Method and device for predicting service life of battery, equipment and storage medium

Abstract

The invention relates to the field of batteries, and provides a battery service life prediction method and device, equipment and a storage medium. A training method of a prediction model used in the method comprises the following steps: acquiring a parameter feature set in sample data related to a to-be-predicted battery; determining the correlation between the parameter characteristics in the parameter characteristic set and the service life of the battery; according to the correlation, determining related parameter features used for predicting the service life of the battery in the parameter feature set; and training a prediction model of the battery service life of the to-be-predicted battery according to the related parameter characteristics in the sample data. As the prediction model adopts the parameter characteristics with relatively high correlation with the service life of the battery, the influence of the parameter characteristics with relatively low correlation on the prediction of the service life of the battery is reduced, the probability that the parameter characteristics with relatively high correlation are not introduced during prediction is reduced, and the prediction efficiency is improved while the training or prediction calculation complexity is reduced. And the prediction precision of the prediction model can be effectively improved.

Description

Battery life prediction method, device, equipment and storage medium Technical Field

[0001] The present application relates to the field of batteries, and in particular to a method, apparatus, device and storage medium for predicting battery service life. Background Art

[0002] With the development of battery technology, batteries have gained advantages such as high energy density, high open-circuit voltage, high output power, no memory effect, low self-discharge, a wide operating temperature range, and fast charge and discharge speeds. They are widely used in industries such as agriculture, communications, transportation, and national defense. During battery use, to prevent accidents, batteries must be replaced before they age. However, blindly replacing batteries prematurely inevitably results in significant waste. Therefore, accurately predicting the remaining battery life is of great significance.

[0003] Battery life prediction methods typically use data models to predict battery life. However, the lack of clarity about the impact of these parameters on battery life hinders accurate prediction when selecting these parameters.

[0004] Summary of the Invention

[0005] In view of this, the embodiments of the present application provide a training method, device, equipment and storage medium for a battery service life prediction model to solve the problem that in some cases, when predicting battery service life, it is not conducive to improving the accuracy of the prediction results of battery service life.

[0006] A first aspect of an embodiment of the present application provides a method for training a prediction model for battery service life, the method comprising: obtaining a parameter feature set from sample data related to a battery to be predicted; determining the correlation between the parameter features in the parameter feature set and the battery service life; based on the correlation, determining the relevant parameter features in the parameter feature set for predicting the battery service life; and training a prediction model for the battery service life of the battery to be predicted based on the relevant parameter features in the sample data.

[0007] When predicting the battery life of a battery to be predicted, the embodiment of the present application first determines sample data related to the battery to be predicted, that is, battery data with the battery life and various parameter characteristics of the battery pre-calibrated. According to the calibrated sample data, the correlation between the parameter characteristics and the battery life is calculated. Based on the correlation between the parameter characteristics and the battery life, the relevant parameter characteristics for predicting the battery life are determined. The battery life is predicted based on the relevant parameter characteristics, thereby determining parameter characteristics with a high correlation with the battery life, reducing the influence of parameter characteristics with a low correlation on the battery life prediction, and reducing the probability of not introducing parameter characteristics with a high correlation during prediction. While reducing the training complexity of the prediction model, the accuracy of the prediction model can be effectively improved.

[0008] In combination with the first aspect, in a first possible implementation of the first aspect, determining the correlation between the parameter features in the parameter feature set and the battery life includes: determining the mutual information value between each parameter feature in the parameter feature subset and the battery life, the parameter feature subset includes a predetermined number of parameter features, and the parameter features belong to the parameter feature set; determining a first average value of the mutual information values ​​of a predetermined number of the parameter features and the battery life based on the mutual information value, and determining the correlation between the parameter features and the battery life based on the first average value.

[0009] By separately determining the mutual information value between each parameter feature and the battery life, and then based on the mutual information values ​​of each parameter feature, determining a first average value of the mutual information values ​​between a parameter feature subset including a predetermined number of parameter features and the battery life, the parameter feature subset with the highest correlation with the battery life is determined based on the first average value. For example, the parameter feature subset corresponding to the maximum first average value is selected as the parameter feature subset for the battery to be predicted. By calculating the correlation, the correlation of the selected parameter feature subset can be made higher, which is conducive to improving the accuracy of the selected feature parameters.

[0010] In combination with the first possible implementation of the first aspect, in the second possible implementation of the first aspect, before determining the mutual information value between each parameter feature in the parameter feature subset and the battery life, the method also includes: calculating the battery life predicted by the parameter feature subset including different numbers of parameter features; determining the difference between the predicted battery life and the actual battery life of the sample data, and obtaining the number of parameter features corresponding to the minimum difference as the predetermined number.

[0011] For different battery types, the number of parameter features included in a parameter feature subset may also vary. Therefore, before determining the most relevant parameter subset, the sample data can be calculated to determine that the parameter subset includes different numbers of parameter features. Based on the difference between the calculated battery life and the battery life calibrated in the sample data, the number of parameter features included in the parameter subset that minimizes the difference is selected as the predetermined number. Screening parameter features using this predetermined number is conducive to improving the accuracy of battery life prediction.

[0012] In combination with the first possible implementation manner of the first aspect, in the third possible implementation manner of the first aspect, before determining the relevant parameter features in the parameter feature set for predicting the battery service life based on the correlation, the method also includes: calculating the second average value of the mutual information values ​​between the parameter features in the parameter feature subset; determining the relevant parameter features in the parameter feature set for predicting the battery service life based on the correlation, including: determining the relevant parameter features in the parameter feature set for predicting the battery service life based on the maximum value of the difference between the first average value and the second average value.

[0013] After determining the correlation between the battery parameter characteristics and the battery service life, the correlation between the parameter features in the battery parameter feature set is further obtained, and the redundancy between the parameter features is determined based on the correlation between the parameter features, including deleting the parameter features with large correlation between the parameter features, so that the correlation between the parameter features after de-redundancy is smaller, which is beneficial to reducing the repeated impact of the parameter features in the parameter feature subset on the prediction results, and is beneficial to improving the accuracy of the prediction results while improving the efficiency of the prediction.

[0014] In combination with the first aspect, in a fourth possible implementation of the first aspect, training the prediction model of the battery service life of the battery to be predicted based on the relevant parameter features in the sample data includes: inputting the relevant parameter features in the sample data into a preset battery service life prediction model for calculation, outputting the predicted battery service life, and adjusting the parameters of the prediction model based on the difference between the predicted battery service life and the calibrated battery service life in the sample data until the difference between the two meets the predetermined requirements, thereby completing the training of the prediction model.

[0015] After determining a subset of parameter features of the battery to be predicted, the embodiment of the present application selects values ​​of the relevant parameter features in the sample data based on the determined relevant parameter features, and trains a battery life prediction model according to the values ​​of the relevant parameter features, that is, inputs the relevant parameter features of the sample data into the battery life prediction model, and adjusts the parameters in the battery life prediction model based on the difference between the battery life output by the battery life prediction model and the battery life calibrated in the sample data until they are adjusted to meet the preset requirements, thereby being able to effectively predict the battery life of the battery to be predicted.

[0016] In combination with the fourth possible implementation manner of the first aspect, in a fifth possible implementation manner of the first aspect, the prediction model for battery service life is a support vector machine model.

[0017] When predicting battery life, the embodiments of the present application are not limited to support vector machine models and may also include other network models, such as different neural network models. Compared to neural network models, support vector machine models have the advantage of processing nonlinear data and can more accurately predict battery life.

[0018] In combination with the fifth possible implementation method of the first aspect, in the sixth possible implementation method of the first aspect, a preset battery life prediction model is calculated using the relevant parameter characteristics, and a predicted battery life is output. The parameters of the battery life prediction model are adjusted according to the difference between the predicted battery life and the battery life calibrated in the sample data, including: determining the initial value of the parameter, the parameter value range and the parameter adjustment step in the support vector machine model; inputting the relevant parameter characteristics in the sample data into the support vector machine model, gradually adjusting the parameters in the support vector machine model according to the adjustment step, and calculating and outputting multiple battery lifespans; comparing the output multiple battery lifespans with the battery lifespan calibrated in the sample data respectively, and taking the parameter value when the output battery life is closest to the battery life calibrated in the sample data as the parameter value of the support vector machine model.

[0019] The parameter features in the embodiments of this application are different from the parameters of the support vector machine model. The parameter features refer to the characteristics of the battery. The parameters are the calculation parameters used in the support vector machine model to predict the battery life.

[0020] When training a support vector machine model, the initial parameter values, parameter value ranges, and parameter adjustment steps in the support vector machine model can be predetermined. The initial parameter values ​​are the values ​​determined during initialization of the support vector machine model, and the parameter adjustment step size is the magnitude of the parameter change each time the parameter is adjusted. The parameter value range refers to the range of parameter adjustment when the parameter is adjusted. In general, the first endpoint of the parameter value range can be selected as the initial value until it is adjusted to the second endpoint of the parameter value range. The battery life in the sample data is calculated after gradually changing the parameters, and the optimal parameters are determined based on the difference between the calculated battery life and the calibrated battery life, which serve as the calculation parameters of the support vector machine model.

[0021] In combination with the sixth possible implementation method of the first aspect, in the seventh possible implementation method of the first aspect, the parameters of the support vector machine model include one or more of the penalty coefficient of the support vector machine model, the gamma parameter of the radial basis function parameter as the kernel function, the adjustment step size of the penalty coefficient and the adjustment step size of the gamma parameter.

[0022] Based on the determined value range of the gamma parameter, an initial value of the gamma parameter can be determined. Based on the value range of the penalty coefficient, an initial value of the penalty coefficient can be determined. The penalty coefficient is gradually adjusted based on the adjustment step size of the penalty coefficient and the initial value of the penalty coefficient. The gamma parameter is gradually adjusted based on the adjustment step size and the initial value of the gamma parameter until all value combinations are adjusted, thereby obtaining battery service life predicted by the support vector machine model for different gamma parameters and different penalty coefficients. The parameter value that produces the most accurate prediction result is selected as the parameter value of the support vector machine model.

[0023] In combination with the seventh possible implementation method of the first aspect, in the eighth possible implementation method of the first aspect, determining the parameter value range in the support vector machine model includes: determining multiple battery lifespans output by the support vector machine model based on a first parameter value range of the support vector machine model and a first adjustment step corresponding to the first parameter value range; determining a second parameter value range and a second adjustment step corresponding to the second parameter value range based on the multiple battery lifespans output by the support vector machine model, wherein the second parameter value range belongs to the first parameter value range and is smaller than the first parameter value range, and the second adjustment step is smaller than the first adjustment step.

[0024] In order to further improve the adjustment efficiency, based on the first parameter value range, a first adjustment step with a larger step size can be first used for coarse adjustment. When a better prediction result is determined, the second value range of the corresponding parameter value is determined, and then a second adjustment step with a finer step size than the first adjustment step size is used for adjustment, so that more accurate parameter values ​​can be obtained, which is beneficial to improving the accuracy of the prediction results of battery service life.

[0025] In combination with the fourth to eighth possible implementation methods of the first aspect, in the ninth possible implementation method of the first aspect, a prediction model for the battery service life of the battery to be predicted is trained based on the relevant parameter characteristics in the sample data, including: inputting the relevant parameter characteristics of the battery to be predicted into the trained prediction model for the battery service life, and outputting the predicted value of the battery service life of the battery to be predicted through calculation of the battery service life prediction model.

[0026] After determining the relevant parameter features required for the battery life prediction model using sample data and completing the parameter adjustment training of the battery life prediction model, the values ​​of the relevant parameter features of the battery to be predicted can be input into the trained battery life prediction model. The battery life prediction model then performs a prediction calculation based on the values ​​of the relevant parameter features to obtain a prediction result for the battery life of the battery to be predicted. Since the sample data and the battery to be predicted are of the same type, the battery life prediction model trained based on the sample data and the determined relevant parameter features can accurately obtain the battery life of the battery to be predicted.

[0027] In combination with the first aspect to the ninth possible implementation of the first aspect, in the tenth possible implementation of the first aspect, the parameter feature set includes one or more of the charge and discharge rate, number of charge and discharge cycles, battery state of charge, charge and discharge state, vehicle operating status, battery temperature, battery internal resistance, battery capacitance, and impedance data of different frequencies of a battery of the same type as the battery to be predicted.

[0028] The vehicle operating state includes one or more of a driving state, an idling state, and an engine-off state. By identifying the rich parameter features included in the parameter feature set, subsequent battery life prediction can more comprehensively and accurately determine the factors that affect battery life, thereby improving the accuracy of battery life prediction.

[0029] The second aspect of an embodiment of the present application provides a method for predicting battery service life, the method comprising: obtaining relevant parameter characteristics of a battery to be predicted, and inputting the relevant parameter characteristics into any training method of the first aspect to obtain a prediction model, and predicting the battery service life of the battery to be predicted.

[0030] In a possible implementation manner of the second aspect, obtaining relevant parameter characteristics of the battery to be predicted may include: obtaining relevant parameter characteristics of the battery to be predicted according to any training method of the first aspect.

[0031] A third aspect of an embodiment of the present application provides a training device for a battery service life prediction model, the device comprising: a parameter feature set acquisition unit, for acquiring a parameter feature set from sample data related to a battery to be predicted; a correlation determination unit, for determining the correlation between parameter features in the parameter feature set and the battery service life; a relevant parameter feature determination unit, for determining relevant parameter features in the parameter feature set for predicting the battery service life based on the correlation; and a training unit, for training the battery service life prediction model of the battery to be predicted based on the relevant parameter features in the sample data.

[0032] The fourth aspect of an embodiment of the present application provides a battery service life prediction device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method described in any one of the first aspects when executing the computer program.

[0033] A fifth aspect of an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method described in any one of the first aspects are implemented.

[0034] It can be understood that the beneficial effects of the second to fifth aspects mentioned above can be found in the relevant description of the first aspect mentioned above, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments or descriptions of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0036] FIG1 is a schematic diagram of an implementation flow of a method for training a battery life prediction model provided in an embodiment of the present application;

[0037] FIG2 is a schematic diagram of an implementation flow of determining a parameter feature subset with the greatest correlation provided by an embodiment of the present application;

[0038] FIG3 is a schematic diagram of an implementation flow of determining the number of parameter features of a parameter feature subset provided by an embodiment of the present application;

[0039] FIG4 is a schematic diagram of an implementation flow of a model training method provided in an embodiment of the present application;

[0040] FIG5 is a flow chart of another model training method provided in an embodiment of the present application;

[0041] FIG6 is a schematic diagram of a training device for a battery life prediction model provided by an embodiment of the present application;

[0042] FIG7 is a schematic diagram of a device for predicting battery service life provided in an embodiment of the present application. DETAILED DESCRIPTION

[0043] In the following description, specific details such as specific system structures and techniques are provided for purposes of illustration rather than limitation to facilitate a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid obscuring the description of the present application with unnecessary detail.

[0044] In order to illustrate the technical solution described in this application, specific embodiments are provided below.

[0045] To prevent battery accidents during use, batteries need to be replaced before they age. However, blindly replacing batteries prematurely will inevitably result in huge waste. Therefore, accurately predicting the remaining battery life is of great significance.

[0046] However, when predicting battery life, if the battery parameter characteristics are blindly input into the prediction model, due to the lack of understanding of the impact of different battery parameter characteristics on battery life, the parameter characteristics used for calculation may include parameter characteristics that have little impact on battery life or are irrelevant, which will affect the prediction accuracy of battery life and increase the complexity of the prediction. Alternatively, the parameter characteristics used for prediction may include parameter characteristics with repeated effects, which will increase the complexity of the prediction and may reduce the accuracy of the battery life prediction.

[0047] Based on this, the embodiment of the present application provides a method for training a prediction model for battery life. By screening the correlation between the battery parameter characteristics and the battery life, as well as the correlation between the battery parameter characteristics, the parameter characteristics are screened, and the parameter characteristics with high correlation and low redundancy between parameters are extracted. The prediction model is trained and calculated based on the screened relevant parameter characteristics, thereby reducing the complexity of the parameter feature calculation and improving the prediction accuracy of the battery life. As shown in Figure 1, the method includes:

[0048] In S101 , a parameter feature set in sample data related to a battery to be predicted is obtained.

[0049] Specifically, the battery to be predicted in the embodiment of the present application is a battery for which battery service life prediction is to be performed.

[0050] Before predicting the battery life of the battery to be predicted, the battery type of the battery to be predicted can be obtained. The battery type can be determined based on the battery materials. For example, the battery type can be determined based on the main materials such as the positive electrode, negative electrode, separator, and electrolyte. However, other battery materials may also be included in the determination of the battery type, such as the battery's conductive agent, additives, and other materials. Alternatively, in a possible implementation, the battery type to be predicted can also be determined based on one or more of the main materials.

[0051] After determining the type of battery to be predicted, sample data of the same type can be searched for based on the type, that is, sample data related to the battery to be predicted. The sample data is data that calibrates the parameter characteristics and battery service life of the battery.

[0052] The battery life can be the remaining life of the battery when the battery parameter feature set is acquired, that is, the duration of normal use of the battery after the relevant parameter feature set is acquired.

[0053] The parameter characteristics of the battery can be all the characteristics included in the parameter characteristic set of the battery, including one or more of the battery charge and discharge rate, number of charge and discharge cycles, battery state of charge, charge and discharge status, vehicle operating status, battery temperature, battery internal resistance, battery capacitance, and impedance data at different frequencies.

[0054] Among them, the vehicle operating state refers to the operating state of the vehicle in which the battery is installed and powered by the battery. The vehicle operating state may include the vehicle driving state, the vehicle idling state and the vehicle flameout state. Among them, the vehicle driving state refers to the vehicle's engine being in a working state and the vehicle's driving speed being greater than 0, that is, the vehicle is in a moving state. The vehicle idling state may include the state in which the vehicle's engine is in a working state and the vehicle's driving speed is 0. The vehicle flameout state may include the state in which the vehicle's engine is not working. Since the parameter characteristics of the battery will be affected when the vehicle is in different operating states, the parameter characteristic set includes the operating state of the vehicle in which the battery is located.

[0055] If the battery is used in non-vehicle equipment, the operating status of the equipment can be obtained from the determined parameter feature set. For example, if the battery is used in other industrial equipment, the different operating states of the equipment can also be determined based on the status of the motor in the industrial equipment.

[0056] When determining impedance data at different frequencies, the current signal i(n) and the voltage signal u(n) can be Fourier transformed to obtain Fi(f) and Fu(f), respectively. The impedance calculation formula Z(f) = Fu(f) / Fi(f) can be used to obtain the impedance values ​​corresponding to different frequency points. Here, n is the count value with an interval of Ts period.

[0057] In S102, the correlation between the parameter features in the parameter feature set and the battery life is determined.

[0058] Each sample data item is calibrated with the values ​​of each parameter feature in the parameter feature set and the battery life. Based on the differences between the different sample data items, including the differences in the parameter feature values ​​and the differences in the battery life, the impact of the battery parameter features on the battery life can be determined, that is, the correlation between the battery parameter features and the battery life can be determined. This correlation can be represented by the mutual information between the battery parameter features and the battery life.

[0059] The mutual information between the battery parameter characteristics and the battery life can be calculated using the following formula:

[0060] Where I(F,c) is the mutual information between the battery parameter characteristics and the battery life, p(fi) is the probability of the parameter characteristic fi appearing, p(ci) is the probability of the battery life ci appearing, and p(fi,ci) is the probability of the parameter characteristic fi and the battery life ci appearing at the same time.

[0061] In the embodiment of the present application, in addition to determining the correlation between a single parameter feature and battery life, a first average value of the correlation between a subset of parameter features and battery life can also be determined. Based on the magnitude of the first average value, the subset of parameter features with the greatest correlation with battery life is determined. The specific implementation process is shown in Figure 2 and includes:

[0062] In S201 , a mutual information value between each parameter feature in the parameter feature subset and the battery life is determined.

[0063] The parameter feature subset includes a predetermined number of parameter features, and the parameter features belong to the parameter feature set. That is, a predetermined number of parameter features can be selected from the parameter feature set to form a parameter feature subset. Because different combination methods can determine different parameter feature subsets, when the number of parameter features in the parameter feature subset is less than the number of parameter features in the parameter feature set, two or more parameter feature subsets can be obtained through grouping.

[0064] When determining the mutual information value between each parameter feature in the parameter feature subset and the battery life, the mutual information value between each parameter feature in the parameter feature set and the battery life can be determined in advance, and the corresponding mutual information value can be selected from the determined mutual information values ​​according to the parameter features included in the parameter feature subset.

[0065] The number of parameter features in the embodiment of the present application can determine the optimal number of parameter features included in the parameter feature subset based on the accuracy of the prediction results corresponding to the parameter feature subsets determined by different numbers of parameter features in the sample data, which can be specifically shown in Figure 3, including:

[0066] In S301 , the battery service life predicted by parameter feature subsets including different numbers of parameter features is calculated.

[0067] The parameter feature set can be reduced in number to obtain parameter feature subsets including different numbers of parameter features. The parameter feature subsets including different numbers of parameter features are combined with sample data to determine data for training a battery life prediction model.

[0068] For example, if the number of parameter features included in the parameter feature set is N, then N-1, N-2, N-3...2 and 1 can be selected as the number of parameter features included in the feature subset. Based on the number of parameter features included in the determined parameter feature subset, a corresponding number of parameter features are selected from the parameter features for combination to obtain a variety of different parameter feature subsets. Based on the obtained multiple parameter feature subsets, the value data of the parameter features in the parameter feature subset is obtained from the sample data, and the battery service life prediction model is trained based on the obtained value data. Or the training result meets the preset requirements, such as after reaching a predetermined number of iterative updates, the battery service life of the sample data is calculated or predicted based on the trained battery service life prediction model.

[0069] In S302, the difference between the predicted battery service life and the actual battery service life of the sample data is determined, and the number of parameter features corresponding to the minimum difference is obtained as the predetermined number.

[0070] After predicting the battery service life based on the parameter feature subsets determined by different numbers of parameter features, the difference between the predicted result and the calibration result is obtained, that is, the difference between the predicted battery service life and the calibrated battery service life.

[0071] According to different numbers of parameter features and different combinations of the same number of parameter features, by training the battery life prediction model and predicting the battery life based on the trained battery life prediction model, the number corresponding to the parameter feature subset with the highest prediction accuracy can be selected as the predetermined number.

[0072] For example, when dividing parameter feature subsets of different numbers, there are parameter feature subsets of n1 and parameter feature subsets of n2. The parameter feature subset of n1 includes x1 combinations, and the parameter feature subset of n2 includes x2 combinations. Since the parameter feature subset of n1 has x1 combinations, x1 prediction accuracies can be calculated. The parameter feature subset of n2 has x2 combinations, and x2 prediction accuracies can be calculated. Among all the prediction accuracies, the parameter feature subset with the highest accuracy can be selected, and according to the number of parameter features corresponding to the parameter feature subset, it can be determined as the predetermined number, that is, the optimal number of parameter features included in the parameter feature subset.

[0073] In S202, a first average value of mutual information values ​​of a predetermined number of the parameter features and the battery life is determined according to the mutual information value.

[0074] After determining the optimal number of parameter features included in the parameter feature subset, that is, the predetermined number of parameter features, the predetermined number of parameter features can be searched in the sample data, and multiple groups of parameter features can be obtained by combination to obtain multiple parameter feature subsets.

[0075] After calculating the mutual information value between the parameter feature fi in each parameter feature subset S and the battery service life c, multiple mutual information values ​​are averaged to obtain a first average value.

[0076] In S203 , the correlation between the parameter feature and the battery service life is determined according to the first average value.

[0077] The correlation between a predetermined number of parameter features in the parameter feature subset and the battery life is determined based on the first average value. The larger the first average value, the stronger the correlation.

[0078] For example, the first average value of the mutual information value between the parameter feature fi and the battery life c can be expressed as:

[0079] Among them, I(f i; c) is the mutual information value between the parameter feature fi and the battery life c, |S| is the number of parameter features included in the parameter feature subset, that is, the predetermined number.

[0080] In S103, the relevant parameter features in the parameter feature set for predicting the battery service life are determined based on the correlation.

[0081] In the embodiment of the present application, the parameter feature subset with the greatest correlation may be selected for battery service life prediction based on the parameter features in the parameter feature set, for example, based on the difference between the parameter feature subset and the battery service life.

[0082] In a possible implementation, in order to further improve the prediction accuracy of battery service life, the embodiment of the present application can further calculate the correlation between the parameter features in the parameter feature subset, and determine the redundant parameter features included in the parameter feature subset based on the correlation, so as to further simplify the calculation amount and improve the prediction accuracy of battery service life.

[0083] Among them, the calculation formula of the mutual information value between parameter features can be shown as follows:

[0084] Among them, I(fi,fj) is the mutual information between the parameter features fi and fj of the battery, p(fi) is the probability of the parameter feature fi appearing, p(fj) is the probability of the parameter feature fj appearing, and p(fi,fj) is the probability of the parameter features fi and parameter features fj appearing at the same time.

[0085] After calculating the mutual information values ​​between all parameter features in the parameter feature subset, the average value of the mutual information between all parameter features in the parameter feature subset, i.e., the second average value, can be calculated. The specific calculation formula can be as follows:

[0086] Wherein, is the second average value, |S| is the number of parameter features included in the parameter feature subset, that is, the predetermined number, and fi and fj are both parameter features in the parameter feature subset.

[0087] A second average value corresponding to the parameter feature subset is determined based on the mutual information between the parameter features. The redundancy of the parameter features in the parameter feature subset can be determined based on the second average value. The larger the second average value, the greater the redundancy.

[0088] In order to optimize the prediction calculation, the difference between the first average value and the second average value can be calculated, and the parameter feature subset corresponding to the maximum difference can be used as the parameter feature subset for the prediction calculation. The parameter feature type in the parameter feature subset is the relevant parameter feature used for battery service life prediction.

[0089] The parameter feature subset, i.e., the related parameter features, is determined by maximizing the difference between the first average value and the second average value, so that the parameter features used for predictive calculations have a greater correlation with the battery service life, while the redundancy between the parameter features is smaller, so that subsequent predictive calculations can effectively reduce the complexity of the calculations and improve the prediction accuracy of the battery service life.

[0090] In S104, a prediction model for the battery service life of the battery to be predicted is trained according to relevant parameter features in the sample data.

[0091] In the embodiments of the present application, the model used for battery life prediction may include a support vector machine model. However, this is not limited to a support vector machine model and may also include neural network models of various structures. When using a support vector machine model for battery life prediction, due to its advantage in processing nonlinear data, it can more accurately predict battery life.

[0092] When predicting the battery life of a battery to be predicted, the model can be trained using sample data in combination with the determined parameter features for prediction. After the training is completed, the battery life of the battery to be predicted is predicted based on the updated parameters.

[0093] During the training process, the corresponding values ​​can be extracted from the sample data based on the determined relevant parameter characteristics, and the values ​​of the relevant parameter characteristics can be input into the prediction model of the battery life to be trained. The prediction model of the battery life is used to calculate and output the predicted battery life. The output battery life is compared with the battery life pre-calibrated in the sample data to determine the size of the difference between the two. The parameters of the prediction model of the battery life are adjusted according to the difference until the difference meets the predetermined requirements, and the training of the model is completed. After the model training is completed, the values ​​of the relevant parameter characteristics of the battery to be predicted can be input into the trained prediction model of the battery life, and the predicted value of the battery life to be predicted can be obtained through calculation and prediction.

[0094] When the battery life prediction model is a support vector machine model, the process of training the model may be as shown in FIG4 , including:

[0095] In S401, the initial value of the parameter, the parameter value range and the parameter adjustment step size in the support vector machine model are determined.

[0096] The support vector model in the embodiment of the present application can be expressed by the following formula: st y i (ω T x i +b)≥1-ε iε i ≥0,i=1,2,…,m.

[0097] Among them, ω is the normal vector, which determines the direction of the hyperplane used for support vector machine segmentation; b is the displacement term; ε i is a slack variable; C is a constant; (xi,yi) is a training sample determined based on the relevant parameter features; m is the number of parameter features in the relevant parameter features.

[0098] In an embodiment of the present application, the parameters of the support vector machine model may include one or more of the penalty coefficient c of the support vector machine model, the gamma parameter g of the radial basis function parameter as the kernel function, the adjustment step size of the penalty coefficient and the adjustment step size of the gamma parameter.

[0099] The penalty coefficient c is the tolerance for misjudgment. The larger the c, the less tolerant the model is to errors, which makes it prone to overfitting. The smaller the c, the more prone to underfitting.

[0100] The gamma parameter g is a parameter added by the radial basis function after selecting it as the kernel function. It implicitly determines the distribution of the data after mapping it to the new feature space. A larger g value indicates fewer support vectors, while a smaller g value indicates more support vectors. The number of support vectors affects the speed of training and prediction.

[0101] Among them, the gamma parameter g and the penalty coefficient c can each determine a value range during initialization. For example, the SVMcgForRegress function can be selected, and in the set value range, the minimum penalty coefficient cmin is set to any value between -6 and -10, such as -6, -8, or -10, the maximum penalty coefficient cmax is set to any value between 6 and 10, such as 6, 8, or 10, the minimum gamma parameter gmin is set to any value between -6 and -10, such as -6, -8, or -10, and the maximum gamma parameter gmax is set to any value between 6 and 10, such as 6, 8, or 10. According to the determined value range, any endpoint thereof can be taken as the initialization value, and according to the predetermined adjustment step size, for example, the adjustment step size cstep of the penalty coefficient c is set to any value between 0.3 and 0.8, such as 0.3, 0.5, or 0.8, and the adjustment step size gstep of the gamma parameter g is set to any value between 0.3 and 0.8, such as 0.3, 0.5, or 0.8. In a possible implementation, the misjudgment step size msestep can also be set to any value between 0.005 and 0.05, such as 0.008, 0.01, 0.02, 0.03, 0.04, or 0.05.

[0102] In a possible implementation, the parameters of the support vector machine model may further include a cross-validation fold parameter v, which indicates that v fold cross-validation is performed on the training set. For example, the cross-validation fold parameter v may be set to 4, which means that 4 fold cross-validation is performed on the training set.

[0103] In S402, relevant parameter features in the sample data are input into the support vector machine model, parameters in the support vector machine model are gradually adjusted according to the adjustment step size, and multiple battery service lives are calculated and output.

[0104] After initializing the support vector machine model and determining the adjustment strategy (i.e., the adjustment step size), the corresponding values ​​of the relevant parameter features can be searched in the sample data based on the relevant parameter features. The found values ​​of the relevant parameter features are input into the initialized support vector machine model, which then predicts and calculates the relevant parameter features and outputs a predicted battery life. By gradually adjusting the parameter values ​​according to the adjustment step size of the adjustment strategy, multiple battery lifespans corresponding to the parameter values ​​are predicted.

[0105] In a possible implementation, the embodiment of the present application can also adopt different precision level adjustment methods when training the model to complete the training of the model. Specifically, as shown in Figure 5, it includes:

[0106] In S501, a plurality of battery service lives output by the support vector machine model are determined according to a first parameter value range of the support vector machine model and a first adjustment step corresponding to the first parameter value range.

[0107] When performing preliminary adjustment, a larger first parameter value range may be used, and within the first parameter value range, a larger first adjustment step may be used for adjustment.

[0108] The first parameter value range may include the value range of the penalty coefficient c and / or the value range of the gamma parameter g. Accordingly, the first adjustment step may include the adjustment step of the penalty coefficient c and / or the adjustment step of the gamma parameter g.

[0109] For example, in the first parameter value range, the value range of c can be [-8, 8], and the value range of g can be [-8, 8]. In the first adjustment step, the adjustment step of g can be 0.5, and the adjustment step of c can be 0.5.

[0110] In S502, a second parameter value range and a second adjustment step corresponding to the second parameter value range are determined based on multiple battery service lives output by the support vector machine model, wherein the second parameter value range belongs to the first parameter value range and is smaller than the first parameter value range, and the second adjustment step is smaller than the first adjustment step.

[0111] After performing a prediction calculation based on the first adjustment step and the first value range, a parameter value range corresponding to a greater prediction accuracy can be obtained, i.e., a second parameter value range. Based on the second parameter value range, a more refined second adjustment step can be used to perform detailed adjustments to obtain a more accurate parameter value.

[0112] For example, in the second parameter value range, the value range of c can be [-3, 3], and the value range of g can be [-3, 3]. In the first adjustment step, the adjustment step of g can be 0.1, and the adjustment step of c can be 0.1.

[0113] In S403, the output multiple battery service lives are compared with the battery service lives calibrated in the sample data respectively, and the parameter values ​​when the output battery service life is closest to the battery service life calibrated in the sample data are used as the parameter values ​​of the support vector machine model.

[0114] By adjusting the parameters, the sample data is predicted and calculated to obtain the difference between the predicted battery life and the calibrated battery life, or to obtain the statistical average of the differences between the battery life predicted by different sample data and the calibrated battery life.

[0115] Based on the obtained differences, the parameters corresponding to the minimum differences can be selected as the parameter values ​​of the support vector machine model, that is, the parameter training of the support vector machine model is completed. Based on the trained support vector machine model, the values ​​of the relevant parameter characteristics of the battery to be predicted can be obtained, input into the trained support vector machine model, and the predicted battery life is output. Since the parameter characteristics used in the support vector model in this application during training and prediction are parameter characteristics after correlation screening, training the prediction model based on the relevant parameter characteristics after screening can effectively improve the training efficiency of the prediction model and improve the accuracy of the prediction model.

[0116] In addition, an embodiment of the present application also provides a method for predicting battery life. This method can determine the relevant parameter characteristics of the battery to be predicted based on the training method of the above-mentioned battery life prediction model, input the determined relevant parameter characteristics into the trained prediction model, and predict the battery life of the battery to be predicted. Alternatively, the parameter characteristics of the battery can be directly obtained, and the obtained parameter characteristics of the battery can be input into the trained prediction model to predict the battery life of the battery to be predicted. Since the prediction model only needs to use the relevant parameter characteristics for prediction calculations, it is beneficial to improve the prediction accuracy of the battery life while reducing the calculation amount of the prediction model.

[0117] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0118] FIG6 is a schematic diagram of a training device for a battery life prediction model provided by an embodiment of the present application. As shown in FIG6 , the device includes:

[0119] The parameter feature set acquisition unit 601 is configured to acquire a parameter feature set from sample data related to the battery to be predicted.

[0120] The correlation determination unit 602 is configured to determine the correlation between the parameter features in the parameter feature set and the battery life.

[0121] The relevant parameter feature determination unit 603 is configured to determine, based on the correlation, relevant parameter features in the parameter feature set for predicting the battery service life.

[0122] The training unit 604 is configured to train a prediction model for the battery service life of the battery to be predicted based on relevant parameter features in the sample data.

[0123] The training device for the battery life prediction model shown in FIG6 corresponds to the training method for the battery life prediction model shown in FIG1 .

[0124] In addition, an embodiment of the present application further provides a device for predicting battery life, the device comprising:

[0125] The relevant parameter feature acquisition module is used to obtain the relevant parameter features of the battery to be predicted based on the relevant parameter features determined by the training method of the battery life prediction model shown in Figure 1, and input the relevant parameter features into the prediction model trained by the training method shown in Figure 1 to calculate the battery life of the battery to be predicted. Performing prediction calculations based on the selected relevant parameter features reduces the calculation of irrelevant parameter features, improving calculation efficiency while also facilitating improved battery life prediction accuracy.

[0126] FIG7 is a schematic diagram of a battery life prediction device provided in an embodiment of the present application. As shown in FIG7 , the battery life prediction device 7 of this embodiment includes: a processor 70, a memory 71, and a computer program 72 stored in the memory 71 and executable on the processor 70, such as a battery life prediction program. When the processor 70 executes the computer program 72, the training method of each battery life prediction model described above, or the steps in the battery life prediction method embodiment, is implemented. Alternatively, when the processor 70 executes the computer program 72, the functions of each module / unit in the above-mentioned device embodiments are implemented.

[0127] Exemplarily, the computer program 72 may be divided into one or more modules / units, which are stored in the memory 71 and executed by the processor 70 to implement the present application. The one or more modules / units may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program 72 in the battery life prediction device 7.

[0128] The battery life prediction device 7 can be a computing device such as a desktop computer, laptop, PDA, vehicle-mounted unit, or cloud server. The battery life prediction device may include, but is not limited to, a processor 70 and a memory 71. Those skilled in the art will appreciate that FIG7 is merely an example of the battery life prediction device 7 and does not limit the battery life prediction device 7. The battery life prediction device 7 may include more or fewer components than shown, or may combine certain components or different components. For example, the battery life prediction device may also include input / output devices, network access devices, buses, and the like.

[0129] The processor 70 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0130] The memory 71 can be an internal storage unit of the battery life prediction device 7, such as a hard disk or memory of the battery life prediction device 7. The memory 71 can also be an external storage device of the battery life prediction device 7, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the battery life prediction device 7. Furthermore, the memory 71 can also include both an internal storage unit and an external storage device of the battery life prediction device 7. The memory 71 is used to store the computer program and other programs and data required by the battery life prediction device. The memory 71 can also be used to temporarily store data that has been output or is to be output.

[0131] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0132] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0133] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0134] In the embodiments provided in this application, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are merely illustrative. For example, the division of the modules or units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0135] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0136] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0137] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the process in the above-mentioned embodiment method, which can also be completed by hardware related to computer program instructions. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by the processor, it can implement the steps of the above-mentioned various method embodiments. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0138] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A training method for a battery life prediction model, It is characterized in that The method comprises: Obtaining a parameter feature set in sample data related to the battery to be predicted; Determining the correlation between the parameter characteristics in the parameter characteristic set and the battery life; Determining, according to the correlation, relevant parameter features in the parameter feature set for predicting battery service life; A prediction model for the battery service life of the battery to be predicted is trained according to relevant parameter features in the sample data.

2. The method according to claim 1, It is characterized in that Determining the correlation between the parameter characteristics in the parameter characteristic set and the battery life includes: Determine a mutual information value between each parameter feature in a parameter feature subset and a battery life, the parameter feature subset comprising a predetermined number of parameter features, and the parameter features belong to the parameter feature set; Determining a first average value of mutual information values ​​of a predetermined number of the parameter features and the battery life; The correlation between the parameter characteristic and the battery life is determined according to the first average value.

3. The method according to claim 2, It is characterized in that Before determining the mutual information value between each parameter feature in the parameter feature subset and the battery life, the method further includes: calculating predicted battery life when the parameter feature subset includes different numbers of parameter features; The difference between the predicted battery life and the actual battery life of the sample data is determined, and the number of parameter features corresponding to the minimum difference is obtained as the predetermined number.

4. The method according to claim 2, It is characterized in that Before determining, according to the correlation, relevant parameter features in the parameter feature set for predicting the battery service life, the method further includes: Calculating a second average value of mutual information values ​​between parameter features in the parameter feature subset; Determining, according to the correlation between the parameter features in the parameter feature set and the battery service life, the relevant parameter features in the parameter feature set for predicting the battery service life, including: According to the maximum value of the difference between the first average value and the second average value, the relevant parameter feature in the parameter feature set for predicting the battery service life is determined.

5. The method according to claim 1, It is characterized in that Training a prediction model for the battery service life of the battery to be predicted according to relevant parameter features in the sample data includes: The relevant parameter features in the sample data are input into a preset battery life prediction model for calculation, and the predicted battery life is output. The parameters of the prediction model are adjusted using the difference between the lifespans until the difference between the two meets the predetermined requirements, thereby completing the training of the prediction model.

6. The method according to claim 5, It is characterized in that The prediction model of battery life is a support vector machine model.

7. The method according to claim 6, It is characterized in that The method comprises: calculating a preset battery life prediction model through the relevant parameter characteristics, outputting a predicted battery life, and adjusting parameters of the battery life prediction model according to a difference between the predicted battery life and the battery life calibrated in the sample data, including: Determining the initial value of the parameter, the parameter value range and the adjustment step of the parameter in the support vector machine model; Inputting relevant parameter features in the sample data into the support vector machine model, gradually adjusting the parameters in the support vector machine model according to the adjustment step, and calculating and outputting a plurality of battery lifespans; The outputted multiple battery lifespans are compared with the battery lifespans calibrated in the sample data respectively, and the parameter values ​​when the outputted battery lifespans are closest to the battery lifespans calibrated in the sample data are used as the parameter values ​​of the support vector machine model.

8. The method according to claim 7, It is characterized in that The parameters of the support vector machine model include one or more of a penalty coefficient of the support vector machine model, a gamma parameter whose kernel function is a radial basis function parameter, an adjustment step size of the penalty coefficient, and an adjustment step size of the gamma parameter.

9. The method according to claim 8, It is characterized in that Determining the parameter value range in the support vector machine model includes: Determine multiple battery service lives output by the support vector machine model according to a first parameter value range of the support vector machine model and a first adjustment step corresponding to the first parameter value range; A second parameter value range and a second adjustment step corresponding to the second parameter value range are determined based on multiple battery service lives output by the support vector machine model, wherein the second parameter value range belongs to the first parameter value range and is smaller than the first parameter value range, and the second adjustment step is smaller than the first adjustment step.

10. The method according to any one of claims 5 to 9, It is characterized in that Training a prediction model for the battery service life of the battery to be predicted according to relevant parameter features in the sample data includes: The relevant parameter characteristics of the battery to be predicted are input into the trained battery service life prediction model, and the predicted value of the battery service life of the battery to be predicted is output through calculation of the battery service life prediction model.

11. The method according to any one of claims 1 to 10, It is characterized in that The parameter feature set includes one or more of the charge and discharge rate, charge and discharge cycle number, battery state of charge, charge and discharge state, vehicle operating state, battery temperature, battery internal resistance, battery capacitance, and impedance data of different frequencies of a battery of the same type as the battery to be predicted.

12. The method according to claim 11, It is characterized in that The vehicle running state includes one or more of a vehicle driving state, a vehicle idling state and a vehicle shutdown state.

13. A method for predicting battery service life. It is characterized in that The method comprises: Obtain relevant parameter characteristics of the battery to be predicted, and input the relevant parameter characteristics into any training method of claims 1-12 to obtain a prediction model, and predict the battery service life of the battery to be predicted.

14. The method according to claim 13, It is characterized in that Obtain relevant parameter characteristics of the battery to be predicted, including: According to the method according to any one of claims 1 to 12, relevant parameter characteristics of the battery to be predicted are obtained.

15. A training device for a prediction model of battery life, It is characterized in that The device comprises: A parameter feature set acquisition unit, used to acquire a parameter feature set from sample data related to the battery to be predicted; A correlation determination unit, used to determine the correlation between the parameter features in the parameter feature set and the battery life; A relevant parameter feature determination unit, used for determining, according to the correlation, relevant parameter features in the parameter feature set for predicting the battery service life; A training unit is used to train a prediction model for the battery service life of the battery to be predicted according to relevant parameter features in the sample data.

16. A battery life prediction device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. It is characterized in that When the processor executes the computer program, the steps of the method according to any one of claims 1 to 12 are implemented.

17. A computer-readable storage medium storing a computer program. It is characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 12 are implemented.

Images