A battery health status assessment method, device and electronic equipment

Abstract

The present invention discloses a battery health status evaluation method, device and electronic device, the method comprising: determining the candidate features related to the health status of the lead-acid battery according to the incremental capacity expression of the lead-acid battery; determining the importance of the candidate features by a random forest algorithm, and determining the target features related to the health status of the lead-acid battery based on the importance; constructing a source model of the lead-acid battery based on the training data to be used corresponding to the target features, and determining the weight coefficient of the source model, and evaluating the health status of the target lead-acid battery based on the source model and the weight coefficient; wherein the source model is a support vector regression model. The technical solution of the present invention realizes the evaluation of the health status of the lead-acid battery through data corresponding to a small number of features.

Description

Technical Field

[0001] The present invention relates to the technical field of battery management, and particularly to a method, device and electronic device for evaluating the state of health of a battery. Background Art

[0002] The battery management system is an essential part of the battery energy storage system and plays an important role in the state monitoring and operation control of lead-acid battery packs. The evaluation of the state of health (SOH) of lead-acid batteries is the core function of the lead-acid battery management system. Accurate SOH can help the Battery Management System (BMS) correctly judge the aging state of lead-acid batteries, which is of great significance for improving the fault prediction performance and ensuring the safe operation of the batteries.

[0003] Most of the existing SOH estimation methods still require a large amount of lead-acid battery aging data, and the established models often lack generalization ability. Therefore, it is urgent to improve the existing SOH estimation methods. Summary of the Invention

[0004] The present invention provides a method, device and electronic device for evaluating the state of health of a battery, which realizes the evaluation of the state of health of lead-acid batteries based on a small amount of characteristic data.

[0005] According to one aspect of the present invention, there is provided a method for evaluating the state of health of a battery, including:

[0006] Determining candidate features related to the state of health of the lead-acid battery according to the incremental capacity expression of the lead-acid battery;

[0007] Determining the importance degree of the candidate features through a random forest algorithm, and determining target features related to the state of health of the lead-acid battery based on the importance degree;

[0008] Constructing a source model of the lead-acid battery based on the training data to be used corresponding to the target features, and determining the weight coefficient of the source model, and evaluating the state of health of the target lead-acid battery based on the source model and the weight coefficient;

[0009] Wherein, the source model is a support vector regression model.

[0010] According to another aspect of the present invention, there is provided an apparatus for evaluating the state of health of a battery, including:

[0011] A candidate feature determination module, configured to determine candidate features related to the state of health of the lead-acid battery according to the incremental capacity expression of the lead-acid battery;

[0012] A target feature determination module, configured to determine the importance degree of the to-be-selected features through a random forest algorithm, and determine the target features related to the state of health of the lead-acid battery based on the importance degree;

[0013] A state of health assessment module, configured to construct a source model of the lead-acid battery based on the to-be-used training data corresponding to the target features, and determine the weight coefficients of the source model, and based on the source model and the weight coefficients, assess the state of health of the target lead-acid battery;

[0014] Wherein, the source model is a support vector regression model.

[0015] According to another aspect of the present invention, there is provided an electronic device, which includes:

[0016] At least one processor; and,

[0017] A memory communicatively connected to the at least one processor; wherein,

[0018] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the method for evaluating the state of health of the battery according to any embodiment of the present invention.

[0019] According to another aspect of the present invention, there is provided a computer-readable storage medium, which stores computer instructions for causing a processor to implement the method for evaluating the state of health of the battery according to any embodiment of the present invention when executed.

[0020] The technical solution of the embodiment of the present invention determines the to-be-selected features related to the state of health of the lead-acid battery according to the incremental capacity expression of the lead-acid battery; determines the importance degree of the to-be-selected features through a random forest algorithm, and determines the target features related to the state of health of the lead-acid battery based on the importance degree; constructs a source model of the lead-acid battery based on the to-be-used training data corresponding to the target features, and determines the weight coefficients of the source model, and based on the source model and the weight coefficients, assesses the state of health of the target lead-acid battery. By using the random forest algorithm to select the target features that have a greater impact on the state of health of the lead-acid battery as the basis for evaluating the state of health of the lead-acid battery, the problem that most of the existing SOH estimation methods still require a large amount of battery aging data is solved, and the evaluation of the state of health of the target lead-acid battery based on a small amount of data is realized.

[0021] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present invention, nor is it used to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. Description of the Drawings

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0023] Figure 1 It is a flowchart of a method for evaluating the state of health of a battery provided in the first embodiment of the present invention;

[0024] Figure 2 It is a flowchart of a method for evaluating the state of health of a battery provided in the second embodiment of the present invention;

[0025] Figure 3 It is a schematic structural diagram of a device for evaluating the state of health of a battery provided in the third embodiment of the present invention;

[0026] Figure 4 It shows a schematic structural diagram of an electronic device that can be used to implement the embodiments of the present invention. Detailed implementation manners

[0027] In order to enable those skilled in the art to better understand the solutions of the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some of the embodiments of the present invention, rather than all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0028] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above accompanying drawings are used to distinguish similar objects, and do not necessarily need to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present invention described here can be implemented in an order other than those illustrated or described here. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or are inherent to these processes, methods, products, or devices.

[0029] Embodiment 1

[0030] Figure 1The flowchart of a method for evaluating the state of health of a battery provided in Embodiment 1 of the present invention. This embodiment is applicable to the situation of evaluating the health degree of a battery through a small amount of battery data. This method can be executed by an evaluation device for the state of health of a battery, and this device can be implemented in the form of hardware and / or software, and this device can be configured in an electronic device, such as a computer device. For example Figure 1 As shown, the method includes:

[0031] S110. Determine candidate features related to the state of health of the lead-acid battery according to the incremental capacity expression of the lead-acid battery.

[0032] Among them, the incremental capacity expression of the lead-acid battery is a mathematical expression used to describe the incremental capacity of the lead-acid battery on continuous voltage steps, and is obtained by comparing the capacity increment in the constant current charging stage with the voltage change. The candidate features can be various features that affect the state of health of the lead-acid battery. For example, the candidate feature can be the constant current charging time of the lead-acid battery. According to the incremental capacity expression of the lead-acid battery, various features related to the state of health of the lead-acid battery can be determined.

[0033] Based on the above solution, determining candidate features related to the state of health of the lead-acid battery according to the incremental capacity expression of the lead-acid battery includes: establishing the incremental capacity expression of the lead-acid battery based on the battery capacity and the battery voltage, and drawing the incremental capacity curve of the lead-acid battery based on the incremental capacity expression; determining candidate features related to the state of health of the lead-acid battery based on the incremental capacity curve.

[0034] It can be understood that the essence of the integrated circuit is to analyze the constant current charging data with a differential equation. Therefore, the lead-acid battery can be made to be in a constant current charging state, and then the incremental capacity expression of the lead-acid battery can be established based on the differential of the battery capacity and the differential of the battery voltage. Further collect data related to the incremental capacity expression in the lead-acid battery, such as battery capacity data, voltage data, etc., and draw the incremental capacity curve of the lead-acid battery based on the collected data. After the incremental capacity curve is drawn, some features that can describe the battery aging process are extracted from the curve as candidate features.

[0035] Exemplarily, the expression of incremental capacity (IC) is: In the formula, Q represents the battery capacity, and V represents the battery voltage.

[0036] Further, after plotting the IC curve, the candidate features are determined. The candidate features can be the voltage, peak area, peak value, incremental variance, fixed voltage increment difference, constant current charging time, maximum slope of constant current charging, curve slope at the end of constant current charging, curve slope of constant current charging, curve area of constant current charging, voltage change amount within a preset time interval, average voltage, voltage standard deviation, voltage mean absolute deviation, maximum voltage difference, voltage skewness, and voltage kurtosis corresponding to the incremental capacity curve.

[0037] Specifically, select the peak value of the IC curve, the voltage corresponding to the peak value, and the peak area as the candidate features reflecting the battery health state, denoted as F1 - F3. The incremental variance of the IC curve and the incremental difference between two fixed voltages are F4 and F5 respectively, and the two fixed voltages can be preset based on previous research and experience.

[0038]

[0039] F 5 = D a - D b

[0040] In the formula, D i is the IC value; is the average value of IC; D a and D b are the values of IC at two fixed voltages respectively.

[0041] During the constant current (CC) charging process of lead-acid batteries, take the CC charging time, the maximum slope of the CC curve, the curve slope at the end of CC charging, the CC charging times at two specific voltages, the amount of voltage change within a certain specific time interval, and the curve area in the CC mode as F6 - F11 respectively. Select the features related to voltage distribution, such as the average voltage, voltage standard deviation, voltage mean absolute deviation, and maximum voltage difference as F12 - F15 respectively. Finally, the skewness and kurtosis of the voltage at different cycles can also well describe the SOH of the battery. The skewness and kurtosis of the voltage are expressed as F16 and F17, that is, F1 - F17 are all candidate features.

[0042]

[0043]

[0044] In the formula, F16 is the voltage skewness; F17 is the voltage kurtosis; T is the sampling time; Sv is the voltage standard deviation; Vt is the voltage value; is the average value of the voltage.

[0045] S120, determining the importance of the candidate feature by using a random forest algorithm, and determining the target feature related to the health status of the lead-acid battery based on the importance.

[0046] The target feature refers to the feature that has a greater impact on the health status of the lead-acid battery, which is determined from the candidate features by the random forest algorithm.

[0047] Specifically, the importance of each candidate feature can be calculated by the random forest algorithm, and the importance can be represented by the corresponding importance value. After obtaining the importance value, the candidate features can be sorted according to the importance value, and can be sorted in descending order according to the size of the feature importance value, and then multiple candidate features with the highest ranking can be selected as target features related to the health status of the lead-acid battery.

[0048] In this embodiment, the importance of the extracted candidate features can be obtained through the random forest algorithm, and the target features are selected based on the importance to achieve feature dimensionality reduction. When evaluating the health status of the lead-acid battery, the data corresponding to some highly important candidate features are processed to evaluate the health status of the lead-acid battery. On the basis of ensuring the accuracy of the health status evaluation, the data required for the health status evaluation is reduced, and only the data corresponding to some target features are required, rather than a large amount of data corresponding to the candidate features, thereby reducing the complexity of the calculation.

[0049] In an embodiment of the present invention, determining the importance of the candidate feature by a random forest algorithm includes: determining a training data set corresponding to the random forest algorithm according to the candidate feature, training a random forest model to be used based on the training data set to obtain a target random forest model; performing importance analysis on each node in the random forest tree based on the target random forest model to obtain the amount of impurity reduction before and after each node is split; determining the importance of the candidate feature based on the amount of impurity reduction of the node corresponding to the candidate feature.

[0050] Wherein, the training data set may consist of a data vector containing the features to be selected and the health status value of the lead-acid battery corresponding to the data vector. For example, the data vector may be a vector consisting of specific values ​​of the voltage and peak area corresponding to the peak value of the IC curve, and the health status value of the lead-acid battery corresponding to the data vector may be a value measuring the health status of the lead-acid battery, for example, the health status value is 70%. This embodiment does not limit the number and type of features to be selected contained in the data vector.

[0051] To illustrate in detail, the data vector may be a vector consisting of voltage V=3 and peak area S=1, and the health status value of the lead-acid battery may be that when the lead-acid battery is at voltage V=3 and peak area S=1, the corresponding health status value is 60%.

[0052] It should be noted that the data vector can be determined from the historical charge and discharge data of the lead-acid battery, and the health state value of the lead-acid battery can be an observed value, that is, the value measured by the sensor. Each data vector and the corresponding health state value of the lead-acid battery are used as a set of training samples in the training dataset. The target random forest model is obtained by training the initial random forest model to be used with multiple sets of training samples in the training dataset. Among them, the random forest model to be used is the initial random forest model that has not undergone training iteration and model parameter optimization, while the target random forest model can be the random forest model obtained after training is completed.

[0053] In the embodiment of the present invention, the training process of the random forest model to be used may be to randomly select samples from the training dataset to construct a new data subset. Each tree of the random forest model to be used is trained separately, where each tree corresponds to a data subset. Each tree is trained through the corresponding data subset to obtain the input-output relationship of each tree, and the target random forest model is composed of the trained trees.

[0054] Exemplarily, the training dataset is the set Tn, T n ={(X 1 ,Y 1 ),…,(X n ,Y n ),X∈R m ,Y∈R

[0055] where each input vector is X = {x1, x2,..., xj}, x1, x2,..., xj represent feature variables, that is, the specific values of the features to be selected. The input vector is the data vector, and Y1....Yn are the observed values of the health state of the lead-acid battery corresponding to the input vector.

[0056] After determining the training dataset, by randomly sampling T n to obtain where k is the index of the tree in the random forest. That is, samples are randomly selected from the training dataset T n to construct a new data subset The tree with index k in the random forest is trained through this data subset

[0057] It should also be noted that during the random sampling process, after each sample is selected and placed in the new data subset, the sample is returned to the original dataset, making it possible to be selected again. This means that the new data subset may contain duplicate samples, and at the same time, some samples in the training dataset may be omitted. Similar operations are performed for the other trees in the random forest.

[0058] ​Let \(p\) be the number of trees in the random forest, \(d\) be the maximum depth of the trees, and \(L\) be the ratio of the training set to the test set. Train and test all batteries with the same parameters, and obtain the input-output relationships of each tree as follows:

[0059] Among them, \(Y\) k is the output value of the \(k\)-th tree in the random forest, and \(X\) m : represents the input vector, which contains \(m\) features.

[0060] Obtain the average estimated output of the random forest, that is, average the prediction results of all trees in the random forest to obtain the final prediction output

[0061]

[0062] Construct the target random forest model based on the above process.

[0063] To determine the importance degree of each candidate feature, the importance of each tree node in the random forest can be analyzed, and then the importance of the candidate features can be determined according to the importance of the nodes. In the random forest, the importance of each feature is determined by calculating the average value of the reduction in impurity for it in all trees. For multiple batteries, normalize the sum of the importance weights of all features to:

[0064]

[0065] Among them, \(f\) i,j represents the importance of the \(j\)-th feature in the \(i\)-th battery, and \(f_j\) represents the total reduction in impurity of feature \(j\) in all trees in the random forest.

[0066] Considering the differences between different batteries, define a new index to evaluate the comprehensive performance of each feature.

[0067]

[0068] Among them, \(N\) is the number of model batteries; is the total importance of the \(j\)-th feature.

[0069] Obtain the importance of each candidate feature in all batteries through the above calculations, and then evaluate and select the appropriate target features accordingly.

[0070] Based on the above embodiments, the number of candidate features is at least two, and the target features related to the state of health of the lead-acid battery are determined based on the importance degree, including: sorting at least two candidate features based on the importance degree, and selecting the target features from at least two candidate features based on the sorting result.

[0071] S130. Construct a source model of the lead-acid battery based on the training data to be used corresponding to the target feature, determine the weight coefficients of the source model, and evaluate the health state of the target lead-acid battery based on the source model and the weight coefficients.

[0072] Among them, the training data to be used corresponding to the target feature may include the specific values of the target feature and the health degree values of the battery, and the source model is a support vector regression model.

[0073] Specifically, a support vector regression model can be trained based on the data corresponding to the target feature, the model parameters can be optimized to obtain the corresponding source model. Multiple lead-acid batteries with the same parameters can be selected, and the source model can be obtained by training based on the training data to be used corresponding to them. Then, the weight coefficients of each source model are calculated, and multiple source models are superimposed to obtain a battery health state evaluation model for the target lead-acid battery. Furthermore, the health state of the target lead-acid battery can be evaluated through the battery health state evaluation model.

[0074] The technical solution of the embodiment of the present invention determines the candidate features related to the health state of the lead-acid battery according to the incremental capacity expression of the lead-acid battery; determines the importance degree of the candidate features through the random forest algorithm, and determines the target features related to the health state of the lead-acid battery based on the importance degree; constructs a source model of the lead-acid battery based on the training data to be used corresponding to the target feature, and determines the weight coefficients of the source model, and evaluates the health state of the target lead-acid battery based on the source model and the weight coefficients. By using the random forest algorithm to select the target features that have a greater impact on the health state of the lead-acid battery as the evaluation basis for the health state of the lead-acid battery, the problem that most of the existing SOH estimation methods still require a large amount of battery aging data is solved, and the health state of the target lead-acid battery can be evaluated based on a small amount of data.

[0075] Embodiment 2

[0076] Figure 2 It is a flowchart of a method for evaluating the health state of a battery provided by Embodiment 2 of the present invention. This embodiment is a preferred embodiment of the above embodiment. As Figure 2 shown, the method includes:

[0077] S210. Determine the candidate features related to the health state of the lead-acid battery according to the incremental capacity expression of the lead-acid battery.

[0078] S220. Determine the importance degree of the candidate features through the random forest algorithm, and determine the target features related to the health state of the lead-acid battery based on the importance degree.

[0079] S230. Divide the training data to be used corresponding to the target feature into a training set to be used and a test set to be used.

[0080] Among them, the training data to be used includes the sample feature values and sample health level values corresponding to the target feature; the sample feature values can be the specific values corresponding to the target feature, and the sample health level can be the health state evaluation value of the lead-acid battery when the lead-acid battery is in the state corresponding to the sample feature value.

[0081] Specifically, some lead-acid batteries can be selected as sample lead-acid batteries, and then the historical data of the sample lead-acid batteries during use or charging operation is determined, and the historical data corresponding to the target feature is determined from the historical data to form the training data to be used.

[0082] Exemplarily, when the target feature includes the peak value of the IC curve, the historical IC curve peak value of the lead-acid battery and the health state value of the battery when at this peak value are correspondingly obtained. For example, when the IC curve peak value is A, the health state of the battery is 80%, and when the IC curve peak value is B, the health level of the battery is 70%. Among them, the IC curve peak values A and B both belong to the sample feature values, and the health levels of the battery 80% and 70% both belong to the sample health level values. It should also be noted that the target feature can also include the average voltage difference, etc., and the corresponding sample feature values are the specific values corresponding to various target features.

[0083] Multiple sample feature values and sample health level values form the training data to be used, and then the training data to be used is divided into a training set to be used and a test set to be used according to a preset division ratio. For example, the number of the training set to be used is 70%, and the number of the test set to be used is 30%.

[0084] S240. According to the minimization problem of the source model to be trained, the source model to be trained is respectively trained and tested through the training set to be used and the test set, and the source model is obtained.

[0085] In the embodiment of the present invention, a regression function (i.e., the source model to be trained) can be constructed according to the principle of minimizing structural risk as follows:

[0086] Among them, h(x) is the output variable of Support Vector Regression (SVR); ω is the weight vector; φ(x) is the function that maps x from a low-dimensional space to a high-dimensional space; b is the bias. Define ε as the insensitive loss coefficient, and set the training set xd, yd and the test set xt, yt (common ratios include 70% training set and 30% test set, or 80% training set and 20% test set, etc.). To find the values of ω and b, the following minimization problem is established:

[0087]

[0088] Here, T is the transpose symbol

[0089] Subject to:

[0090]

[0091] To solve for ω and b, it can be transformed into:

[0092]

[0093] where and β i are Lagrange multipliers; K(x d , x t ) is the kernel function in SVR.

[0094] Since the generated target features are highly correlated with the SOH of the battery, considering that the kernel function needs to be accurate and easy to calculate, the present invention selects a linear function as the kernel function in SVR.

[0095] The expression of the linear kernel function is as follows:

[0096]

[0097] S250. Establish a source model for each of the lead-acid batteries, and train each source model based on the target sample data corresponding to the target lead-acid battery.

[0098] To improve the generalization ability of the model, multiple lead-acid batteries can be selected, and the corresponding source models for each lead-acid battery are established in the above manner.

[0099] Among them, the target sample data includes the target sample feature values corresponding to the target features and the target sample health degree values in the target lead-acid battery; the target lead-acid battery can be a lead-acid battery whose health state is to be evaluated, the target sample feature value can be the specific value corresponding to the target feature in the target lead-acid battery, and the target sample health degree value can be the health degree value of the target battery.

[0100] Specifically, use the target sample data as the training samples to train each source model. For example, for each SVR source model HB = {h1,..., hB}, substitute part of the data of the target lead-acid battery, that is, the target sample data T = {(x1, y1),..., (xn, yn)} into each source model as the training set, and the output can be expressed as:

[0101] O i,j = h j (x i )

[0102] where i and j represent the sample serial number and the source model index respectively.

[0103] Perform k-fold cross-validation on the target model. Then, use a small portion of the experimental data of the target battery to establish an SVR model of the target battery, denoted as hB+1. Its output is expressed as Oi,B+1.

[0104] S260. During the training process of the source model, by solving the optimization problem of the source model, the weight coefficients of the source model are obtained.

[0105] Specifically, after solving the optimization problem, the weight coefficients obtained by each model are as follows:

[0106]

[0107] Subject to the constraints:

[0108]

[0109] In the formula, a j is the weight coefficient; O i,j is the output of each model; yi is the true value of SOH; B+1 represents an additional model added on the basis of the original B source models. B+2 represents an additional parameter in this optimization problem, usually used for the bias term introduced in the optimization process or a parameter set to meet certain constraint conditions.

[0110] S270. Weight each of the source models based on the weight coefficients corresponding to the source models to obtain the target state-of-health evaluation model of the target lead-acid battery.

[0111] Among them, the target state-of-health evaluation model refers to a model used for evaluating the health state of the battery. Specifically, an SVR model based on TS (TS-SVR) is obtained, and its expression is:

[0112] h f (x) = a 1 h 1 (x) + … + a B+1 h B+1 (x) + a B+2

[0113] Specifically, if a certain specific source model h is highly correlated with the state of health of the battery, the magnitude of the coefficient a will be large, which means that h is given a large weight. On the other hand, if the source model has nothing to do with the state of health of the battery, a smaller coefficient will be learned, which means that the corresponding model is given a smaller weight.

[0114] S280. Use the feature value to be evaluated corresponding to the target feature in the target lead-acid battery as the input of the target state-of-health evaluation model to obtain the target lead-acid battery health state value output by the target state-of-health evaluation model.

[0115] Specifically, for a lead-acid battery whose health status needs to be evaluated, the parameter value corresponding to the target feature in the battery is the feature value to be evaluated. This value is used as the input to the target health degree evaluation model, and the model can output the result, that is, the health status value of the target lead-acid battery. For example, if the health status value of the target lead-acid battery is 80%, it means the health degree of the target lead-acid battery is 80%.

[0116] The technical solution of the embodiment of the present invention determines candidate features related to the health status of the lead-acid battery according to the incremental capacity expression of the lead-acid battery. The importance degree of the candidate features is determined by the random forest algorithm, and the target features related to the health status of the lead-acid battery are determined based on the importance degree. The training data to be used corresponding to the target features is divided into a training set to be used and a test set to be used. According to the minimization problem of the source model to be trained, the source model to be trained is trained and tested respectively through the training set to be used and the test set to be used, and the source model is obtained. The source model of each lead-acid battery is established, and each source model is trained based on the target sample data corresponding to the target lead-acid battery. During the training process of the source model, the weight coefficient of the source model is obtained by solving the optimization problem of the source model. Based on each source model and the weight coefficient corresponding to the source model, the source models corresponding to multiple lead-acid batteries are superimposed to obtain the target health degree evaluation model finally used to evaluate the health degree of the target lead-acid battery, which solves the problem that the model established by the prior art lacks generalization ability, improves the generalization ability of the target health degree evaluation model, and enables it to adapt to different battery data.

[0117] Embodiment III

[0118] Figure 3 FIG. is a schematic structural diagram of an evaluation device for the health status of a battery provided in Embodiment III of the present invention. As Figure 3 shown, the device includes:

[0119] A candidate feature determination module 310, configured to determine candidate features related to the health status of the lead-acid battery according to the incremental capacity expression of the lead-acid battery;

[0120] A target feature determination module 320, configured to determine the importance degree of the candidate features by the random forest algorithm, and determine target features related to the health status of the lead-acid battery based on the importance degree;

[0121] A health status evaluation module 330, configured to construct a source model of the lead-acid battery based on the training data to be used corresponding to the target features, and determine the weight coefficient of the source model. Based on the source model and the weight coefficient, the health status of the target lead-acid battery is evaluated;

[0122] Wherein, the source model is a support vector regression model.

[0123] In the technical solution of the embodiment of the present invention, according to the incremental capacity expression of the lead-acid battery, candidate features related to the health state of the lead-acid battery are determined; the importance degree of the candidate features is determined through the random forest algorithm, and target features related to the health state of the lead-acid battery are determined based on the importance degree; a source model of the lead-acid battery is constructed based on the training data to be used corresponding to the target features, and the weight coefficient of the source model is determined. Based on the source model and the weight coefficient, the health state of the target lead-acid battery is evaluated. By using the random forest algorithm to select target features that have a greater impact on the health state of the lead-acid battery as the evaluation basis for the health state of the lead-acid battery, the problem that most of the existing SOH estimation methods still require a large amount of battery aging data is solved, and the health state evaluation of the target lead-acid battery based on a small amount of data is realized.

[0124] Based on the above device, the candidate feature determination module 310 includes:

[0125] A curve drawing module, configured to establish an incremental capacity expression of the lead-acid battery based on the battery capacity and the battery voltage, and draw an incremental capacity curve of the lead-acid battery based on the incremental capacity expression;

[0126] A candidate feature extraction module, configured to determine candidate features related to the health state of the lead-acid battery based on the incremental capacity curve.

[0127] Based on the above device, the candidate features include:

[0128] At least one of the voltage, peak area, peak value, incremental variance, fixed voltage increment difference, constant current charging time, maximum slope of constant current charging, curve slope at the end of constant current charging, curve slope of constant current charging, curve area of constant current charging, voltage change amount in a preset time interval, average voltage, voltage standard deviation, voltage mean absolute deviation, maximum voltage difference, voltage skewness, and voltage kurtosis corresponding to the incremental capacity curve.

[0129] Based on the above device, the target feature determination module 320 includes:

[0130] A target random forest model establishment module, configured to determine a training data set corresponding to the random forest algorithm according to the candidate features, and train a random forest model to be used based on the training data set to obtain a target random forest model;

[0131] A node importance analysis module, configured to perform importance analysis on each node in the random forest tree based on the target random forest model to obtain the reduction amount of impurity before and after splitting of each node;

[0132] A feature importance determination module, configured to determine the importance degree of the to-be-selected feature based on the impurity reduction amount of the node corresponding to the to-be-selected feature.

[0133] Based on the above device, the number of to-be-selected features is at least two, and the target feature determination module 320 includes:

[0134] A to-be-selected feature sorting module, configured to sort at least two to-be-selected features based on the importance degree, and select the target feature from at least two to-be-selected features based on the sorting result.

[0135] Based on the above device, the health status evaluation module 330 includes:

[0136] A to-be-used training data partitioning module, configured to partition the to-be-used training data corresponding to the target feature into a to-be-used training set and a to-be-used test set; wherein, the to-be-used training data includes the sample feature values and sample health degree values corresponding to the target feature.

[0137] A source model establishment module, configured to train and test the to-be-trained source model through the to-be-used training set and the test set respectively according to the minimization problem of the to-be-trained source model, and obtain the source model.

[0138] Based on the above device, the number of lead-acid batteries is at least two, and the health status evaluation module 330 includes:

[0139] A source model retraining module, configured to establish a source model for each lead-acid battery and train each source model based on the target sample data corresponding to the target lead-acid battery; wherein, the target sample data includes the target sample feature values and target sample health degree values corresponding to the target feature in the target lead-acid battery.

[0140] A weight coefficient determination module, configured to obtain the weight coefficient of the source model by solving the optimization problem of the source model during the training process of the source model.

[0141] Based on the above device, the health status evaluation module 330 includes:

[0142] An evaluation model determination module, configured to perform weighting based on each source model and the weight coefficient corresponding to the source model to obtain the target health degree evaluation model of the target lead-acid battery.

[0143] An evaluation output module, configured to use the to-be-evaluated feature value corresponding to the target feature in the target lead-acid battery as the input of the target health degree evaluation model, and obtain the target health degree evaluation model to output the health status value of the target lead-acid battery.

[0144] The battery health state evaluation device provided by the embodiments of the present invention can execute the battery health state evaluation method provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the method.

[0145] Embodiment 4

[0146] Figure 4 FIG. shows a schematic structural diagram of an electronic device that can be used to implement the embodiments of the present invention. The electronic device is intended to represent various forms of digital computers, such as, for example, laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as, for example, personal digital processors, cellular telephones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0147] As Figure 4 shown, the electronic device 10 includes at least one processor 11, and a memory communicatively connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc. The memory stores a computer program executable by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other via a bus 14. The input / output (I / O) interface 15 is also connected to the bus 14.

[0148] Multiple components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0149] The processor 11 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the method for evaluating the battery health status.

[0150] In some embodiments, the method for evaluating the battery health status can be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as the storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the method for evaluating the battery health status described above can be executed. Alternatively, in other embodiments, the processor 11 can be configured to execute the method for evaluating the battery health status in any other suitable manner (e.g., by means of firmware).

[0151] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special-purpose or general-purpose programmable processor, receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0152] The computer program for implementing the method of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to the processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowchart and / or block diagram are implemented. The computer program can be executed entirely on the machine, partially on the machine, as a stand-alone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0153] In the context of the present invention, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, the computer-readable storage medium can be a machine-readable signal medium. More specific examples of the machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0154] To provide for interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds of devices can also be used to provide for interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0155] The systems and techniques described herein can be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer having a graphical user interface or a web browser through which the user can interact with an implementation of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0156] A computing system may include a client and a server. The client and the server are generally far from each other and usually interact via a communication network. The client-server relationship is created by computer programs running on respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system, and solves the defects of difficult management and weak business scalability existing in traditional physical hosts and VPS services.

[0157] It should be understood that various forms of processes shown above can be used, with steps reordered, added or deleted. For example, the steps described in the present invention can be executed in parallel, sequentially or in a different order, as long as the desired results of the technical solution of the present invention can be achieved, and no limitation is made herein.

[0158] The above specific embodiments do not constitute a limitation on the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for evaluating battery health status, characterized in that: include: Determining, according to an incremental capacity expression of a lead-acid battery, a candidate feature related to a health state of the lead-acid battery; Determine the importance of the candidate feature by using a random forest algorithm, and determine the target feature related to the health status of the lead-acid battery based on the importance; Building a source model of the lead-acid battery based on the to-be-used training data corresponding to the target feature, where the number of the lead-acid batteries is at least two, and determining a weight coefficient of the source model, and evaluating the health status of the target lead-acid battery based on the source model and the weight coefficient; Wherein, the source model is a support vector regression model; The importance of the candidate features is determined by a random forest algorithm, including: Determine a training data set corresponding to the random forest algorithm according to the selected features, train the random forest model to be used based on the training data set, and obtain a target random forest model; Based on the target random forest model, an importance analysis is performed on each node in the random forest tree to obtain the reduction in impurity before and after the split of each node; Determining the importance of the feature to be selected based on the reduction in impurity of the node corresponding to the feature to be selected; Among them, for multiple batteries, the sum of the importance weights of all the selected features is normalized to: in, represents the importance of the jth feature in the i-th battery, fj represents the sum of the reduction of impurity of feature j on all trees in the random forest; Define a new metric to evaluate the comprehensive performance of each feature; Where N is the number of model batteries; is the total importance of the jth feature; Among them, the linear function is used as the kernel function in the support vector regression model; The expression of the linear kernel function is as follows: 。 2. The method according to claim 1, characterized in that According to the incremental capacity expression of the lead-acid battery, the candidate features related to the health state of the lead-acid battery are determined, including: Establishing an incremental capacity expression of the lead-acid battery based on the battery capacity and the battery voltage, and drawing an incremental capacity curve of the lead-acid battery based on the incremental capacity expression; Based on the incremental capacity curve, a candidate feature associated with a state of health of the lead-acid battery is determined.

3. The method according to claim 2, characterized in that The features to be selected include: The incremental capacity curve corresponds to at least one of the voltage, peak area, peak value, incremental variance, fixed voltage incremental difference, constant current charging time, maximum slope of constant current charging, curve slope at the end of constant current charging, curve slope of constant current charging, curve area of ​​constant current charging, voltage change in a preset time interval, average voltage, voltage standard deviation, voltage mean absolute deviation, maximum voltage difference, voltage skewness and voltage kurtosis.

4. The method according to claim 1, characterized in that The number of the features to be selected is at least two, and the target features related to the health status of the lead-acid battery are determined based on the importance level, including: Based on the importance degree, at least two of the candidate features are sorted, and the target feature is selected from the at least two candidate features based on the sorting result.

5. The method according to claim 1, characterized in that Constructing a source model of the lead-acid battery based on the to-be-used training data corresponding to the target feature, including: The training data to be used corresponding to the target feature is divided into a training set to be used and a test set to be used; wherein the training data to be used includes a sample feature value and a sample health value corresponding to the target feature; According to the minimization problem of the source model to be trained, the source model to be trained is trained and tested respectively by using the training set to be used and the test set to be used to obtain the source model.

6. The method according to claim 5, characterized in that The number of the lead-acid batteries is at least two, and determining the weight coefficient of the source model includes: Establishing a source model for each of the lead-acid batteries, and training each of the source models based on target sample data corresponding to the target lead-acid battery; wherein the target sample data includes a target sample feature value and a target sample health value corresponding to the target feature in the target lead-acid battery; In the process of training the source model, the weight coefficient of the source model is obtained by solving the optimization problem of the source model.

7. The method according to claim 6, characterized in that Based on the source model and the weight coefficient, the health status of the target lead-acid battery is evaluated, including: Weighting is performed based on each of the source models and the weight coefficient corresponding to the source model to obtain a target health assessment model of the target lead-acid battery; The characteristic value to be evaluated corresponding to the target characteristic in the target lead-acid battery is used as the input of the target health assessment model to obtain the target lead-acid battery health status value output by the target health assessment model.

8. A battery health status assessment device, characterized in that: include: A candidate feature determination module, used to determine the candidate features related to the health status of the lead-acid battery according to the incremental capacity expression of the lead-acid battery; A target feature determination module, used to determine the importance of the candidate feature by using a random forest algorithm, and determine the target feature related to the health status of the lead-acid battery based on the importance; A health status assessment module, configured to construct a source model of the lead-acid battery based on the to-be-used training data corresponding to the target feature, wherein the number of the lead-acid batteries is at least two, and determine a weight coefficient of the source model, and assess the health status of the target lead-acid battery based on the source model and the weight coefficient; Wherein, the source model is a support vector regression model; The importance of the candidate features is determined by a random forest algorithm, including: Determine a training data set corresponding to the random forest algorithm according to the selected features, train the random forest model to be used based on the training data set, and obtain a target random forest model; Based on the target random forest model, an importance analysis is performed on each node in the random forest tree to obtain the reduction in impurity before and after the split of each node; Determining the importance of the feature to be selected based on the reduction in impurity of the node corresponding to the feature to be selected; Among them, for multiple batteries, the sum of the importance weights of all the selected features is normalized to: in, represents the importance of the jth feature in the i-th battery, fj represents the sum of the reduction of impurity of feature j on all trees in the random forest; Define a new metric to evaluate the comprehensive performance of each feature; Where N is the number of model batteries; is the total importance of the jth feature; Among them, the linear function is used as the kernel function in the support vector regression model; The expression of the linear kernel function is as follows: 。 9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can perform the battery health status assessment method according to any one of claims 1 to 7.

Images