Method and device for early prediction of remaining life of battery based on information characteristics

Abstract

The invention discloses a battery remaining life early prediction method and device based on information characteristics, and the method comprises the steps: carrying out the data pairing of battery signals in different use periods, so as to construct a cyclic evolution characteristic; constructing information statistical characteristics according to the statistical characteristics of the battery signals; the cyclic evolution features and the information statistical features form a battery feature set, and effective features are selected from the battery feature set; constructing a regression prediction model, and determining model parameter distribution of the regression prediction model through Bayesian ridge regression based on the effective features; and determining a group of model parameters in the model parameter distribution, setting the model parameters in a regression prediction model, and inputting an electrical measurement signal to be measured into the regression prediction model to predict the service life of the battery. The problems that an existing battery life prediction method is poor in generalization ability and low in prediction accuracy are solved. The service life of the battery can be predicted more accurately, so that the use safety of the battery can be improved.

Description

Technical Field

[0001] The present application relates to the field of battery technology, and in particular to a method and device for early prediction of remaining battery life based on information features. Background Art

[0002] Lithium-ion batteries have gained great popularity in recent years in a variety of applications from smartphones to electric vehicles due to their excellent properties such as high energy efficiency, long cycle life and reasonable cost. However, the performance of batteries seriously degrades after long cycles due to various factors such as inappropriate usage strategies and material loss. In addition, due to the long battery life, it takes a long time to obtain feedback on its performance.

[0003] Existing battery life prediction methods can be divided into two categories: model-based methods and machine learning (ML)-based methods. In model-based methods, battery capacity is modeled as a function of time or charge amount, and empirical mathematical equations are proposed to fit the capacity decay pattern. The parameters of the mathematical equations are usually obtained using least squares methods, genetic algorithms, and filtering techniques. Unlike model-based methods, machine learning (ML)-based methods do not rely on explicit battery degradation mechanisms, but instead study the statistical relationship between extracted features and battery life.

[0004] Model-based methods have been shown to be effective in battery life prediction, but their performance is highly dependent on the specific form of the formula and the estimation of model parameters, which leads to poor generalization under various operating conditions. Machine learning-based studies only consider changes within a fixed cycle and fail to fully utilize data from early cycles to fully explore battery characteristics. In addition, most existing methods extract features empirically, so they can only find suboptimal low-dimensional feature subsets, making it difficult to accurately predict battery life. Summary of the invention

[0005] The embodiments of the present application provide a method and device for early prediction of remaining battery life based on information features, thereby solving the problems of poor generalization ability and low prediction accuracy of existing battery life prediction methods.

[0006] In a first aspect, an embodiment of the present application provides an early prediction method for the remaining battery life based on information features, comprising: performing data pairing on battery signals of different usage cycles to construct a cycle evolution feature; constructing an information statistical feature based on the statistical characteristics of the battery signal; forming a battery feature set with the cycle evolution feature and the information statistical feature, and selecting effective features therefrom; constructing a regression prediction model, and determining the model parameter distribution of the regression prediction model through Bayesian Ridge regression based on the effective features; determining a set of model parameters in the model parameter distribution, setting them in the regression prediction model, and inputting the electrical measurement signal to be measured into the regression prediction model to predict the battery life.

[0007] In combination with the first aspect, in a possible implementation, data pairing is performed on battery signals of different usage cycles to construct a cyclic evolution feature, including: constructing a battery signal curve based on the battery signal, and data pairing the signal curve based on the usage cycle; calculating the characteristic difference between the paired signal curves, and constructing a preliminary evolution feature based on the characteristic difference; determining the correlation coefficient between the preliminary evolution feature and its corresponding remaining battery life; screening a set number of the preliminary evolution features based on the correlation coefficient, and performing dimensionality reduction processing on them to obtain the cyclic evolution feature.

[0008] In combination with the first aspect, in a possible implementation, determining the correlation coefficient between the preliminary evolution feature and the remaining battery life includes: converting the preliminary evolution feature and its corresponding remaining battery life into vectors respectively to obtain a feature vector and a remaining life vector; determining their covariance matrix and variance matrix based on the feature vector and the remaining life vector; and determining the correlation coefficient based on the covariance matrix and the variance matrix.

[0009] In combination with the first aspect, in a possible implementation manner, the information statistical features include charging-related features, discharging-related features, capacity-related features, and internal resistance-related features.

[0010] In combination with the first aspect, in a possible implementation method, the selecting effective features includes: evaluating the contribution value of each element in the battery feature set to the regression prediction model, and screening the elements in the battery feature set according to the contribution value to construct a feature subset; evaluating the importance of each element in the feature subset, and eliminating non-important elements according to the importance to obtain the effective features.

[0011] In combination with the first aspect, in a possible implementation, before obtaining the effective features, the method further includes: using the Akaike information criterion to perform evaluation and cross-validation to optimize the feature subset.

[0012] In combination with the first aspect, in a possible implementation method, determining the model parameter distribution of the regression prediction model through Bayesian ridge regression based on the effective features includes: setting initial model parameters and the prior distribution of the initial model parameters for the regression prediction model; determining the likelihood function of the initial model parameters based on the prior distribution and the effective features; using Bayes' theorem to determine the posterior distribution of the initial model parameters based on the prior distribution and the likelihood function; and determining the model parameter distribution based on the posterior distribution.

[0013] In the second aspect, an embodiment of the present application provides an early prediction device for the remaining battery life based on information features, including: a cycle evolution feature module, used to perform data pairing on battery signals of different usage cycles to construct a cycle evolution feature; an information statistical feature module, used to construct an information statistical feature based on the statistical characteristics of the battery signal; an effective feature module, used to form a battery feature set with the cycle evolution feature and the information statistical feature, and select effective features therefrom; a model parameter module, used to construct a regression prediction model, and determine the model parameter distribution of the regression prediction model through Bayesian Ridge regression based on the effective features; a prediction module, used to determine a set of model parameters in the model parameter distribution, set them in the regression prediction model, and input the battery signal to be tested into the regression prediction model to predict the battery life.

[0014] In a third aspect, an embodiment of the present application provides a device, comprising: a processor; a memory for storing processor-executable instructions; when the processor executes the executable instructions, it implements the method described in the first aspect or any possible implementation method of the first aspect.

[0015] In a fourth aspect, an embodiment of the present application provides a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium includes a medium for storing a computer program or instructions, which, when executed, enables the method described in the first aspect or any possible implementation method of the first aspect to be implemented.

[0016] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0017] The embodiment of the present application forms a battery feature set with cycle evolution features and information statistical features, and selects effective features therefrom, which can fully explore battery changes, improve the prediction accuracy of the model, remove noise, and improve computing efficiency. By determining the model parameter distribution of the regression prediction model through Bayesian Ridge regression, the generalization ability and robustness of the model can be improved. It effectively solves the problem of poor generalization ability and low prediction accuracy of existing battery life prediction methods. It can more accurately predict battery life, thereby improving the safety of battery use, reducing the number of scrapped batteries, and improving energy efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 A flowchart of a method for early prediction of remaining battery life based on information features provided in an embodiment of the present application;

[0020] Figure 2 A flowchart of performing data pairing on battery signals of different usage cycles to construct cycle evolution characteristics provided in an embodiment of the present application;

[0021] Figure 3 A device structure diagram of a method for early prediction of remaining battery life based on information features provided in an embodiment of the present application. DETAILED DESCRIPTION

[0022] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments 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 creative work are within the scope of protection of the present invention.

[0023] The following describes some of the techniques involved in the embodiments of the present application to facilitate understanding, and they should be considered as merely exemplary. Therefore, it should be appreciated by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present application. Similarly, for the sake of clarity and conciseness, some descriptions of well-known functions and structures are omitted in the following description.

[0024] Figure 1 1 is a flowchart of a method for early prediction of remaining battery life based on information features provided by an embodiment of the present application, including steps 101 to 105. Figure 1 This is only an execution order shown in the embodiment of the present application, and does not represent the only execution order of the method for early prediction of the remaining battery life based on information features. If the final result can be achieved, Figure 1 The steps shown may be performed in parallel or reversed.

[0025] Step 101: Pair the data of battery signals of different usage cycles to construct cycle evolution characteristics. The specific implementation method of step 101 is as follows: Figure 2 As shown, it includes steps 201 to 204, which are specifically as follows.

[0026] Step 201: construct a battery signal curve according to the battery signal, and perform data pairing on the signal curve according to the use cycle. Specifically, in order to fully explore the evolution of the battery between different use cycles, the signal curves in all available cycles are data paired.

[0027] Exemplarily, the public MIT dataset from the Massachusetts Institute of Technology is selected. This dataset contains a variety of different battery fast charging data, including a combination of different charging strategies in the range of SOC (remaining power) from 0% to 80%. The experiments in the MIT dataset were carried out at 30°C, using a constant current discharge method with a current of 4C, where 1C represents a current of 1.1A, until the battery reaches a cut-off voltage of 2.0V. Multiple parameters in the recording process, including voltage, current, temperature, etc. This application performs 10-fold cross validation on the MIT dataset, using battery signals in the first 100 cycles to compare the performance of different feature extraction methods, feature selection methods, and prediction models. The battery signals of 100 cycles of the battery are obtained, and the signal curves of each usage cycle are constructed according to the battery signals. The signal curves from the 1st cycle to the 50th cycle are taken as a group, and the signal curves from the 51st cycle to the 100th cycle are taken as a group. The signal curves of the two groups are paired separately to obtain 2500 cycle pair data.

[0028] Step 202: Calculate the characteristic differences between paired signal curves, and construct preliminary evolution characteristics based on the characteristic differences. Specifically, calculate the statistical changes of each pair of cycle pairs of data. Calculate the differences between 2500 cycle pairs of data respectively, obtain the characteristic differences between signal curves, and determine preliminary evolution characteristics based on the characteristic differences. By analyzing the statistical differences of signal curves of different usage cycles, the changes in the battery can be fully explored.

[0029] Step 203: Determine the correlation coefficient between the preliminary evolution feature and its corresponding remaining battery life. In the embodiment of the present application, the correlation coefficient between each preliminary evolution feature and the remaining battery life is calculated to evaluate the correlation between them. The Pearson correlation coefficient is used exemplarily here, which can be efficiently calculated by matrix multiplication and summation.

[0030] Specifically, the initial evolution characteristics and their corresponding remaining battery life are converted into vectors to obtain feature vectors and remaining life vectors. Based on the feature vectors and the remaining life vectors, their covariance matrix and variance matrix are determined. The correlation coefficient is determined based on the covariance matrix and the variance matrix. The variance matrix includes the variance matrix of the feature vector and the variance matrix of the remaining life vector.

[0031] The Pearson correlation coefficient is as follows:

[0032] Where ρ(X,Y) represents the Pearson correlation coefficient, Cov(X,Y) represents the covariance matrix between the eigenvector and the remaining useful life vector, Cov(X,X) represents the variance matrix of the eigenvector, Cov(Y,Y) represents the variance matrix of the remaining useful life vector, X represents the eigenvector, and Y represents the remaining useful life vector.

[0033] The calculated Pearson correlation coefficient can evaluate the linear correlation between each preliminary evolution feature and the remaining battery life. The value range of the Pearson correlation coefficient is between (-1,1). When the Pearson correlation coefficient is close to 1, it indicates that there is a strong positive correlation between the preliminary evolution feature and the remaining battery life. When the Pearson correlation coefficient is close to -1, it indicates that there is a strong negative correlation between the preliminary evolution feature and the remaining battery life. When the Pearson correlation coefficient is close to 0, it indicates that there is basically no relationship between the feature and the remaining battery life.

[0034] Step 204: Filter a set number of preliminary evolution features according to the correlation coefficient, and perform dimensionality reduction processing on them to obtain cyclic evolution features. Specifically, by calculating the Pearson correlation coefficient, select preliminary evolution features that are highly correlated with the remaining service life of the battery to improve the accuracy and reliability of the prediction model. Exemplarily, the set number here is 21.

[0035] The cycle evolution characteristics are constructed based on the evolutionary behavior of the battery during the cyclic charge and discharge process. They reflect the performance changes and degradation trends of the battery during the cyclic use process and help reveal the cycle life characteristics of the battery.

[0036] In addition, the preliminary evolution features are multidimensional, and the selected preliminary evolution features are subjected to dimensionality reduction. Exemplarily, the k-dimensional features of the selected preliminary evolution features are converted into one-dimensional features using partial least squares regression (PLSR), and 21 cycle evolution features that are highly correlated with battery life are obtained. By performing dimensionality reduction on the preliminary evolution features, preliminary evolution features that are irrelevant or weakly correlated with the remaining service life of the battery can be eliminated, reducing noise and redundant information, thereby improving the effect of the later regression prediction model.

[0037] Among them, partial least squares regression is a statistical method. By performing partial least squares regression, the screened preliminary evolution features can be projected into a low-dimensional space to find the projection direction that is most relevant to the battery life. The advantage of this is that the dimensions of the screened preliminary evolution features can be reduced and the features with the highest correlation with the battery life can be extracted, thereby improving the prediction performance of the subsequent regression prediction model. Finally, 21 cycle evolution features that are highly correlated with the battery life are obtained for each battery cell.

[0038] As before, the battery signal difference between the two cycles is rich in information and highly correlated with the battery life, so the proposed 21-cycle cycle evolution features contribute significantly to improving the prediction accuracy. Since each preliminary evolution feature constructed in the first stage is high-dimensional, directly using these preliminary evolution features as input to the model will result in a large amount of redundant information, resulting in poor model prediction performance and high computational complexity.

[0039] Directly performing partial least squares regression on the preliminary evolution features is quite expensive and contains information about the noise data, which will take more time. If the preliminary evolution features with the highest correlation are directly selected without performing partial least squares regression, information loss may occur because many differences in the data of the information-rich period pairs are ignored. Therefore, the Pearson correlation coefficient is calculated first, and then the partial least squares regression is used to reduce the dimension, and the high-dimensional preliminary evolution features screened out are converted into scalar values.

[0040] In the embodiment of the present application, the cycle evolution characteristics of the 21 use cycles finally constructed can be divided into six categories to cover multiple aspects of information of the battery during the cycle process. These categories include: voltage capacity curve, incremental capacity curve, time-voltage curve, voltage-temperature curve, internal resistance and capacity. The details are as follows:

[0041] Voltage-capacity curve: This type of cycle evolution feature is based on the relationship between voltage and capacity, and includes the mean, maximum, minimum, and variance of the difference between the two voltage-capacity curves in the data for each cycle pair.

[0042] Incremental capacity curve: This type of cycle evolution feature is calculated based on the rate of change of capacity in each cycle, which can reflect the changing trend of battery capacity, including the mean, maximum, minimum and variance of the difference between the two incremental capacity curves in the data for each cycle.

[0043] Time-voltage curve: This type of cyclic evolution characteristics describes the change trend of voltage over time, including the mean, maximum, minimum, skewness and kurtosis of the difference between two time-voltage curves in the data for each cycle.

[0044] Voltage-Temperature Curves: This type of cyclic evolution feature is based on the relationship between voltage and temperature and includes the mean, maximum, minimum, and variance of the difference between two voltage-temperature curves in the data for each cycle.

[0045] Internal resistance: This type of cycle evolution characteristics reflects the changes in the internal resistance of the battery, including the change in internal resistance.

[0046] Capacity: This type of cycle evolution characterization focuses on the change in battery capacity, including the amount of capacity per cycle.

[0047] Step 102: Construct information statistical features based on the statistical characteristics of the battery signal. In the embodiment of the present application, 24 information statistical features are constructed based on the statistical characteristics of the battery signal (average value, variance, maximum value, minimum value, etc.). These information statistical features can be divided into four categories: charging-related features, discharging-related features, capacity-related features, and internal resistance-related features. The details are as follows.

[0048] Charging-related characteristics: Charging-related characteristics are related to the performance changes of the battery during the charging process. For example, the statistical characteristics of charging time, charging current, and charging voltage can reflect the charging behavior and charging performance of the battery.

[0049] Discharge-related characteristics: Discharge-related characteristics are related to the performance changes of the battery during the discharge process. For example, the statistical characteristics of discharge time, discharge current, and discharge voltage can reflect the discharge behavior and discharge performance of the battery.

[0050] Capacity-related characteristics: Capacity-related characteristics are related to the change and attenuation of battery capacity. For example, characteristics such as the initial capacity, capacity attenuation rate, and capacity retention rate of the battery can reflect the capacity attenuation of the battery.

[0051] Internal resistance related features: The internal resistance related features are related to the change of the internal impedance of the battery. The increase of the internal resistance of the battery will lead to the degradation of battery performance and capacity attenuation. The internal resistance related features can include the average value, maximum value, minimum value, etc. of the internal resistance of the battery.

[0052] These information statistics provide a description of battery performance and degradation, and can be used as an important indicator to measure battery health and predict the remaining battery life. Information statistics are constructed based on the statistical characteristics of battery signals, and can describe the distribution, fluctuation and extreme value of battery signals, thereby reflecting the stability and degree of change of the battery.

[0053] Step 103: The cycle evolution features and the information statistical features constitute a battery feature set, and select effective features from them. In the embodiment of the present application, the contribution value of each element in the battery feature set to the regression prediction model is evaluated, and the elements in the battery feature set are screened according to the contribution value to construct a feature subset. The importance of each element in the feature subset is evaluated, and non-important elements are eliminated according to the importance to obtain effective features.

[0054] Specifically, in order to evaluate the significance of each feature element of the battery feature set, the available feature elements are divided into k groups, and k training and testing are performed, each time using a different group as the test set. In this way, the prediction value of each sample can be obtained and the AIC (Akaike Information Criterion) value can be conveniently calculated.

[0055] The Akaike information criterion is used for evaluation and cross-validation to optimize the feature subset. Specifically, the Akaike information criterion is a criterion for balancing model complexity and goodness of fit. The specific formula is as follows:

[0056] AIC = 2k-2ln(L). In the formula, k is the number of model parameters and L is the likelihood function of the model.

[0057] The core idea of ​​using the Akaike Information Criterion is to avoid overfitting while achieving goodness of fit. The smaller the AIC value, the better. A smaller AIC value indicates that the model has better balance when fitting the data.

[0058] Step 104: Construct a regression prediction model, and determine the model parameter distribution of the regression prediction model through Bayesian ridge regression based on the effective features. In the embodiment of the present application, Lasso (Least Absolute Shrinkage and Selection Operator) regression is used as the regression prediction model for AIC value calculation, and the regularization coefficient of the regression prediction model is determined by cross-validation.

[0059] Among them, Lasso regression is a linear regression model that constrains the complexity of the regression prediction model by adding an L1 regularization term. Cross-validation is a method for evaluating the generalization ability of a model. It divides the data set into a training set and a validation set, and repeatedly trains and evaluates the model through different divisions to select the best regularization coefficient.

[0060] In an embodiment of the present application, initial model parameters and prior distributions of the initial model parameters are set for the regression prediction model. The likelihood function of the initial model parameters is determined based on the prior distribution and the effective features. Using Bayes' theorem, the posterior distribution of the initial model parameters is determined based on the prior distribution and the likelihood function. The distribution of model parameters is determined based on the posterior distribution.

[0061] Specifically, determine the prior distribution. Assume that the regression coefficient w of the regression prediction model has a mean of 0 and a standard deviation of λ -1 I p The output y follows a normal distribution with a mean of Xw and a standard deviation of α, as follows:

[0062] Among them, X represents the input of the regression prediction model, and p represents the probability.

[0063] The random variables α and λ are assumed to follow the gamma distribution, which is a conjugate prior of the normal distribution, thus simplifying the computation. The parameters of these gamma distributions are hyperparameters and are set to 10 in this work. -6. α and λ are estimated by iteratively maximizing the marginal likelihood. Compared with other prediction models such as support vector machine (SVM), the adopted Bayesian linear ridge regression can not only adaptively adjust the model complexity based on the available effective features, but also provide prediction confidence intervals.

[0064] Prior distribution:

[0065] Calculate the posterior distribution. Using Bayes' theorem, combine the prior distribution and the likelihood function to calculate the posterior distribution of the regression coefficient w.

[0066] Likelihood function:

[0067] Posterior distribution: p(w|X,y,α,λ)∝p(y|X,w,α)p(w|λ).

[0068] The model parameters α and λ are estimated by maximizing the marginal likelihood. The expectation maximization (EM) algorithm or iterative method can be used for optimization. Based on the posterior distribution, the new data X new Make predictions, giving the predicted distribution of model parameters:

[0069] Among them, y new Represents the new data X new The predicted new output, T represents the transpose, μ w Represents the mean of the distribution.

[0070] Through these steps, Bayesian linear ridge regression can provide not only predicted values, but also confidence intervals for the predicted values, thus reflecting the uncertainty of the predicted results.

[0071] Step 105: Determine a set of model parameters in the model parameter distribution, set them in the regression prediction model, and input the electrical measurement signal to be measured into the regression prediction model to predict the battery life. Specifically, a set of model parameters is randomly determined in the model parameter distribution, set them in the regression prediction model, and input the electrical measurement signal to be measured into the regression prediction model to predict the battery life.

[0072] Although the present application provides method operation steps as described in the embodiments or flowcharts, more or fewer operation steps may be included based on conventional or non-creative labor. The order of steps listed in this embodiment is only one way of executing the order of many steps and does not represent the only execution order. When the actual device or client product is executed, it can be executed in sequence or in parallel according to the method shown in this embodiment or the accompanying drawings (for example, in a parallel processor or multi-threaded processing environment).

[0073] like Figure 3As shown, the embodiment of the present application also provides an early prediction device 300 for the remaining battery life based on information features. The device includes: a cycle evolution feature module 301, an information statistics feature module 302, an effective feature module 303, a model parameter module 304 and a prediction module 305, as follows.

[0074] The cycle evolution feature module 301 is used to perform data pairing on battery signals of different usage cycles to construct a cycle evolution feature.

[0075] The information statistical feature module 302 is used to construct information statistical features according to the statistical characteristics of the battery signal.

[0076] The effective feature module 303 is used to form a battery feature set by combining cycle evolution features and information statistical features, and select effective features therefrom.

[0077] The model parameter module 304 is used to construct a regression prediction model and determine the model parameter distribution of the regression prediction model through Bayesian Ridge regression based on effective features.

[0078] The prediction module 305 is used to determine a set of model parameters in the model parameter distribution, set them in the regression prediction model, and input the battery signal to be tested into the regression prediction model to predict the battery life.

[0079] Some modules in the apparatus described in the present application can be described in the general context of computer executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, classes, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0080] The devices or modules described in the above application embodiments can be implemented by computer chips or entities, or by products with certain functions. For the convenience of description, the above devices are described in various modules according to their functions. When implementing the embodiments of the present application, the functions of each module can be implemented in the same or multiple software and / or hardware. Of course, the module that implements a certain function can also be implemented by combining multiple sub-modules or sub-units.

[0081] The methods, devices or modules described in this application can be implemented in the form of computer-readable program codes. The controller can be implemented in any appropriate manner. For example, the controller can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program codes (such as software or firmware) that can be executed by the (micro) processor, logic gates, switches, application-specific integrated circuits (English: Application Specific Integrated Circuit; Abbreviation: ASIC), programmable logic controllers and embedded microcontrollers. Examples of controllers include but are not limited to the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20 and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art also know that in addition to implementing the controller in a purely computer-readable program code, the controller can be implemented in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers and embedded microcontrollers by logically programming the method steps. Therefore, this controller can be considered as a hardware component, and the devices included in it for implementing various functions can also be regarded as structures within the hardware component. Or even, the means for realizing various functions may be regarded as both a software module for realizing the method and a structure within a hardware component.

[0082] An embodiment of the present application further provides a device, comprising: a processor; a memory for storing processor executable instructions; when the processor executes the executable instructions, the method described in the embodiment of the present application is implemented.

[0083] The embodiments of the present application also provide a non-volatile computer-readable storage medium on which a computer program or instruction is stored. When the computer program or instruction is executed, the method described in the embodiments of the present application is implemented.

[0084] In addition, each functional module in each embodiment of the present invention may be integrated into one processing module, or each module may exist independently, or two or more modules may be integrated into one module.

[0085] The above storage media include but are not limited to random access memory (RAM), read-only memory (ROM), cache, hard disk (HDD) or memory card. The memory can be used to store computer program instructions.

[0086] It can be known from the description of the above implementation methods that those skilled in the art can clearly understand that the present application can be implemented by means of software plus necessary hardware. Based on such an understanding, the technical solution of the present application can be essentially or partly contributed to the prior art in the form of a software product, or it can be reflected in the implementation process of data migration. The computer software product can be stored in a storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a mobile terminal, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present application or certain parts of the embodiments.

[0087] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. All or part of this application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld devices or portable devices, tablet devices, mobile communication terminals, multi-processor systems, microprocessor-based systems, programmable electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc.

[0088] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit the present application. Although the present application has been described in detail with reference to the aforementioned embodiments, a person of ordinary skill in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some or all of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the present application.

Claims

1. A method for early prediction of remaining battery life based on information features, characterized in that: include: Data pairing of battery signals at different usage cycles to construct cycle evolution features; Constructing information statistical features according to the statistical characteristics of the battery signal; The cycle evolution feature and the information statistical feature constitute a battery feature set, and select effective features from them; Constructing a regression prediction model, and determining the distribution of model parameters of the regression prediction model through Bayesian ridge regression based on the effective features; A set of model parameters is determined in the model parameter distribution, and is set in the regression prediction model. The battery signal to be tested is input into the regression prediction model to predict the battery life.

2. The method according to claim 1, characterized in that: The data pairing of battery signals of different usage cycles to construct cycle evolution characteristics includes: Constructing a battery signal curve according to the battery signal, and performing data pairing on the signal curve according to a usage cycle; Calculating feature differences between pairs of the signal curves, and constructing preliminary evolution features based on the feature differences; Determining a correlation coefficient between the preliminary evolution feature and its corresponding remaining battery life; A set number of the preliminary evolution features are screened according to the correlation coefficient, and dimension reduction processing is performed on them to obtain the cyclic evolution features.

3. The method according to claim 2, characterized in that Determining the correlation coefficient between the preliminary evolution feature and the remaining service life of the battery includes: Convert the preliminary evolution feature and its corresponding remaining battery life into vectors to obtain a feature vector and a remaining battery life vector; Based on the eigenvector and the remaining useful life vector, determining a covariance matrix and a variance matrix thereof; The correlation coefficient is determined according to the covariance matrix and the variance matrix.

4. The method according to claim 1, characterized in that: The information statistical features include charging-related features, discharging-related features, capacity-related features and internal resistance-related features.

5. The method according to claim 1, characterized in that The selecting of valid features includes: Evaluate the contribution value of each element in the battery feature set to the regression prediction model, and select the elements in the battery feature set according to the contribution value to construct a feature subset; The importance of each element in the feature subset is evaluated, and unimportant elements are eliminated according to the importance to obtain the effective features.

6. The method according to claim 5, characterized in that Before obtaining the effective features, the method further includes: The feature subset was optimized using Akaike's information criterion for evaluation and cross-validation.

7. The method according to claim 1, characterized in that The determining the distribution of model parameters of the regression prediction model by Bayesian Ridge regression based on the effective features comprises: Setting initial model parameters and prior distribution of the initial model parameters for the regression prediction model; Determine a likelihood function of the initial model parameters based on the prior distribution and the effective features; Determine the posterior distribution of the initial model parameters based on the prior distribution and the likelihood function using Bayes' theorem; The model parameter distribution is determined based on the posterior distribution.

8. An early prediction device for remaining battery life based on information features, characterized in that: include: The cycle evolution feature module is used to perform data pairing on battery signals of different usage cycles to construct cycle evolution features; An information statistical feature module, used to construct information statistical features according to the statistical characteristics of the battery signal; An effective feature module, used for forming a battery feature set from the cycle evolution feature and the information statistical feature, and selecting effective features therein; A model parameter module, used to construct a regression prediction model and determine the model parameter distribution of the regression prediction model through Bayesian Ridge regression based on the effective features; The prediction module is used to determine a set of model parameters in the model parameter distribution, set the set of model parameters in the regression prediction model, and input the battery signal to be tested into the regression prediction model to predict the battery life.

9. A device for executing an early prediction method of battery remaining life based on information features, characterized in that: include: processor; a memory for storing processor-executable instructions; When the processor executes the executable instructions, the method according to any one of claims 1 to 7 is implemented.

10. A non-volatile computer-readable storage medium, characterized in that: The device comprises a computer program or an instruction for storing the computer program or the instruction, which, when executed, enables the method according to any one of claims 1 to 7 to be implemented.

Images