CN112834927A - Lithium battery residual life prediction method, system, device and medium - Google Patents
Lithium battery residual life prediction method, system, device and medium Download PDFInfo
- Publication number
- CN112834927A CN112834927A CN202110014352.2A CN202110014352A CN112834927A CN 112834927 A CN112834927 A CN 112834927A CN 202110014352 A CN202110014352 A CN 202110014352A CN 112834927 A CN112834927 A CN 112834927A
- Authority
- CN
- China
- Prior art keywords
- support vector
- regression model
- charge
- vector regression
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/378—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC] specially adapted for the type of battery or accumulator
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/392—Determining battery ageing or deterioration, e.g. state of health
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
Abstract
The invention discloses a method, a system, equipment and a medium for predicting the residual life of a lithium battery, wherein the method comprises the following steps: performing a lithium battery charge-discharge cycle experiment, collecting voltage data and cycle times in each charge-discharge process to establish a charge-discharge curve corresponding to each charge-discharge cycle experiment, and extracting at least three dynamic characteristic data from each charge-discharge curve; performing weighting enhancement processing on the extracted three dynamic characteristic data to form a group of data sets, and dividing the data sets into a training set, a verification set and a test set; establishing a support vector regression model, taking a training set as the input of the model, training the support vector regression model, and utilizing kernel function parameters and weighting coefficients in a verification set optimization model to iterate the support vector regression model to obtain a target support vector regression model; and (4) realizing the life prediction by utilizing a target support vector regression model according to the test set. The invention obtains higher prediction precision and better robustness.
Description
Technical Field
The invention relates to the field of lithium battery residual life prediction, in particular to a method, a system, equipment and a medium for predicting the residual life of a lithium battery.
Background
The lithium ion battery has the advantages of light weight, high energy density, high efficiency, excellent low-temperature performance, low self-discharge rate, long service life and the like, and is widely applied to various fields of airplanes, electric automobiles, spacecrafts and the like. However, the performance of the lithium ion battery is degraded with time, and the remaining life is an implicit state quantity in the power battery and cannot be directly measured and estimated. Therefore, establishing an accurate residual Life prediction (RUL) model is beneficial to the health management of the battery.
The lithium battery residual life prediction method has various methods, such as an equivalent circuit model, an artificial neural network, an AR model and the like. The equivalent circuit model is constructed by analyzing a large amount of state data through an expert with special knowledge and based on the system working principle, and the system dynamic characteristics are equivalent or approximate. However, in the approximation process, the implicit relationship that some parameters in the battery have decisive effects on the system characteristics may be ignored, it is difficult to consider all the complex external conditions, so that the description capability of the model on the battery characteristics is weak, and due to the accumulation of noise such as model errors and measurement errors, the robustness in the middle and later periods of prediction is easily deteriorated. The artificial neural network is an artificial intelligent network system formed by connecting a plurality of neurons with each other according to a certain rule, and has the characteristics of strong self-organization and self-learning capabilities. However, the network structure is difficult to determine, only the point estimation value of the RUL prediction can be given, uncertain expression capability of the prediction result is not provided, and the robustness of the neural network method is limited by the advantages and disadvantages of a specific algorithm and the selection of evaluation indexes. The AR model is used for obtaining a predicted value of the system state at the current moment based on the recorded results of the system state at a plurality of past moments, and establishing a model to estimate the future state of the system. However, the confidence interval of the prediction result of the AR model is large, and the AR model cannot combine any physical information by only focusing on the change characteristics of the input data itself, so that the prediction result is easily violated from the characteristics of the actual system, and the AR model has the problems of low prediction accuracy and poor robustness.
Disclosure of Invention
In view of the above-mentioned shortcomings of the prior art, an object of the present invention is to provide a method, a system, a device and a medium for predicting remaining life of a lithium battery, which are used to solve the problem in the prior art that the performance of a lithium battery is degraded with the passage of time, and the remaining life is an implicit state quantity in a power battery and cannot be directly measured and estimated.
To achieve the above and other related objects, the present invention provides a method, system, device and medium for predicting remaining life of a lithium battery, wherein the method comprises:
performing a lithium battery charge-discharge cycle experiment, collecting voltage data and cycle times in each charge-discharge process to establish a charge-discharge curve corresponding to each charge-discharge cycle experiment, and extracting at least three dynamic characteristic data from each charge-discharge curve;
performing weighting enhancement processing on three dynamic characteristic data extracted from each charge and discharge curve to form a group of data sets, and dividing the data sets into a training set, a verification set and a test set;
establishing a Support vector Regression model, taking the training set as the input of the Support Vector Regression (SVR) model, training the SVR model, optimizing kernel function parameters and weighting coefficients in the SVR model by using the verification set, and iterating the SVR model to obtain a target SVR model;
and according to the test set, utilizing the target support vector regression model to realize life prediction.
In an embodiment of the present invention, the three dynamic characteristic data are a constant current charging time, a constant voltage charging time and a time consumed for the voltage to decrease from 3.5V to 3.3V during the discharging process, wherein the constant current charging time is a time consumed during the charging process with the constant current, and the constant voltage charging time is a time consumed during the charging process with the constant voltage until the charging current decreases to 20 mA.
In one embodiment of the present invention, the interval between each acquisition of the dynamic characteristic data is 1 second.
In an embodiment of the present invention, the performing the weighted enhancement processing on the three dynamic feature data includes:
carrying out standardization processing on the three dynamic characteristic data;
and performing weighting enhancement processing on the dynamic characteristic data subjected to the standardization processing.
In an embodiment of the present invention, the support vector regression model establishing process is:
establishing an initial model of
And taking the training set as the input of the initial model, wherein the constraint conditions are as follows:
where C is a regularization parameter, ξiAnd xii' is the relaxation factor, [ epsilon ] is the soft interval, [ omega ] is the weight vector, [ omega ] isTIs the transpose of ω, b is the offset, yiIs the target model, n is the cycle number;
solving an inequality planning problem for the initial model by using a Lagrange dual method, and obtaining the target model as follows:
wherein beta isi' and betaiIs a parameter, K (x), obtained by solving a planning problemi,xj)＝(γxi Txj+r)dIs a polynomial kernel function, gamma and r represent kernel parameters of the kernel function, d represents the order of the polynomial, xiAnd xjFeature vectors, x, representing the input training seti TDenotes the transpose of the feature vector, j is 1,2, …, n.
In one embodiment of the present invention, the ratio of the training set, the validation set, and the test set is 6:2: 2.
In an embodiment of the present invention, the verification set is used to optimize kernel function parameters and weighting coefficients in the support vector regression model, and the optimization method is to perform optimization by using a differential evolution algorithm.
In one embodiment of the present invention, the system comprises:
the data acquisition module is used for performing a lithium battery charge-discharge cycle experiment, acquiring voltage data and cycle times in each charge-discharge process to establish a charge-discharge curve corresponding to each charge-discharge cycle experiment, and extracting at least three dynamic characteristic data from each charge-discharge curve;
the data set dividing module is used for performing weighting enhancement processing on the three dynamic characteristic data extracted from each charge-discharge curve to form a group of data sets, and then dividing the data sets into a training set, a verification set and a test set;
the model establishing module is used for establishing a support vector regression model, taking the training set as the input of the support vector regression model, training the support vector regression model, optimizing kernel function parameters and weighting coefficients in the support vector regression model by using the verification set, and iterating the support vector regression model to obtain a target support vector regression model;
and the data analysis module is used for realizing the life prediction by utilizing the target support vector regression model according to the test set.
In one embodiment of the present invention, a processor is included, the processor coupled with a memory, the memory storing program instructions, which when executed by the processor, implement any of the above-described methods.
In an embodiment of the invention, a program is included which, when run on a computer, causes the computer to perform any of the methods described above.
According to the lithium battery remaining life prediction method, the lithium battery remaining life prediction system, the lithium battery remaining life prediction equipment and the lithium battery remaining life prediction medium, at least three dynamic characteristics are extracted from data, then weighting enhancement processing is carried out on the dynamic characteristics to form a data set, the enhanced data set is divided into a training set, a verification set and a test set, a support vector regression model is established by using a polynomial kernel function, the training model is trained by using the training set, kernel function parameters and weighting coefficients are optimized by using the verification set, and finally the generalization capability of the test set test model is utilized, so that the lithium battery remaining life prediction method obtains higher prediction precision and better robustness.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a flowchart of a method for predicting the remaining life of a lithium battery according to the present invention;
fig. 2 is a schematic diagram of a method for predicting the remaining life of a lithium battery according to the present invention;
FIG. 3 is a comparison graph of the accuracy of the prediction method in the embodiment of the present invention;
FIG. 4 is a comparison graph of robustness in an embodiment of the invention;
fig. 5 is a schematic diagram of a system for predicting the remaining life of a lithium battery according to the present invention.
Detailed Description
The embodiments of the present invention are described below with reference to specific embodiments, and other advantages and effects of the present invention will be easily understood by those skilled in the art from the disclosure of the present specification. The invention is capable of other and different embodiments and of being practiced or of being carried out in various ways, and its several details are capable of modification in various respects, all without departing from the spirit and scope of the present invention.
It should be noted that the drawings provided in the present embodiment are only for illustrating the basic idea of the present invention, and the components related to the present invention are only shown in the drawings rather than drawn according to the number, shape and size of the components in actual implementation, and the type, quantity and proportion of the components in actual implementation may be changed freely, and the layout of the components may be more complicated.
The invention discloses a method for predicting the remaining life of a lithium ion battery, which can obtain higher prediction precision and better robustness compared with the traditional prediction sorting method after being used, and as shown in figure 1, the prediction method comprises the following steps:
s1, carrying out lithium battery charge-discharge cycle experiments, collecting voltage data and cycle times in each charge-discharge process to establish charge-discharge curves corresponding to each charge-discharge cycle experiment, and extracting at least three dynamic characteristic data from each charge-discharge curve;
s2, performing weighting enhancement processing on the three dynamic characteristic data extracted from each charge-discharge curve to form a group of data sets, and dividing the data sets into a training set, a verification set and a test set;
s3, establishing a support vector regression model, taking the training set as the input of the support vector regression model, training the support vector regression model, optimizing kernel function parameters and weighting coefficients in the support vector regression model by using the verification set, and iterating the support vector regression model to obtain a target support vector regression model;
and S4, according to the test set, utilizing the target support vector regression model to realize life prediction.
As shown in fig. 1, in step S1, a lithium battery charge-discharge cycle experiment is performed, voltage data and cycle number in each charge-discharge process are collected to establish a charge-discharge curve corresponding to each charge-discharge cycle experiment, at least three dynamic characteristic data are extracted from each charge-discharge curve, in the present embodiment, the three dynamic characteristics are the constant current charging time, the constant voltage charging time and the time consumed by the voltage decreasing from 3.5V to 3.3V during the discharging process, it should be noted that, in this embodiment, 3.5V to 3.3V is selected as the fixed voltage segment, the time consumed by the test battery in this voltage segment under different cycle periods is intercepted as the third dynamic feature extracted in this embodiment, namely, the constant current charging time is the time consumed in the constant current charging process, and the constant voltage charging time is the time consumed in the constant voltage charging process until the charging current is reduced to 20 mA. In this embodiment, the interval between each time of acquiring the voltage data is 1 second, and further, in order to ensure the accuracy of the acquired and extracted data to improve the subsequent prediction accuracy, in this embodiment, a period of time, for example, 4 hours, needs to be left for standing between each charging and discharging process.
In the present embodiment, for convenience of description, the first charge/discharge experiment process is taken as an example for illustration, carrying out charge and discharge experiments on the lithium battery, collecting voltage data in the experimental process to establish a charge and discharge curve corresponding to the experiments, extracting the three dynamic characteristic data from the curve, performing weighting enhancement processing on the dynamic characteristic data to generate weighted enhanced data, and so on, establishing corresponding charge-discharge curves for each charge-discharge experiment, respectively extracting three dynamic characteristic data from the corresponding charge-discharge curves, weighting and strengthening treatment is respectively carried out, so that each charge and discharge experiment corresponds to data after weighting and strengthening treatment, these data form a data set, and it should be noted that, in order to ensure the accuracy of the model, the weighting coefficient of each data in the data set is the same.
As shown in fig. 1 and 2, in step S2, a set of data sets is formed by performing weighted enhancement processing on the three dynamic feature data extracted from each of the charge and discharge curves, and the data sets are divided into a training set, a verification set, and a test set. Wherein, the weighting enhancement processing of the three dynamic characteristic data comprises:
s21, standardizing the three dynamic characteristic data, wherein the standardization is to scale the dynamic characteristic data to be between 0 and 1;
and S22, performing weighting enhancement processing on the dynamic characteristic data after the normalization processing. And when the three dynamic characteristic data extracted from each charging and discharging curve are subjected to weighting enhancement processing, the weighting coefficients are the same.
In this embodiment, the ratio of the training set, the verification set, and the test set is 6:2:2, it should be noted that the process of dividing the training set, the verification set, and the test set is random, and the first weighting coefficient in the weighting enhancement processing process is randomly selected.
As shown in fig. 1 and 2, in step S3, a support vector regression model is established, the training set is used as an input of the support vector regression model, the support vector regression model is trained, kernel function parameters and weighting coefficients in the support vector regression model are optimized by using the verification set, and the support vector regression model is iterated to obtain a target support vector regression model.
In this embodiment, the support vector regression model establishing process is as follows:
establishing an initial model of
And taking the training set as the input of the initial model, wherein the constraint conditions are as follows:
where C is a regularization parameter, ξiAnd xii' is a relaxation factor, εIs the soft interval, ω is the weight vector, ωTIs the transpose of ω, b is the offset, yiIs the target model, n is the cycle number;
solving an inequality planning problem for the initial model by using a Lagrange dual method, and obtaining the target model as follows:
wherein beta isi' and betaiIs a parameter, K (x), obtained by solving a planning problemi,xj)＝(γxi Txj+r)dIs a polynomial kernel function, gamma and r represent kernel parameters of the kernel function, d represents the order of the polynomial, xiAnd xjFeature vectors, x, representing the input training seti TDenotes the transpose of the feature vector, j is 1,2, …, n, when i is j, xiAnd xjThe same feature vector representing the input training set; when i ≠ j, xiAnd xjDifferent feature vectors representing the input training set. It should be noted that the target model is a support regression vector model. The verification set is used for optimizing the kernel parameters and the weighting coefficients of the kernel function in the support vector regression model, the optimization method is to optimize by using a Differential Evolution (DE) algorithm, and it is to be noted that the kernel parameters of the initial polynomial kernel function are randomly selected. And continuously iterating in the optimization process to optimize the model and improve the accuracy of the residual life prediction.
The principle of the differential evolution algorithm in the step is as follows:
population initialization: m individuals are randomly and uniformly generated in a solution space, and each individual consists of an n-dimensional vector.
Xi(0)＝(xi,1(0),xi,2(0),xi,3(0),…,xi,n(0))，i＝1,2,3,…,M
The j dimension value mode of the ith individual is as follows:
Xj,i(0)＝Xj,i L+rand(0,1)(Xj,i U-Xj,i L)，i＝1,2,3,…,M，j＝1,2,3,…,n，
wherein, Xj,i UAnd Xj,i LRespectively representing the upper bound and the lower bound of the value range of the jth dimension of the ith individual. rand (0,1) represents random numbers uniformly distributed in the interval (0,1), and it should be noted that the population size M is generally between 5 Xn and 10 Xn, but not less than 4 Xn. Meanwhile, the support vector regression model is used as a fitness function, the root mean square error of a prediction result is calculated, the negative root mean square error is used as the fitness of each individual in the initial population and is marked as fit0(Xi)。
And (3) mutation process: randomly select 3 individuals from the g-th generation population: xp1(g),Xp2(g),Xp3(g) And p1 ≠ p2 ≠ p3 ≠ i, the new mutation vector is:
Vi(g+1)＝Xp1(g)+F(Xp2(g)-Xp3(g))
wherein, Deltap2,p3(g)＝Xp2(g)-Xp3(g) Is the difference vector and F is the mutation operator.
And (3) a crossing process: and performing cross operation on the g generation population, and randomly taking a number cr between 0 and 1, so that the generated population is as follows:
where CR is the crossover operator, jrandIs [1,2, …, n ]]Is a random integer of (a).
The selection process comprises the following steps: selecting individuals in the g generation population, firstly calculating a population Ui(g +1) Individual fitness, denoted as fitnessg+1(Ui) And g is 1,2,3, …, and N is the total iteration number of the algorithm. The new generation of population is:
and circularly executing the steps until the maximum iteration times are reached, obtaining the optimal individual according to the fitness recorded in each generation, taking the optimal individual as a kernel function parameter and a weighting coefficient in the support vector regression model to obtain the target support vector regression model, wherein a new generation of population is generated after initialization, variation, intersection and selection once.
As shown in fig. 1 and fig. 2, in step S4, the target support vector regression model is used to predict the remaining life of the lithium battery according to the test set.
Specifically, as shown in fig. 3, in the present embodiment, in order to compare the prediction accuracy of the method proposed by the present invention with that of other prediction methods, a comparative experiment is designed, in which an experiment using the method proposed by the present invention is referred to as experiment group 1, an experiment using gaussian process regression is referred to as experiment group 2, and an experiment using real-time particle filtering is referred to as experiment group 3. As shown in fig. 3, the method of the present invention can obtain higher prediction accuracy, and the comparison result between the Mean Absolute Percentage Error (MAPE) and the Root Mean Square Error (RMSE) is shown in fig. 3.
Referring to fig. 2 to 4, in an embodiment, in order to compare the enhancement processing in the method of the present invention to improve the robustness of the model, the following comparative experiment is designed, where an experiment using the method of the present invention is used as an experiment one, an experiment without the enhancement processing in the method of the present invention is used as an experiment two, and the experiment is compared 50 times for each method, so as to obtain an experiment result shown in fig. 4. It can be seen that a lower median error and a stronger robustness can be obtained by using the method proposed by the present invention.
As shown in fig. 5, in this embodiment, the present invention further provides a system for predicting remaining life of a lithium battery, where the prediction system 100 includes a data acquisition module 10, a data set partitioning module 20, a model building module 30, and a data analysis module 40.
As shown in fig. 5, in this embodiment, the data acquisition module 10 is configured to perform a lithium battery charge-discharge cycle experiment, acquire voltage data and cycle times in each charge-discharge process to establish a charge-discharge curve corresponding to each charge-discharge cycle experiment, and extract at least three dynamic characteristic data from each charge-discharge curve;
as shown in fig. 5, in this embodiment, the data set partitioning module 20 is configured to perform weighting enhancement processing on three dynamic feature data extracted from each of the charge and discharge curves to form a group of data sets, and then partition the data sets into a training set, a verification set, and a test set;
as shown in fig. 5, in this embodiment, the model establishing module 30 is configured to establish a support vector regression model, take the training set as an input of the support vector regression model, train the support vector regression model, optimize kernel function parameters and weighting coefficients in the support vector regression model by using the verification set, and iterate the support vector regression model to obtain a target support vector regression model;
in the present embodiment, as shown in fig. 5, the data analysis module 40 is configured to implement life prediction according to the test set and by using the target support vector regression model. In this embodiment, the prediction system 100 is connected to the test battery 200 to predict the service life of the test battery 200, and it should be noted that the prediction system 100 may be internally disposed in the test battery 200 or externally disposed with the test battery 200.
In an embodiment, a processor is included, the processor coupled with a memory, the memory storing program instructions, which when executed by the processor, implement any of the above methods.
In an embodiment of the invention, a program is included which, when run on a computer, causes the computer to perform any of the methods described above.
The invention relates to a method, a system, equipment and a medium for predicting the residual life of a lithium battery, which are used for performing charge-discharge cycle experiments on the lithium battery, acquiring voltage data and corresponding cycle times in each charge-discharge process to establish a charge-discharge curve of each charge-discharge cycle experiment, extracting at least three dynamic characteristics from the charge-discharge curve, performing weighted enhancement treatment on the dynamic characteristics to form a data set, dividing the enhanced data set into a training set, a verification set and a test set, establishing a support vector regression model by using a polynomial kernel function, performing the training model by using the training set as the input of the support vector regression model, optimizing kernel function parameters and weighting coefficients by using the verification set, and finally testing the generalization capability of the model by using the test set, so that the invention obtains higher prediction precision and better robustness.
The above description is only a preferred embodiment of the present application and a description of the applied technical principle, and it should be understood by those skilled in the art that the scope of the present application is not limited to the technical solution of the specific combination of the above technical features, and also covers other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the inventive concept, for example, the technical solutions formed by mutually replacing the above technical features (but not limited to) having similar functions disclosed in the present application.
Other technical features than those described in the specification are known to those skilled in the art, and are not described herein in detail in order to highlight the innovative features of the present invention.
Claims (10)
1. A method for predicting the residual life of a lithium battery is characterized by comprising the following steps:
performing a lithium battery charge-discharge cycle experiment, collecting voltage data and cycle times in each charge-discharge process to establish a charge-discharge curve corresponding to each charge-discharge cycle experiment, and extracting at least three dynamic characteristic data from each charge-discharge curve;
performing weighting enhancement processing on three dynamic characteristic data extracted from each charge and discharge curve to form a group of data sets, and dividing the data sets into a training set, a verification set and a test set;
establishing a support vector regression model, taking the training set as the input of the support vector regression model, training the support vector regression model, optimizing kernel function parameters and weighting coefficients in the support vector regression model by using the verification set, and iterating the support vector regression model to obtain a target support vector regression model;
and according to the test set, utilizing the target support vector regression model to realize life prediction.
2. The method as claimed in claim 1, wherein the three dynamic characteristics are a constant current charging time, a constant voltage charging time and a time consumed for a voltage to decrease from 3.5V to 3.3V during discharging, wherein the constant current charging time is a time consumed during charging with a constant current, and the constant voltage charging time is a time consumed for charging with a constant voltage until a charging current decreases to 20 mA.
3. The method as claimed in claim 1, wherein the interval between each acquisition of the voltage data is 1 second.
4. The method for predicting the remaining life of a lithium battery as claimed in claim 1, wherein the weighting enhancement processing of the three dynamic characteristic data comprises:
carrying out standardization processing on the three dynamic characteristic data;
carrying out weighting enhancement processing on the dynamic characteristic data subjected to the standardization processing;
and when the three dynamic characteristic data extracted from each charging and discharging curve are subjected to weighting enhancement processing, the weighting coefficients are the same.
5. The method for predicting the residual life of a lithium battery as claimed in claim 1, wherein the support vector regression model establishing process comprises:
establishing an initial model of
And taking the training set as the input of the initial model, wherein the constraint conditions are as follows:
where C is a regularization parameter, ξiAnd xii' is the relaxation factor, [ epsilon ] is the soft interval, [ omega ] is the weight vector, [ omega ] isTIs the transpose of ω, b is the offset, yiIs the target model, n is the cycle number;
solving an inequality planning problem for the initial model by using a Lagrange dual method, and obtaining the target model as follows:
wherein beta isi' and betaiIs a parameter, K (x), obtained by solving a planning problemi,xj)＝(γxi Txj+r)dIs a polynomial kernel function, gamma and r represent kernel parameters of the kernel function, d represents the order of the polynomial, xiAnd xjFeature vectors, x, representing the input training seti TDenotes the transpose of the feature vector, j is 1,2, …, n.
6. The method as claimed in claim 1, wherein the ratio of the training set, the validation set and the test set is 6:2: 2.
7. The method as claimed in claim 1, wherein the verification set is used to optimize kernel function parameters and weighting coefficients in the support vector regression model, and the optimization method is to optimize by using a differential evolution algorithm.
8. A system for predicting remaining life of a lithium battery, the system comprising:
the data acquisition module is used for performing a lithium battery charge-discharge cycle experiment, acquiring voltage data and cycle times in each charge-discharge process to establish a charge-discharge curve corresponding to each charge-discharge cycle experiment, and extracting at least three dynamic characteristic data from each charge-discharge curve;
the data set dividing module is used for performing weighting enhancement processing on the three dynamic characteristic data extracted from each charge-discharge curve to form a group of data sets, and then dividing the data sets into a training set, a verification set and a test set;
the model establishing module is used for establishing a support vector regression model, taking the training set as the input of the support vector regression model, training the support vector regression model, optimizing kernel function parameters and weighting coefficients in the support vector regression model by using the verification set, and iterating the support vector regression model to obtain a target support vector regression model;
and the data analysis module is used for realizing the life prediction by utilizing the target support vector regression model according to the test set.
9. A lithium battery remaining life predicting device comprising a processor coupled to a memory, the memory storing program instructions that when executed by the processor implement the method of any of claims 1 to 7.
10. A computer-readable storage medium, characterized by comprising a program which, when run on a computer, causes the computer to perform the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110014352.2A CN112834927A (en) | 2021-01-06 | 2021-01-06 | Lithium battery residual life prediction method, system, device and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110014352.2A CN112834927A (en) | 2021-01-06 | 2021-01-06 | Lithium battery residual life prediction method, system, device and medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112834927A true CN112834927A (en) | 2021-05-25 |
Family
ID=75926314
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110014352.2A Pending CN112834927A (en) | 2021-01-06 | 2021-01-06 | Lithium battery residual life prediction method, system, device and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112834927A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113205233A (en) * | 2021-06-10 | 2021-08-03 | 湖北东湖实验室 | Lithium battery life prediction method based on wolf algorithm and multi-core support vector regression |
| CN113589172A (en) * | 2021-08-12 | 2021-11-02 | 国网江苏省电力有限公司常州供电分公司 | Service life estimation method for power grid components |
| CN114814631A (en) * | 2022-04-25 | 2022-07-29 | 浙江大学 | Cloud computing and feature selection based lithium battery online life prediction method |
| CN114881316A (en) * | 2022-04-24 | 2022-08-09 | 上海玫克生储能科技有限公司 | Lithium battery residual life prediction method and system, terminal device and storage medium |
| CN116593904A (en) * | 2023-07-17 | 2023-08-15 | 中国华能集团清洁能源技术研究院有限公司 | Model training method and method for predicting battery SOH and battery RUL |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170205466A1 (en) * | 2014-12-29 | 2017-07-20 | Hefei University Of Technology | Method for predicting remaining useful life of lithium battery based on wavelet denoising and relevance vector machine |
| CN108805217A (en) * | 2018-06-20 | 2018-11-13 | 山东大学 | A kind of health state of lithium ion battery method of estimation and system based on support vector machines |
| CN110161425A (en) * | 2019-05-20 | 2019-08-23 | 华中科技大学 | A kind of prediction technique of the remaining life divided based on lithium battery catagen phase |
| CN110443377A (en) * | 2019-06-24 | 2019-11-12 | 南方电网调峰调频发电有限公司信息通信分公司 | A kind of support vector machines life of storage battery prediction technique based on immune algorithm optimization |
| CN110501646A (en) * | 2019-08-29 | 2019-11-26 | 中国人民解放军国防科技大学 | Off-line lithium battery residual capacity estimation method |
-
2021
- 2021-01-06 CN CN202110014352.2A patent/CN112834927A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170205466A1 (en) * | 2014-12-29 | 2017-07-20 | Hefei University Of Technology | Method for predicting remaining useful life of lithium battery based on wavelet denoising and relevance vector machine |
| CN108805217A (en) * | 2018-06-20 | 2018-11-13 | 山东大学 | A kind of health state of lithium ion battery method of estimation and system based on support vector machines |
| CN110161425A (en) * | 2019-05-20 | 2019-08-23 | 华中科技大学 | A kind of prediction technique of the remaining life divided based on lithium battery catagen phase |
| CN110443377A (en) * | 2019-06-24 | 2019-11-12 | 南方电网调峰调频发电有限公司信息通信分公司 | A kind of support vector machines life of storage battery prediction technique based on immune algorithm optimization |
| CN110501646A (en) * | 2019-08-29 | 2019-11-26 | 中国人民解放军国防科技大学 | Off-line lithium battery residual capacity estimation method |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113205233A (en) * | 2021-06-10 | 2021-08-03 | 湖北东湖实验室 | Lithium battery life prediction method based on wolf algorithm and multi-core support vector regression |
| CN113589172A (en) * | 2021-08-12 | 2021-11-02 | 国网江苏省电力有限公司常州供电分公司 | Service life estimation method for power grid components |
| CN114881316A (en) * | 2022-04-24 | 2022-08-09 | 上海玫克生储能科技有限公司 | Lithium battery residual life prediction method and system, terminal device and storage medium |
| CN114881316B (en) * | 2022-04-24 | 2023-06-16 | 上海玫克生储能科技有限公司 | Lithium battery remaining life prediction method, system, terminal equipment and storage medium |
| CN114814631A (en) * | 2022-04-25 | 2022-07-29 | 浙江大学 | Cloud computing and feature selection based lithium battery online life prediction method |
| CN116593904A (en) * | 2023-07-17 | 2023-08-15 | 中国华能集团清洁能源技术研究院有限公司 | Model training method and method for predicting battery SOH and battery RUL |
| CN116593904B (en) * | 2023-07-17 | 2023-10-03 | 中国华能集团清洁能源技术研究院有限公司 | Model training method and method for predicting battery SOH and battery RUL |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Sui et al. | A review of non-probabilistic machine learning-based state of health estimation techniques for Lithium-ion battery | |
| CN112834927A (en) | Lithium battery residual life prediction method, system, device and medium | |
| CN110221225B (en) | Spacecraft lithium ion battery cycle life prediction method | |
| CN111860982A (en) | Wind power plant short-term wind power prediction method based on VMD-FCM-GRU | |
| CN113805064B (en) | Lithium ion battery pack health state prediction method based on deep learning | |
| CN113064093B (en) | Method and system for jointly estimating state of charge and state of health of energy storage battery | |
| WO2022198616A1 (en) | Battery life prediction method and system, electronic device, and storage medium | |
| CN113504483B (en) | Integrated prediction method for residual life of lithium ion battery considering uncertainty | |
| CN115201686B (en) | Lithium ion battery health state assessment method under incomplete charge and discharge data | |
| CN112434848A (en) | Nonlinear weighted combination wind power prediction method based on deep belief network | |
| CN114660497A (en) | Lithium ion battery service life prediction method aiming at capacity regeneration phenomenon | |
| CN112803398A (en) | Load prediction method and system based on empirical mode decomposition and deep neural network | |
| Ding et al. | Remaining useful life prediction for lithium-ion batteries based on CS-VMD and GRU | |
| CN113917336A (en) | Lithium ion battery health state prediction method based on segment charging time and GRU | |
| CN114726045A (en) | Lithium battery SOH estimation method based on IPEA-LSTM model | |
| Cruz et al. | Neural network prediction interval based on joint supervision | |
| Li et al. | A hybrid framework for predicting the remaining useful life of battery using Gaussian process regression | |
| Mohammad et al. | Short term load forecasting using deep neural networks | |
| Mazzi et al. | Lithium-ion battery state of health estimation using a hybrid model based on a convolutional neural network and bidirectional gated recurrent unit | |
| CN111337833B (en) | Lithium battery capacity integrated prediction method based on dynamic time-varying weight | |
| CN113033898A (en) | Electrical load prediction method and system based on K-means clustering and BI-LSTM neural network | |
| Wang et al. | An efficient state-of-health estimation method for lithium-ion batteries based on feature-importance ranking strategy and PSO-GRNN algorithm | |
| Al-Hawani et al. | Short-term forecasting of electricity consumption using gaussian processes | |
| Song et al. | Capacity estimation method of lithium-ion batteries based on deep convolution neural network | |
| CN115980588A (en) | Lithium ion battery health state estimation method based on self-encoder extraction features |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |