CN110045288A - A kind of capacity of lithium ion battery On-line Estimation method based on support vector regression - Google Patents

Abstract

The invention relates to an online capacity estimation method of lithium ion battery based on support vector regression. The method includes the following steps: S1. Perform a cycle life test on the lithium-ion battery, and obtain a corresponding DC equivalent internal resistance spectrum; S2. Extract a potential health factor that can reflect the performance degradation of the lithium-ion battery based on the DC equivalent internal resistance spectrum. It performs correlation analysis; S3. Constructs a capacity estimation model based on the support vector regression algorithm; S4. Obtains the current charging data of the battery to be estimated and the charging equivalent internal resistance curve; S5. According to the established capacity estimation model, the extracted health The factor parameter value determines the current capacity of the battery. The method solves the problem of on-line estimation of the capacity of lithium-ion batteries under different cycle conditions, and has high estimation accuracy and strong adaptability.

Description

An online estimation method of lithium-ion battery capacity based on support vector regression

technical field

The present application relates to the field of battery management and battery state analysis, in particular to an online estimation method of lithium-ion battery capacity based on support vector regression.

Background technique

Lithium-ion batteries are widely used in electric vehicles due to their high energy density, long cycle life, and high safety. As the number of cycles of a lithium-ion battery increases during use, its external characteristics will deteriorate in all aspects, such as a decrease in effective capacity and an increase in charge-discharge internal resistance.

The capacity estimation of lithium-ion batteries is one of the core issues of battery management. It is of great significance in judging the current deterioration state of the battery, estimating the remaining service life of the battery, and avoiding premature battery failure. However, in practical applications, the capacity of the battery is generally obtained by an offline test method of fully discharging the battery, and it is impossible to update the real-time capacity change of the battery.

SUMMARY OF THE INVENTION

In view of the above problems, the present invention provides an online capacity estimation method of lithium ion battery based on support vector regression, which can estimate the effective capacity of the battery in real time by obtaining the equivalent internal resistance value of battery charging online as the input of the capacity estimation model.

The technical scheme of the present invention comprises the following steps:

S1. Perform a cycle life test on the lithium-ion battery to obtain the corresponding DC equivalent internal resistance spectrum.

(1) Cyclic charge-discharge test for lithium-ion batteries. Set the experimental conditions for the cycle life test, and use a fixed ambient temperature and a fixed charge-discharge rate to charge and discharge the battery in an uninterrupted cycle.

(2) Evaluate lithium-ion batteries. After each cycle for a period of time, the cycle is suspended, and the capacity evaluation and charging equivalent internal resistance evaluation are performed on the battery to obtain the effective capacity value of the battery. , charging equivalent internal resistance curve. Among them, the equivalent internal resistance of charging is calculated by the following formula:

(1)

in, It is the terminal voltage of the battery when it is charging and working with load, which is called the charging working voltage. is the terminal voltage of the battery after it has been fully shelved, called the charging open circuit voltage, It is the current size monitored in real time during the charging process.

(3) Determine the termination condition of the cycle life test. remember is the nominal capacity of the battery. If the ratio of the effective capacity of the battery to the nominal capacity If it is less than 0.7, stop the battery cycle life test, otherwise, return to step (1) to continue the test.

Through the cyclic charge-discharge test in step (1), the deterioration of the lithium-ion battery is accelerated, and through the evaluation test in step (2), the effective capacity value and the charging equivalent internal resistance spectrum under the corresponding degree of deterioration are obtained. The resistance spectrum is the curve of the battery state of charge on the abscissa and the equivalent internal resistance on the ordinate. By completing the cycle life test, a series of internal resistance spectra of lithium-ion batteries in the whole life cycle composed of multiple charging equivalent internal resistance curves were obtained.

S2. Extract potential health factors that can reflect the performance degradation of lithium-ion batteries based on DC equivalent internal resistance spectrum, and perform correlation analysis on them.

(1) Using the method of linear interpolation, starting from the state of charge of 0, the equivalent internal resistance of the battery samples under different deterioration states is extracted every 10%, and 9 potential health factors are obtained.

(2) Do a differential operation between the charging equivalent internal resistance and the state of charge, obtain the charging equivalent internal resistance increment curve, extract the peak value of the internal resistance increment and its peak point of the battery samples under different deterioration states, and obtain 2 potential health factor.

(3) Select the amount of capacity attenuation As a measure of battery deterioration, capacity fade is defined as follows:

(2)

(3)

in, is the nominal capacity of the battery, is the effective capacity value of the battery under a certain cycle, It is a measure to characterize the deterioration state of the battery.

(4) Using the Spearman correlation coefficient to compare the 11 potential health factors and deterioration indicators obtained in the above steps (1) and (2) Do a correlation analysis.

S3. Build a capacity estimation model based on the support vector regression algorithm.

According to the correlation analysis result in step S2, select the health factor with the correlation coefficient greater than 0.8 as the input quantity, select the effective capacity value under the corresponding deterioration degree as the output quantity, and use the support vector regression algorithm to construct a nonlinear regression model for capacity estimation.

S4. Obtain the current charging data and the charging equivalent internal resistance curve of the battery to be estimated, and the charging method used is the same as that in step S1.

S5. According to the capacity estimation model established in step S3, the current capacity of the battery is determined by the extracted health factor parameter value.

According to the nonlinear regression model established in step S3, the health factor parameter value extracted in step S4 is used as the input, and the output effective capacity value is obtained through model calculation, which is the current capacity of the battery.

Beneficial effects of the present invention: Compared with the prior art, the method of the present invention establishes a capacity modeling based on the support vector regression algorithm for the lithium ion battery under different cycle conditions, and realizes the health status of the lithium ion battery. Online estimation; support vector regression, compared with other algorithms, is a machine learning that specializes in solving limited sample situations, and can solve high-dimensional problems at the same time. It has the advantages of low computational complexity, low generalization error, and easy interpretation. The invention can monitor the aging state only through the external characteristics of the battery without knowing the internal characteristics of the battery, and has strong operability.

Description of drawings

Below in conjunction with accompanying drawing and embodiment, the present invention is further described:

Fig. 1 is the overall flow chart of the present invention;

Fig. 2 is the experimental condition of cycle life test;

Figure 3 is the DC equivalent internal resistance spectrum of the lithium-ion battery;

Figure 4 is the DC equivalent internal resistance increment curve of the lithium-ion battery;

Fig. 5 is the estimation result that utilizes the support vector regression algorithm to establish the capacity estimation model;

Figure 6 shows the correlation between 11 potential health factors and deterioration indicators using the Spearman correlation coefficient. Make a correlation analysis table;

FIG. 7 is a table of performance evaluation indexes of a lithium-ion battery capacity estimation model.

Detailed ways

The present invention provides an online capacity estimation method of lithium ion battery based on support vector regression, and the overall flow chart is shown in FIG. 1 . According to an embodiment of the present invention, the following steps are specifically included:

S1. Perform a cycle life test on the lithium-ion battery to obtain the corresponding DC equivalent internal resistance spectrum.

During the cycle life test, the charge and discharge rate, discharge interval, discharge depth and temperature conditions should be controlled, and the battery should be charged and discharged in an uninterrupted cycle. For samples of the same type of lithium-ion battery under different working conditions, the charge-discharge rate is set to 0.5C, 1C; the discharge depth is set to 0.30, 0.60; the discharge interval is set to the upper section (average SOC is 85%), middle section (average SOC is 50%) ) and the lower segment (average SOC is 30%); the temperature condition is set to 40°C. The specific experimental conditions are set as shown in Figure 2.

After each cycle for a period of time, the cycle is suspended, and the capacity evaluation and charging equivalent internal resistance evaluation are performed on the battery to obtain the effective capacity value of the battery. , charging equivalent internal resistance curve. Due to the lithium-ion battery When it is less than 0.7, its performance can no longer meet the needs of electric vehicles. Therefore, when When it is less than 0.7, the cycle life test can be ended, among which, is the nominal capacity of the battery.

Through the evaluation of charging internal resistance, the equivalent internal resistance spectrum can be obtained. The internal resistance spectrum reflects that as the battery ages, the equivalent internal resistance of the battery will gradually increase, as shown in Figure 3.

Based on the equivalent internal resistance spectrum, the health factor of the battery can be extracted, including the internal resistance under different SOC, the peak value of the internal resistance increment and the peak point. Figure 4 shows the peak value and peak point of the internal resistance increment in different deterioration states of the battery.

S2. Extract potential health factors that can reflect the performance degradation of lithium-ion batteries based on the DC equivalent internal resistance spectrum, and perform correlation analysis on them.

In this step, firstly, the method of linear interpolation is used to extract the equivalent DC internal resistance of battery samples with different deterioration states at 10% SOC intervals, and a total of 9 potential health factors are obtained; Quantitative curve, extract the peak value of internal resistance increment and its peak point of battery samples in different deterioration states, this step obtains 2 potential health factors; select capacity decay as a measure of battery deterioration degree; use Spearman correlation coefficient to compare the 11 obtained above Potential health factors and volume decay Do a correlation analysis, as shown in Figure 6.

S3. According to the result of the correlation analysis in step S2, select the health factor related to the capacity, and use the battery training sample to construct a capacity estimation model based on the support vector regression algorithm.

In this example, features with a correlation coefficient greater than 0.8 are selected: R(SOC=0.7, R(SOC=0.8), the peak value of the internal resistance increment and the peak point as the input of the model. Compared with other algorithms, support vector regression is a special solution to Machine learning with limited samples can solve high-dimensional problems at the same time, and has the advantages of low computational complexity, low generalization error, and easy interpretation. At the same time, the degree of deterioration of lithium-ion batteries is not linear with the change of internal resistance in different working environments. , the use of the support vector regression algorithm can obtain the deterioration of the lithium-ion battery without knowing the change of the internal characteristics of the battery.

S4. Obtain the current charging data and the charging equivalent internal resistance curve of the battery to be estimated, and the charging method used is the same as that in step S1.

S5. According to the capacity estimation model established in step S3, the current capacity of the battery is determined by the extracted health factor parameter value. Figure 5 shows the support vector regression results of selecting 70% of the samples as the training set and 30% of the samples as the test set. Figure 7 shows the performance evaluation index of the lithium-ion battery capacity estimation model. The correlation error is small and the estimation accuracy is high. The estimation method is practical and reliable.

Claims (6)

1. A lithium ion battery capacity online estimation method based on support vector regression is characterized by comprising the following steps:

s1, carrying out cycle life test on a lithium ion battery to obtain a corresponding direct current equivalent internal resistance spectrum;

s2, extracting potential health factors capable of reflecting the performance degradation of the lithium ion battery based on the direct current equivalent internal resistance spectrum, and performing correlation analysis on the potential health factors;

s3, constructing a capacity estimation model based on a support vector regression algorithm;

s4, acquiring the current charging data and the equivalent internal resistance curve of the battery to be estimated, wherein the charging method is the same as that in the step S1;

and S5, determining the current capacity of the battery according to the capacity estimation model established in the step S3 and the extracted health factor parameter value.

2. The online lithium ion battery capacity estimation method based on support vector regression according to claim 1, characterized in that: the specific steps of performing the cycle life test on the lithium ion battery in the step S1 and obtaining the corresponding direct current equivalent internal resistance spectrum include:

(1) carrying out cyclic charge and discharge tests on the lithium ion battery, setting experimental conditions of the cyclic life tests, and carrying out uninterrupted cyclic charge and discharge on the battery by utilizing fixed environment temperature and fixed charge and discharge multiplying power;

(2) evaluating the lithium ion battery, pausing the cycle after each cycle for a period of time, and carrying out capacity evaluation and charging equivalent internal resistance evaluation on the battery to obtain the effective capacity value of the batteryAnd a charging equivalent internal resistance curve, wherein the charging equivalent internal resistance is calculated by the following formula:

（1）

wherein,the voltage of the battery at the time of charging and loading operation, called charging operation voltage,the voltage at the terminal of the battery after sufficient rest, referred to as the charge open circuit voltage,to be chargedMonitoring the magnitude of the current in real time in the process;

(3) judging the end condition of the cycle life test and recordingThe nominal capacity of the battery is determined by the ratio of the effective capacity value of the battery to the nominal capacityIf the current battery life is less than 0.7, stopping the battery cycle life test, otherwise, returning to the step (1) to continue the test;

through the cyclic charge and discharge test in the step (1), the lithium ion battery is accelerated to deteriorate, and through the evaluation test in the step (2), an effective capacity value and a charging equivalent internal resistance spectrum under the corresponding deterioration degree are obtained, wherein the charging equivalent internal resistance spectrum is a curve with the abscissa as the battery charge state and the ordinate as the charging equivalent internal resistance; and (4) obtaining the lithium ion battery full life cycle internal resistance spectrum series consisting of a plurality of charging equivalent internal resistance curves by completing the cycle life test.

3. The online lithium ion battery capacity estimation method based on support vector regression according to claim 1, characterized in that: the step S2 is to extract potential health factors capable of reflecting the performance degradation of the lithium ion battery based on the dc equivalent internal resistance spectrum, and the specific steps of performing correlation analysis on the potential health factors include:

(1) extracting the equivalent internal resistance of the battery sample in different degradation states every 10% from the charge state of 0 by using a linear interpolation method to obtain 9 potential health factors;

(2) performing differential operation on the charging equivalent internal resistance and the charge state to obtain a charging equivalent internal resistance increment curve, extracting internal resistance increment peak values and peak value points of battery samples in different degradation states, and obtaining 2 potential health factors;

(3) selecting capacity fadeAs a measure of the degree of battery degradation, the capacity fade is defined as follows:

（2）

（3）

wherein,is the nominal capacity of the battery and,is the value of the effective capacity of the battery at a certain cycle,is a measure for representing the degradation state of the battery;

(4) utilizing Spearman correlation coefficient to carry out on 11 potential health factors and degradation indexes obtained in the steps (1) and (2)And (5) performing correlation analysis.

4. The online lithium ion battery capacity estimation method based on support vector regression according to claim 1, characterized in that: the step of constructing the capacity estimation model based on the support vector regression algorithm in step S3 is as follows: according to the correlation analysis result in the step S2, selecting a health factor having a correlation coefficient greater than 0.8 as an input quantity, selecting an effective capacity value at a corresponding degradation degree as an output quantity, and constructing a nonlinear regression model for capacity estimation by using a support vector regression algorithm.

5. The online lithium ion battery capacity estimation method based on support vector regression according to claim 1, characterized in that: in the step S4, the current charging data and the charging equivalent internal resistance curve of the battery to be estimated are obtained, and the charging method and the charging equivalent internal resistance calculation formula are the same as those in the step S1.

6. The online lithium ion battery capacity estimation method based on support vector regression according to claim 1, characterized in that: the specific steps of step S5 are: according to the nonlinear regression model established in step S3, the value of the health factor parameter extracted in step S4 is used as an input, and an output effective capacity value, that is, the current capacity of the battery, is obtained through model calculation.