiple charging modes provided by the present invention.

[0062] Figure 3 This is a structural diagram of the battery SOH vehicle-cloud collaborative estimation system that is adaptable to multiple charging modes provided by the present invention.

[0063] Figure 4a This is a slow charging current curve diagram of the lithium-ion battery provided by the present invention.

[0064] Figure 4b This is a fast charging current curve diagram of the lithium-ion battery provided by the present invention.

[0065] Figure 5 A schematic diagram of the correlation analysis between the charging capacity and SOH in the optimal charging SOC interval provided by the present invention.

[0066] Figure 6a This is a diagram of the slow charging SOH estimation result provided by the present invention.

[0067] Figure 6b This is a diagram of the fast charging SOH estimation results provided by the present invention.

[0068] Figure 7 This is a diagram of the SOH data fusion and filtering results provided by the present invention. DETAILED DESCRIPTION

[0069] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0070] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work should fall within the scope of protection of the present invention.

[0071] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so as to describe the embodiments of the present invention described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes 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 devices.

[0072] In this embodiment, a battery SOH vehicle-cloud collaborative estimation method that is adaptable to multiple charging modes is provided. Figure 1 The flowchart of the battery SOH vehicle-cloud collaborative estimation method adapted to multiple charging modes provided by the present invention is as follows: Figure 1 As shown, the battery SOH vehicle-cloud collaborative estimation method adapted to multiple charging modes includes:

[0073] Step S1: obtaining current charging data of the battery, and calculating the current charging capacity of the battery in the optimal charging SOC interval according to the current charging data of the battery;

[0074] Preferably, Figure 2 As shown, before step S1, it also includes:

[0075] Obtain battery historical aging data; wherein the battery historical aging data includes battery historical charging data and its corresponding battery historical discharge data, and the battery historical charging data is divided into battery historical slow charging data and battery historical fast charging data according to the charging mode; usually, slow charging is constant current charging, and fast charging adopts multi-stage charging current with different rates for charging. Figure 4a is the slow charging current curve, such as Figure 4b This is the fast charging current curve.

[0076] Filter out the optimal charging SOC interval in the slow charging mode, calculate the battery historical slow charging capacity in the optimal charging SOC interval according to the battery historical slow charging data, and perform normalization processing; and calculate the first battery historical health state according to the first battery historical discharge data corresponding to the battery historical slow charging data;

[0077] Filter out the optimal charging SOC interval in the fast charging mode, calculate the battery's historical fast charging capacity in the optimal charging SOC interval based on the battery's historical fast charging data, and perform normalization processing; and calculate the second battery's historical health status based on the second battery's historical discharge data corresponding to the battery's historical fast charging data;

[0078] The Pearson correlation analysis method is used to analyze the correlation between the battery historical slow charging capacity and the first battery historical health status in the normalized optimal charging SOC interval, such as Figure 5 As shown, in the slow charging mode, the correlation coefficient is r=0.954; and the slow charging data set is constructed with the battery historical slow charging capacity in the normalized optimal charging SOC interval as the independent variable and the first battery historical health status as the dependent variable;

[0079] The Pearson correlation analysis method is used to analyze the correlation between the battery's historical fast-charge capacity and the second battery's historical health status in the normalized optimal charging SOC interval, such as Figure 5 As shown, in the fast charging mode, the correlation coefficient is r=0.988; and the fast charging data set is constructed with the battery historical fast charging capacity in the normalized optimal charging SOC interval as the independent variable and the second battery historical health status as the dependent variable;

[0080] The slow charging aging model is trained according to the slow charging data set to obtain a trained slow charging aging model; and the fast charging aging model is trained according to the fast charging data set to obtain a trained fast charging aging model.

[0081] Specifically, it also includes:

[0082] Extracting battery historical charging data from the battery historical aging data every N cycles, wherein the battery historical charging data includes charging time, charging current and charging voltage; assuming that the battery is charged L times in total during its full life cycle (SOH drops from 100% to 80%), and extracting battery historical charging data every 10 cycles, then extracting M times of battery historical charging data in total, that is, there are M groups of battery historical charging data, M=L / 10;

[0083] Preprocessing the battery historical charging data extracted each time to obtain a plurality of preprocessed battery historical charging data;

[0084] It should be noted that due to sampling errors and measurement noise, some current or voltage sampling data are abnormal, so it is necessary to check each group of battery historical charging data (charging time-charging current, charging time-charging voltage), delete the outlier battery historical charging data in the M groups of battery historical charging data, and interpolate the lost battery historical charging data.

[0085] In this example, for each extracted battery historical charging data, the battery historical charging data that differs from the mean by more than three standard deviations within the data window with a window length of 10 is defined as outlier battery historical charging data, and the outlier battery historical charging data will be deleted. The missing battery historical charging data will be supplemented by the linear interpolation method of adjacent non-missing values.

[0086] According to the plurality of pre-processed battery historical charging data, the battery charging capacity Q of each historical charging process is calculated by using the ampere-hour integration method. fullchrg and battery charge SOC, where the battery charge capacity Q fullchrg And the calculation formula of battery charging SOC is as follows:

[0087]

[0088] Where I is the charging current, t0 is the time when charging starts, t1 is the time when charging ends, SOC0 is the initial state of charge of the battery, and Qmax is the maximum available capacity of the battery;

[0089] The optimal charging SOC interval is determined based on the charging SOC interval commonly used by electric vehicle users, and the battery historical charging capacity Q of the optimal charging SOC interval is calculated using the ampere-hour integration method based on the battery charging SOC during each historical charging process. opt ; Among them, the optimal charging SOC interval is set to [50%, 80%], then the battery historical charging capacity Q in this interval opt The calculation formula is as follows:

[0090]

[0091] Among them, t_SOC0 is the charging time when the battery charging SOC is 50%, and t_SOC1 is the charging time when the battery charging SOC is 80%;

[0092] It should be noted that the commonly used charging SOC interval for electric vehicles is [30%, 80%]. The optimal charging SOC interval is shortened to [50%, 80%]. In this way, the probability of each charge passing through the defined optimal charging SOC interval is higher, thereby further increasing the frequency of SOH estimation; if a certain charge does not pass through the defined optimal charging SOC interval, the SOH estimation of this charge is skipped. Because battery aging is a long-term process, the loss of a certain charging data will not have a significant impact on the overall SOH estimation value.

[0093] The Min-Max normalization method is used to calculate the battery historical charging capacity Q in the optimal charging SOC interval. opt Normalization is performed to obtain the battery historical charging capacity Q of the normalized optimal charging SOC interval opt_norm ; The normalization formula is as follows:

[0094] Q opt_norm =(Q opt -min(Q opt )) / (max(Q opt )-min(Q opt ))

[0095] Among them, min(Q opt ) represents the historical charging capacity Q of group M batteries opt The minimum value in the opt Represents the historical charging capacity Q of group M batteries opt The maximum value in .

[0096] Specifically, it also includes:

[0097] Extracting battery historical discharge data every N cycles from the battery historical aging data, wherein the extracted battery historical discharge data and the extracted battery historical charging data are discharge data and charging data of the same aging cycle, respectively;

[0098] The battery historical health status SOH during each historical discharge process is calculated based on the battery historical discharge data extracted each time. The calculation formula is as follows:

[0099]

[0100] Among them, Qnow is the current maximum available capacity of the battery, which is calculated by integrating the ampere-hours during the discharge process; I is the discharge current, t0_dis and t1_dis are the start and end times of discharge, respectively, and Qnew is the rated capacity of the battery when it leaves the factory, which is given by the battery manufacturer.

[0101] Specifically, the training of the slow charging aging model according to the slow charging data set to obtain the trained slow charging aging model; and the training of the fast charging aging model according to the fast charging data set to obtain the trained fast charging aging model also include:

[0102] The slow charging data set is divided into a slow charging training set, a slow charging verification set and a slow charging test set; and the fast charging data set is divided into a fast charging training set, a fast charging verification set and a fast charging test set;

[0103] Select a suitable machine learning model, use the slow charging training set to train the slow charging aging model, and during the training process, use the slow charging validation set to verify the slow charging aging model to prevent the slow charging aging model from overfitting; after the training is completed, use the slow charging test set to test and evaluate the performance of the trained slow charging aging model; the evaluation indicators include root mean square error (RMSE), mean absolute error (MAE) and maximum absolute error (MAX), and the evaluation results are used to guide the selection of the machine learning algorithm;

[0104] Select a suitable machine learning model, use the fast charging training set to train the fast charging aging model, and during the training process, use the fast charging validation set to verify the fast charging aging model to prevent the fast charging aging model from overfitting; after the training is completed, use the fast charging test set to test and evaluate the performance of the trained fast charging aging model; the evaluation indicators include root mean square error (RMSE), mean absolute error (MAE) and maximum absolute error (MAX), and the evaluation results are used to guide the selection of machine learning algorithms.

[0105] It should be noted that battery aging is a process that gradually changes over time. Recurrent neural networks have a strong fitting ability for previously and subsequently related time series data. The gated recurrent unit (GRU) adds a gating mechanism on the basis of the recurrent neural network, which can better capture the dependencies with larger step sizes in the time series and solve the problems of gradient attenuation or explosion in long-term memory and back propagation. GRU is selected to train the slow charging training set, and the slow charging validation set is used for verification every 20 iterations to prevent the model from overfitting. Similarly, GRU is used to train and verify the fast charging training set.

[0106] It should be noted that the Q opt_normAs the input of the slow charge aging model, the estimated value of the battery health status under the slow charge mode is obtained, and the error between the model estimate and the measured battery health status in the slow charge test set is calculated, and the root mean square error (RMSE), mean absolute error (MAE) and maximum absolute error (MAX) are further calculated; similarly, the fast charge test set is used to estimate the battery health status under the fast charge mode, and the performance of the fast charge aging model is evaluated using the three indicators of RMSE, MAE and MAX. Figure 6a As shown in the figure, the SOH estimation result under slow charging mode, the RMSE is 0.67%, the MAE is 0.74%, and the MAX is 2.32%; Figure 6b As shown, the SOH estimation results under fast charging mode, the RMSE is 1.27%, the MAE is 1.28%, and the MAX is 2.57%.

[0107] It should be noted that the trained slow-charging aging model and the trained fast-charging aging model are both stored in the cloud server.

[0108] Step S2: normalizing the current charging capacity of the battery in the optimal charging SOC interval to obtain the normalized current charging capacity of the battery in the optimal charging SOC interval;

[0109] Step S3: inputting the normalized current charging capacity of the battery in the optimal charging SOC interval into the trained charging aging model to output an estimated value of the current health state of the battery.

[0110] Specifically, Figure 2 As shown, it also includes:

[0111] Determine whether the vehicle is in slow charging mode or fast charging mode each time it is charged;

[0112] If the vehicle is charged in slow charging mode, the current slow charging data of the battery is obtained; it is determined whether the vehicle passes through the optimal charging SOC interval during the slow charging process; if not, the charging capacity is not calculated; when the vehicle passes through the optimal charging SOC interval during the slow charging process, the current slow charging capacity of the battery in the optimal charging SOC interval is calculated according to the current slow charging data of the battery; then the current slow charging capacity of the battery in the optimal charging SOC interval is normalized to obtain the normalized current slow charging capacity of the battery in the optimal charging SOC interval; the normalized current slow charging capacity of the battery in the optimal charging SOC interval is input into the trained slow charging aging model to output the first battery current health state estimation value SOH_S;

[0113] If the vehicle is in fast charging mode, the current fast charging data of the battery is obtained; it is determined whether the vehicle passes through the optimal charging SOC interval during the fast charging process; if not, the charging capacity is not calculated; when the vehicle passes through the optimal charging SOC interval during the fast charging process, the current fast charging capacity of the battery in the optimal charging SOC interval is calculated according to the current fast charging data of the battery; then the current fast charging capacity of the battery in the optimal charging SOC interval is normalized to obtain the normalized current fast charging capacity of the battery in the optimal charging SOC interval; the current fast charging capacity of the battery in the normalized optimal charging SOC interval is input into the trained fast charging aging model to output a second battery current health state estimation value SOH_F;

[0114] After each charging and uploading of battery data, the cloud server will call the stored slow charge aging model or fast charge aging model to obtain an estimated value of the battery's current health state. Arrange the first battery's current health state estimate SOH_S or the second battery's current health state estimate SOH_F output each time in order from far to near in time to obtain the arranged battery's current health state estimate SOH; assuming that the 1st, k-1st, kth, and k+1st times are fast charge, slow charge, fast charge, and slow charge, respectively, then the arranged battery's current health state estimate SOH=[SOH_F (1) ,…,SOH_S (k-1) ,SOH_F (k) ,SOH_S (k+1) ,…], where the superscript is used to indicate the order of data arrangement;

[0115] Because the battery is a complex electrochemical system, in actual applications, the capacity of each charge may fluctuate up and down, resulting in fluctuations in the estimated SOH value. Using a sliding average filter method with a window length of 20 to filter the estimated SOH value of the arranged battery current health state, smooth filtering can be achieved to obtain the filtered estimated value of the battery current health state, so that the SOH finally output by the cloud server changes smoothly, while avoiding obvious fluctuations in the estimated SOH value caused by switching of the charging mode, such as Figure 7 Finally, the SOH output by the cloud server is transmitted back to the vehicle-side BMS to achieve vehicle-cloud collaborative SOH estimation, thereby using cloud server resources to achieve accurate estimation of complex SOH, which can reduce the resource requirements for the vehicle-side BMS chip.

[0116] It should be noted that if the charging does not pass through the defined optimal charging SOC interval, the SOH estimation will not be performed for this charging. Because battery aging is a long-term process, the loss of a certain SOH estimation value will not have a significant impact on the overall SOH estimation.

[0117] The battery SOH vehicle-cloud collaborative estimation method adapted to multiple charging modes provided by the present invention: (1) the health status of the lithium battery can be estimated in both slow charging and fast charging modes, and it has good applicability and practicality; (2) the battery health status is estimated based on the charging capacity of the optimal charging SOC range, which does not rely on complete charging data and has good applicability in practical applications; (3) the battery health status is estimated by relying on cloud server resources, which solves the problem of limited BMS resources on the vehicle side, and the vehicle-cloud collaboration realizes more accurate SOH estimation.

[0118] As another embodiment of the present invention, Figure 3 As shown, a battery SOH vehicle-cloud collaborative estimation system adapting to multiple charging modes is provided, wherein the battery SOH vehicle-cloud collaborative estimation system adapting to multiple charging modes includes a vehicle, a vehicle-side BMS and a cloud server, wherein the vehicle includes a lithium battery, which can be charged and discharged to provide power for the vehicle;

[0119] The vehicle-side BMS is used to obtain the current charging data of the battery, and calculate the current charging capacity of the battery in the optimal charging SOC interval according to the current charging data of the battery; then normalize the current charging capacity of the battery in the optimal charging SOC interval to obtain the normalized current charging capacity of the battery in the optimal charging SOC interval;

[0120] The cloud server is used to input the current charging capacity of the battery in the normalized optimal charging SOC interval into the trained charging aging model to output an estimated value of the current health state of the battery.

[0121] Preferably, the vehicle-side BMS includes a slave controller and a main controller, the slave controller collects voltage data and temperature data of the battery through the front-end sampling chip AFE, and transmits the collected voltage data and temperature data to the main controller through CAN communication or daisy chain communication; the main controller collects current data of the battery through a current sensor, and calculates the current charging capacity of the battery in the optimal charging SOC interval through the MCU and normalizes it, and then uploads the charging mode and the normalized current charging capacity of the battery in the optimal charging SOC interval to the cloud server through the 4G communication module carried by the vehicle-side BMS.

[0122] Preferably, the cloud server calls the trained charging aging model stored in itself according to the received charging mode and the current charging capacity of the battery in the normalized optimal charging SOC range to output an estimated value of the current health state SOH of the battery.

[0123] Preferably, the cloud server will store the battery's current state of health SOH estimate outputted during each charging, and obtain a filtered and stable battery's current state of health SOH estimate through data fusion and filtering algorithms, and then transmit the stable battery's current state of health SOH estimate back to the vehicle-side BMS through the 4G communication module, thereby realizing vehicle-cloud collaborative battery health state estimation.

[0124] The battery SOH vehicle-cloud collaborative estimation system adapted to multiple charging modes provided by the present invention: (1) estimates the health status of lithium batteries in slow charging and fast charging modes respectively, and realizes the estimation of the health status of batteries compatible with slow charging and fast charging modes through data fusion and filtering; (2) extracts features from partial charging data without the need for complete full charging information, thereby improving the adaptability of the proposed battery health status estimation method in practical applications; (3) vehicle-cloud collaboration, uploads the battery charging data of the vehicle-side BMS to the cloud server through the 4G communication module, uses the cloud server resources to realize the health status estimation, and then transmits it back to the vehicle-side BMS through the 4G communication module, solving the problem of limited vehicle-side BMS resources.

[0125] It is to be understood that the above embodiments are merely exemplary embodiments used to illustrate the principles of the present invention, but the present invention is not limited thereto. For those of ordinary skill in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A battery SOH vehicle-cloud collaborative estimation method that adapts to multiple charging modes, characterized in that: The battery SOH vehicle-cloud collaborative estimation method adapted to multiple charging modes includes: Step S1: obtaining current charging data of the battery, and calculating the current charging capacity of the battery in the optimal charging SOC interval according to the current charging data of the battery; Step S2: normalizing the current charging capacity of the battery in the optimal charging SOC interval to obtain the normalized current charging capacity of the battery in the optimal charging SOC interval; Step S3: inputting the normalized current charging capacity of the battery in the optimal charging SOC interval into the trained charging aging model to output an estimated value of the current health state of the battery.

2. The battery SOH vehicle-cloud collaborative estimation method adapted to multiple charging modes according to claim 1 is characterized in that: Before step S1, the method further includes: Obtaining battery historical aging data; wherein the battery historical aging data includes battery historical charging data and its corresponding battery historical discharging data, and the battery historical charging data is divided into battery historical slow charging data and battery historical fast charging data according to the charging mode; Calculate the battery's historical slow-charge capacity in the optimal charging SOC interval based on the battery's historical slow-charge data, and perform normalization processing; and calculate the first battery's historical health state based on the first battery's historical discharge data corresponding to the battery's historical slow-charge data; The historical fast-charge capacity of the battery in the optimal charging SOC interval is calculated according to the historical fast-charge data of the battery, and normalized; meanwhile, the historical health state of the second battery is calculated according to the historical discharge data of the second battery corresponding to the historical fast-charge data of the battery; Analyze the correlation between the historical slow charging capacity of the battery in the normalized optimal charging SOC interval and the historical health state of the first battery, and construct a slow charging data set with the historical slow charging capacity of the battery in the normalized optimal charging SOC interval and the historical health state of the first battery; at the same time, analyze the correlation between the historical fast charging capacity of the battery in the normalized optimal charging SOC interval and the historical health state of the second battery, and construct a fast charging data set with the historical fast charging capacity of the battery in the normalized optimal charging SOC interval and the historical health state of the second battery; The slow charging aging model is trained according to the slow charging data set to obtain a trained slow charging aging model; meanwhile, the fast charging aging model is trained according to the fast charging data set to obtain a trained fast charging aging model; wherein the trained slow charging aging model and the trained fast charging aging model are both stored in a cloud server.

3. The battery SOH vehicle-cloud collaborative estimation method adapted to multiple charging modes according to claim 2 is characterized in that: The slow charging aging model is trained according to the slow charging data set to obtain a trained slow charging aging model; At the same time, the fast charge aging model is trained according to the fast charge data set to obtain the trained fast charge aging model, which also includes: The slow charging data set is divided into a slow charging training set, a slow charging verification set and a slow charging test set; and the fast charging data set is divided into a fast charging training set, a fast charging verification set and a fast charging test set; The slow charging aging model is trained using the slow charging training set. During the training process, the slow charging validation set is used to validate the slow charging aging model to prevent the slow charging aging model from overfitting. After the training is completed, the slow charging test set is used to test and evaluate the performance of the trained slow charging aging model. The fast charging training set is used to train the fast charging aging model. During the training process, the fast charging verification set is used to verify the fast charging aging model to prevent the fast charging aging model from overfitting. After the training is completed, the fast charging test set is used to test and evaluate the performance of the trained fast charging aging model.

4. The battery SOH vehicle-cloud collaborative estimation method adapted to multiple charging modes according to claim 2 is characterized in that: Also includes: Determine whether the vehicle is in slow charging mode or fast charging mode each time it is charged; If the vehicle is charged in slow charging mode, obtain the current slow charging data of the battery; When the vehicle passes through the optimal charging SOC interval during the slow charging process, the current slow charging capacity of the battery in the optimal charging SOC interval is calculated according to the current slow charging data of the battery; then the current slow charging capacity of the battery in the optimal charging SOC interval is normalized to obtain the normalized current slow charging capacity of the battery in the optimal charging SOC interval; the normalized current slow charging capacity of the battery in the optimal charging SOC interval is input into the trained slow charging aging model to output a first battery current health state estimation value; If the vehicle is in fast charging mode, obtain the current fast charging data of the battery; When the vehicle passes through the optimal charging SOC interval during fast charging, the current fast charging capacity of the battery in the optimal charging SOC interval is calculated according to the current fast charging data of the battery; then the current fast charging capacity of the battery in the optimal charging SOC interval is normalized to obtain the normalized current fast charging capacity of the battery in the optimal charging SOC interval; Inputting the current fast-charge capacity of the battery in the normalized optimal charging SOC interval into the trained fast-charge aging model to output a current health state estimation value of the second battery; Arrange the first battery current health state estimation value or the second battery current health state estimation value output each time in chronological order to obtain an arranged battery current health state estimation value; The arranged battery current health state estimation value is filtered by a sliding average filtering method to obtain a filtered battery current health state estimation value.

5. The battery SOH vehicle-cloud collaborative estimation method adapted to multiple charging modes according to claim 2 is characterized in that: Also includes: Extracting battery historical charging data from the battery historical aging data every N cycles, wherein the battery historical charging data includes charging time, charging current and charging voltage; assuming that the battery is charged L times in total during its complete life cycle, and the battery historical charging data is extracted every 10 cycles, then a total of M times of battery historical charging data is extracted, that is, there are a total of M groups of battery historical charging data, M=L / 10; Preprocessing the battery historical charging data extracted each time to obtain a plurality of preprocessed battery historical charging data; The battery charging capacity Q of each historical charging process is calculated according to the plurality of pre-processed battery historical charging data. fullchrg and battery charge SOC, where the battery charge capacity Q fullchrg And the calculation formula of battery charging SOC is as follows: Where I is the charging current, t0 is the time when charging starts, t1 is the time when charging ends, SOC0 is the initial state of charge of the battery, and Qmax is the maximum available capacity of the battery; Determine the optimal charging SOC interval, and calculate the battery historical charging capacity Q of the optimal charging SOC interval according to the battery charging SOC in each historical charging process opt ; Among them, the optimal charging SOC interval is set to [50%, 80%], then the battery historical charging capacity Q in this interval opt The calculation formula is as follows: Among them, t_SOC0 is the charging time when the battery charging SOC is 50%, and t_SOC1 is the charging time when the battery charging SOC is 80%; The Min-Max normalization method is used to calculate the battery historical charging capacity Q in the optimal charging SOC interval. opt Normalization is performed to obtain the battery historical charging capacity Q of the normalized optimal charging SOC interval opt_norm ; The normalization formula is as follows: Q opt_norm =(Q opt -min(Q opt )) / (max(Q opt )-min(Q opt )) Among them, min(Q opt ) represents the historical charging capacity Q of group M batteries opt The minimum value in max(Q opt Represents the historical charging capacity Q of group M batteries opt The maximum value in .

6. The battery SOH vehicle-cloud collaborative estimation method adapted to multiple charging modes according to claim 5 is characterized in that: Also includes: Extracting battery historical discharge data every N cycles from the battery historical aging data, wherein the extracted battery historical discharge data and the extracted battery historical charging data are discharge data and charging data of the same aging cycle, respectively; The battery historical health status SOH during each historical discharge process is calculated based on the battery historical discharge data extracted each time. The calculation formula is as follows: Among them, Qnow is the current maximum available capacity of the battery, I is the discharge current, t0_dis and t1_dis are the start time and end time of discharge respectively, and Qnew is the rated capacity of the battery when it leaves the factory.

7. A battery SOH vehicle-cloud collaborative estimation system adapted to multiple charging modes, used to implement the battery SOH vehicle-cloud collaborative estimation method adapted to multiple charging modes as described in any one of claims 1-6, characterized in that: The battery SOH vehicle-cloud collaborative estimation system adapted to multiple charging modes includes a vehicle, a vehicle-side BMS and a cloud server; wherein the vehicle includes a battery, which can be charged and discharged to provide power for the vehicle; The vehicle-side BMS is used to obtain the current charging data of the battery, and calculate the current charging capacity of the battery in the optimal charging SOC interval according to the current charging data of the battery; then normalize the current charging capacity of the battery in the optimal charging SOC interval to obtain the normalized current charging capacity of the battery in the optimal charging SOC interval; The cloud server is used to input the current charging capacity of the battery in the normalized optimal charging SOC interval into the trained charging aging model to output an estimated value of the current health state of the battery.

8. The battery SOH vehicle-cloud collaborative estimation system adapted to multiple charging modes according to claim 7 is characterized in that: The vehicle-side BMS includes a slave controller and a main controller. The slave controller collects the voltage data and temperature data of the battery through the front-end sampling chip AFE, and transmits the collected voltage data and temperature data to the main controller through CAN communication or daisy chain communication; the main controller collects the current data of the battery through the current sensor, and calculates the current charging capacity of the battery in the optimal charging SOC range through the MCU and normalizes it, and then uploads the normalized current charging capacity of the battery in the optimal charging SOC range to the cloud server through the 4G communication module carried by the vehicle-side BMS.

9. The battery SOH vehicle-cloud collaborative estimation system adapted to multiple charging modes according to claim 8 is characterized in that: The cloud server calls the trained charging aging model stored in itself according to the received normalized current charging capacity of the battery in the optimal charging SOC interval to output an estimated value of the current health state SOH of the battery.

10. The battery SOH vehicle-cloud collaborative estimation system adapted to multiple charging modes according to claim 9 is characterized in that: The cloud server will store the battery's current state of health SOH estimate outputted during each charge, and obtain a filtered and stable battery's current state of health SOH estimate through data fusion and filtering algorithms, and then transmit the stable battery's current state of health SOH estimate back to the vehicle-side BMS through the 4G communication module.

Images