CN111856282B - Vehicle-mounted lithium battery state estimation method based on improved genetic unscented Kalman filtering - Google Patents

Abstract

The invention provides a method for estimating the state of an on-board lithium battery based on an improved genetic unscented Kalman filter, which samples the open-circuit voltage and current of the lithium battery under the condition of constant current discharge of the lithium battery; and establishes the method based on the second-order RC equivalent circuit model. State space equation of lithium battery, calculate the estimated value of open circuit voltage; use Kalman filter algorithm to estimate and update the second-order RC equivalent circuit model in real time; use improved genetic algorithm to find the optimal noise covariance matrix; use unscented Kalman filter The Mann filter algorithm estimates the battery state through the open circuit voltage estimate to obtain the current open circuit voltage estimate; and then outputs the SOC estimate corresponding to the current open circuit voltage estimate through the relationship between the lithium battery open circuit voltage and SOC. The invention solves the problem of serious nonlinearity of the system state variables caused by the internal electrochemical reaction of the lithium battery, and improves the real-time performance and accuracy of estimation.

Description

State estimation method of vehicle lithium battery based on improved genetic unscented Kalman filter

technical field

The invention belongs to the field of on-board lithium batteries, and more particularly relates to a state estimation method for on-board lithium batteries based on improved genetic unscented Kalman filtering.

Background technique

Lithium-ion batteries are widely used in electric vehicles due to their advantages of long cycle life, low self-discharge rate, high energy, and no memory effect. In order to ensure the safe, reliable and efficient operation of lithium-ion batteries, it is necessary to accurately estimate the working state of the battery and accurately establish a battery management system. Among them, the state of charge (SOC) of the lithium-ion battery directly reflects the amount of remaining power. Only by accurately estimating the battery SOC can avoid the overcharge and overdischarge behavior of the lithium-ion battery and keep the battery in a good working state.

At present, the estimation methods of lithium-ion battery SOC at home and abroad mainly include the following: 1) Open circuit voltage method (OCV), which uses the nonlinear relationship between open circuit voltage and SOC to obtain SOC value by measuring open circuit voltage. The battery needs to be left standing for a long time, which is not suitable for online estimation of SOC; 2) Ampere-hour integration method (CC), the current is integrated under the known initial value. This method has extremely high requirements for the initial value of SOC, and at the same time The cumulative error caused by current detection and the effect of capacity decay caused by battery aging are ignored; 3) Electrochemical impedance spectroscopy (EIS), through the AC impedance spectrum to find the relationship between the ohmic-polarization internal resistance and SOC of lithium-ion batteries, This method has poor stability, complex detection, long running time, and is rarely used in real-time monitoring; 4) Kalman filter (KF), based on the equivalent model, uses observation data to correct the state estimation, and the algorithm requires model accuracy. High, not suitable for nonlinear systems; 5) Extended Kalman (EKF), by linearizing the nonlinear function, and then using Kalman filtering to complete the estimation of the target, due to the non-positive definiteness of the variance matrix in the calculation process, it will be 6) Unscented Kalman (UKF), which abandons nonlinear functions for linearization, and uses unscented transformation to deal with the nonlinear transfer problem of mean and covariance. This method has better performance on nonlinear distribution statistics. High precision; 7) Particle filter (PF), which uses discrete particle sets to approximately describe the probability density of random variables in the system, which can more accurately express the posterior probability distribution of observed quantities and control quantities; 8) Neural network method (NN) , which has strong processing ability for nonlinear systems, but needs to train a large amount of data, and the estimation error is greatly affected by the training data and training methods; 9) Sliding Mode Observation Method (SMO), which can effectively solve the problem of nonlinear model However, frequent switching of control states will cause chattering in the system.

In the actual working process of lithium-ion batteries, factors such as ambient temperature, cycle times, and detection accuracy have an important impact on the state estimation of lithium-ion batteries. In practical applications, the problem of divergence of estimated values due to nonlinearity of the system often occurs, and the phenomenon of particle scarcity occurs due to the small number of particles simply using the particle filter algorithm.

SUMMARY OF THE INVENTION

The technical problem to be solved by the present invention is to provide a method for estimating the state of a vehicle lithium battery based on an improved genetic unscented Kalman filter, so as to improve the estimation accuracy.

The technical solution adopted by the present invention to solve the above-mentioned technical problems is: a method for estimating the state of a vehicle-mounted lithium battery based on improved genetic unscented Kalman filtering, which is characterized in that: it comprises the following steps:

S1. Under the condition of constant current discharge of the lithium battery, sample the open circuit voltage and current of the lithium battery;

S2. According to the second-order RC equivalent circuit model, establish the state space equation of the lithium battery, and use the current sampled by S1 to calculate the estimated value of the open circuit voltage by using the state space equation; compare the sampled open circuit voltage and the calculated open circuit voltage estimate value;

S3, using the Kalman filter algorithm to estimate the second-order RC equivalent circuit model and update it in real time through the open-circuit voltage estimated value and the measured value calculated by S2;

S4. Through constant current charging and discharging of lithium batteries with different SOC values, the open circuit voltage is measured after a period of rest, and the relationship between the open circuit voltage and SOC of the lithium battery is obtained by fitting;

S5, using the improved genetic algorithm to seek the optimal noise covariance matrix;

S6. Using the optimal noise covariance matrix obtained in S5, using the unscented Kalman filter algorithm, the battery state is estimated by the open circuit voltage estimated value obtained by S2, and the current open circuit voltage estimated value is obtained; then the lithium battery is open circuit obtained through S4. The relationship between voltage and SOC, and output the SOC estimate value corresponding to the current open circuit voltage estimate value.

According to the above method, the S1 specifically includes:

Select vehicle-mounted lithium battery packs with the same consistency to perform charge and discharge experiments at room temperature, and select the average voltage value of lithium batteries as valid data;

To prevent over-discharge of lithium batteries, set the discharge cut-off voltage;

Set the constant current discharge rate of the lithium battery, and sample the open circuit voltage and current of the lithium battery with a fixed sampling period.

According to the above method, the S3 is specifically:

3-1. According to the second-order RC equivalent circuit model of the lithium battery, the system transfer function is obtained by Kirchhoff's law;

3-2. Perform bilinear transformation on the transfer function, and select the state equation and the observation equation under the condition of a fixed sampling period;

3-3. The state variables are estimated online by the Kalman filter algorithm, and then the inverse transformation of z is used to obtain the identification values of each parameter in the second-order RC equivalent circuit model.

According to the above method, the S4 is specifically:

4-1. Fully discharge/charge the lithium battery and let it stand for a period of time;

4-2. Then carry out the constant current pulse charge/discharge experiment at equal intervals on the lithium battery, let it stand after each pulse experiment, and measure the terminal voltage of the lithium battery;

4-3. Obtain the average voltage of the lithium battery pack at each stage through multiple experiments;

4-4. Obtain the average voltage of lithium battery packs with the same consistency through experiments as the open circuit voltage of a single lithium battery, and fit the function.

According to the above method, the S5 is specifically:

5-1. Select the chromosome encoding method, and randomly generate the initial population according to the selected encoding method;

5-2. Judging whether the convergence conditions are met, generally judge whether the number of iterations has been reached, if it is satisfied, execute 5-6, but not execute step 5-3;

5-3. Use inverse dichotomy to select crossover individuals;

5-4. Use the improved crossover probability algorithm to perform crossover operation on the selected crossover individuals.

5-5. Select mutation and return to 5-2;

5-6. Output the optimal solution.

According to the above method, the S6 specifically includes:

6-1. Analyze the electrochemical process of the battery, and list the state equation and measurement equation of the system;

6-2. Calculation of initial value of each state quantity;

6-3. Establish Sigma point;

6-4. Update the state equation;

6-5. Update the measurement equation;

6-6. Repeat steps 6-2 to 6-5 above.

The beneficial effects of the present invention are:

1. By using S5 and S6, the system noise and observation noise covariance matrices _Qw and _Rv can be better corrected in each iteration process to obtain stable estimation results, and the unscented Kalman filter algorithm is better. The problem of serious nonlinearity of the system state variables caused by the electrochemical reaction inside the lithium-ion battery is solved, so that the SOC estimation accuracy is greatly improved compared with the traditional method.

2. Since step S5 is adopted, the inverse dichotomy method is used to select the crossover individuals, and a new probability algorithm is used to distinguish the improved genetic algorithm of the better individuals and the worse individuals, thereby reducing the blindness of the genetic algorithm operation, which is beneficial to Retention of excellent genes and elimination of poor genes.

Description of drawings

FIG. 1 is a flow chart of the present invention.

Figure 2 is a schematic diagram of a second-order RC equivalent circuit model.

Figure 3 is a parameter identification curve diagram.

FIG. 4 is a graph showing the relationship between open circuit voltage and SOC.

Figure 5 is a flowchart of the improved genetic algorithm.

FIG. 6 is a graph showing the relationship between charge and discharge current and time under pulse charge and discharge current test conditions.

FIG. 7 is a graph of SOC estimation under different initial value settings.

Figure 8 is a graph of the SOC estimation error under different initial values.

Fig. 9 is the current curve diagram of UDDS cycle working condition.

Figure 10 is a graph of SOC estimation curves for different algorithms.

Figure 11 is a graph of SOC estimation error curves for different algorithms.

Detailed ways

The present invention will be further described below in conjunction with specific examples.

The present invention provides a vehicle-mounted lithium battery state estimation method based on improved genetic unscented Kalman filtering, as shown in Figure 1, which includes the following steps:

S1. Under the condition of constant current discharge of the lithium battery, sample the open circuit voltage and current of the lithium battery.

In this embodiment, S1 specifically includes the following sub-steps:

(1-1) Select the lithium iron phosphate battery pack with the same consistency and the rated capacity of the battery cell of 40Ah at room temperature, carry out the charge and discharge experiment, and select the average voltage value of the lithium battery as the valid data;

(1-2) Prevent the lithium battery from over-discharging, and set the discharge cut-off voltage through software;

(1-3) Program the electronic load to realize the constant current discharge rate of 1/5C of the battery, and sample the open-circuit voltage and current of the lithium battery with 1s as the sampling period.

S2. According to the second-order RC equivalent circuit model, the state space equation of the lithium battery is established. Using the current sampled by S1, the state space equation is used to calculate the estimated value of the open circuit voltage; compare the sampled open circuit voltage with the calculated open circuit voltage estimated value.

Specifically, it includes the following sub-steps:

(2-1) First, establish a second-order RC equivalent circuit model based on current and charge-discharge direction as shown in Figure 2. The model includes a voltage-controlled voltage source U _oc , which characterizes the nonlinear relationship between the state of charge and the open-circuit voltage; R ₀ is the ohmic internal resistance of the battery; R _s , C _s are the electrochemical polarization resistance and capacitance of the battery, representing the short-term circuit polarization response; R _l , C _l are the battery concentration polarization resistance and capacitance, representing the long-term The circuit polarization response of ; I _c is the working current; V _c is the battery terminal voltage;

(2-2) According to the second-order RC equivalent circuit model, the state space equation of the lithium-ion battery is established:

V _c =V _oc (SOC)-V _s -V _l -R ₀ I _c (k); ②

Among them: T is the sampling time, _CN is the battery capacity, k is the discrete time variable; the same below;

(2-3) Combine the state space equations ① and ② to obtain the estimated value of the open circuit voltage according to the sampled current value;

(2-4) Compare the estimated value with the measured value.

S3. Using the Kalman filter algorithm, the estimated second-order RC equivalent circuit model is estimated and updated in real time through the estimated value and measured value of the open circuit voltage calculated in S2.

Specifically, it includes the following sub-steps:

(3-1): According to the second-order RC equivalent circuit model of the lithium-ion battery and Kirchhoff's law, the system transfer function can be obtained:

Let the time constants be as follows: T _s = R _s C, T _l = R _l C _l , a = T _s T _l , b = R _s T _l +R _l T _s +R ₀ (T _s +T _l ),c =R ₀ +R _s +R _l , d=T _s +T _l then there are:

由双线性变换得到：(3-2): Order

Obtained by bilinear transformation:

in:

Under the condition that the fixed sampling period is T, select the state equation and the observation equation:

θ=[k ₁ ,k ₂ ,k ₃ ,k ₄ ,k ₅ ] ^T

y(k)=-k ₁ ·y(k-1)-k ₂ ·y(k-1)+k ₃ ·I(k)+k ₄ ·I(k-1)+k ₅ ·I(k -2)

(3-3): The state variables are estimated online by the Kalman filter algorithm, and then the inverse transformation of z is used to obtain the identification values of the parameters in the equivalent circuit model. Through the discharge experiment of lithium iron phosphate battery packs with the same consistency, the average voltage value of the lithium ion battery is selected as the effective data, and the KF algorithm is used to identify the parameters of 100 sets of voltages. Its parameter identification curve is shown in Figure 3. It can be seen that the ohmic internal resistance R ₀ of the battery changes little during the entire 100 consecutive discharge stages; the polarization resistances R _s , R _l and the polarization capacitances C _s and C _l have opposite transformation trends. When the polarization resistance increases, Polarization capacitance is reduced.

S4, by charging and discharging lithium batteries with different SOC values at constant current, measuring the open circuit voltage after a period of rest, and fitting the relationship between the open circuit voltage and SOC of the lithium battery.

Specifically, it includes the following sub-steps:

(4-1): Fully discharge/charge the lithium-ion battery and let it stand for a period of time;

(4-2): Carry out 10 constant-current pulse charge/discharge experiments at equal intervals on the battery. After each pulse experiment, let it stand for 3 hours, and measure the terminal voltage of the battery. The terminal voltage at this moment can be approximated as the open-circuit voltage;

(4-3): Obtain the average value of lithium-ion battery packs at each stage through multiple experiments;

(4-4): Obtain the average voltage of the lithium-ion battery pack with the same consistency through experiments as the open circuit voltage value of the single cell, and use the fifth-order fitting function. The relationship between the identified open-circuit voltage and the state of charge is shown in Figure 4. The open circuit voltage and SOC relationship of the higher order fit is as follows:

V _oc =15.4647SOC ⁵ -44.9905SOC ⁴ +49.5110SOC ³ -25.1893SOC ² +5.8724SOC+2.7669;

S5, using the improved genetic algorithm to find the optimal noise covariance matrix.

As shown in Figure 5, it specifically includes the following sub-steps:

(5-1): Select the chromosome encoding method, and randomly generate the initial population according to the selected encoding method;

(5-2): Judging whether the convergence conditions are met, generally judge whether the number of iterations is reached, if the execution step (5-6) is satisfied, the execution step (5-3) is not satisfied;

(5-3): Use inverse dichotomy to select crossover individuals: divide the n individuals in the chromosome set into n/2 sets at random. Crossover operations are performed on the chromosomes within the set to produce progeny chromosomes. Two sets of progeny chromosomes are randomly selected for binding, generating n/4 bindings and labeling which set each chromosome comes from. Perform pairwise crossover operations on chromosomes from different sets within the set. Two sets of progeny chromosomes are randomly selected for binding, generating n/8 bindings and labeling which set each chromosome comes from. Continue according to the previous operations until finally merged into a set. ;

(5-4): Use the improved crossover probability algorithm to perform the crossover operation on the selected crossover individuals:

为当前种群平均适应度值；F为进入交叉配对操作个体的适应度值。Among them, P _e0 is the reference crossover probability, which can be between 0.85 and 0.95 according to the actual situation; F _best is the optimal individual fitness value in the current population;

is the average fitness value of the current population; F is the fitness value of the individual entering the cross-pairing operation.

(5-5): Select mutation and return to step (5-2);

(5-6): Output the optimal solution.

S6. Using the optimal noise covariance matrix obtained in S5, using the unscented Kalman filter algorithm, the battery state is estimated by the open circuit voltage estimated value obtained by S2, and the current open circuit voltage estimated value is obtained; then the lithium battery is open circuit obtained through S4. The relationship between voltage and SOC, and output the SOC estimate value corresponding to the current open circuit voltage estimate value.

Specifically, it includes the following sub-steps:

(6-1): List the state equation and measurement equation of the system:

X _k =f(X _k-1 ,U _k )+W _k

Y _k =g(X _k-1 )+V _k

k is the current moment f(X _k-1 , U _k ) is the nonlinear system state transition equation, g(X _k-1 ) is the nonlinear measurement equation, X _k is the state variable, U _k is the known input, Y _k is the measurement signal; W _k is the process noise, and V _k is the measurement noise. We assume that W _k and V _k are uncorrelated white Gaussian noise with zero mean, and their covariances are Q _w and R _v , respectively.

The electrochemical processes such as battery self-discharge, electrochemical polarization and concentration difference polarization are analyzed for the aforementioned battery equivalent model. The model can be divided into the following two parts: the model based on the running time and the model based on the voltage-current characteristic. Among them, the part based on the voltage-current model can be obtained by analyzing the external characteristics of lithium-ion battery discharge, and the battery state of charge value is taken between 0 and 1.

Based on the aforementioned battery equivalent model, it is not difficult to list the state equation and measurement equation of the lithium-ion battery system as follows:

V _c =V _OC +I _c R ₀ +V _s +V _l

The current I _c takes a positive value during charging and a negative value during discharging, and C _q is the nominal capacity of the lithium-ion battery.

(6-2): Calculation of initial value of each state quantity:

where k|k-1 is the estimated value of time k based on time k-1.

(6-3): Establish Sigma point:

(6-4): Update the state equation:

(6-5): Update the measurement equation:

(6-6): By repeating the above four steps, the optimal state estimation value X k at time k can be estimated according to the state value at time k-1 and the observation value obtained at time _k .

As described in step (1-1), the lithium-ion battery was tested with pulse charge and discharge current under the conditions of setting the initial SOC value of 0.2, 0.4, and 0.6, respectively. The pulse charge and discharge experiment is shown in Figure 6. Among them, Figure 7 and Figure 8 are respectively the SOC estimation value and estimation error curve based on the unscented particle filter algorithm under the experimental conditions of pulse charge and discharge current. It can be seen that the closer the initial value is to the real value, the faster the convergence speed; even if there is a large error between the initial set value and the real value, after a period of correction and iteration, it will quickly converge to the theoretical value; Under working conditions, when the initial state of charge is large, after about 200s of adjustment, the theoretical value can be tracked stably; when the estimated value is stable, the estimated error is between ±1.5%.

In order to further verify the tracking performance of the unscented particle filter algorithm under complex conditions. The present invention takes the power demand value of the American small electric car under the UDDS working condition as the driving condition, and then reduces the power demand value to the working condition of the single battery according to a certain proportion. Taking two UDDS cycle conditions as the charging and discharging test conditions of the lithium-ion battery pack, the cycle period is 2792s. Under this condition, the unscented particle filter algorithm is compared with EKF, UKF, and EPF for comparative simulation analysis. Among them, the cyclic working current based on UDDS is shown in Figure 9, and the sampling period is set to 1s during the experiment.

Figure 10 and Figure 11 are respectively the estimated value and the estimated error curve of the lithium-ion battery SOC obtained based on different filtering algorithms under the UDDS cycle condition. Figure 10 shows that in the case of unknown initial value, the use of unscented particle filter algorithm to estimate the state of charge of lithium-ion battery is significantly better than other estimation algorithms in terms of convergence speed and tracking accuracy. Under the complex UDDS cycle condition and the initial estimation error is large, the convergence time is about 250s; it can be seen from Figure 11 that the tracking accuracy of the algorithm after the estimated value is stable is less than 2.0% under the UDDS cycle condition.

The invention discloses a method for estimating the state of a vehicle lithium ion battery based on an improved genetic unscented Kalman filter algorithm, aiming at accurately estimating the working state of the lithium ion battery, establishing a battery management system accurately, and ensuring the safety, reliability, and safety of the lithium ion battery. Efficient operation is important. This method firstly uses the Kalman filter algorithm to identify the equivalent model through the lithium-ion battery charging and discharging experiment, establishes a second-order RC equivalent circuit model, and then uses the unscented Kalman algorithm to estimate the SOC of the lithium-ion battery (State of Charge estimation). , SOC estimation for short), at the same time, an improved genetic algorithm is used to optimize the covariance of system noise and observation noise in the iterative process of the unscented Kalman filter algorithm. Compared with the traditional method, the method proposed in the present invention solves the problem of serious nonlinearity of the system state variable caused by the electrochemical reaction inside the lithium ion battery, and significantly improves the real-time performance and accuracy of estimation.

The above embodiments are only used to illustrate the design ideas and features of the present invention, and the purpose is to enable those skilled in the art to understand the contents of the present invention and implement them accordingly, and the protection scope of the present invention is not limited to the above embodiments. Therefore, all equivalent changes or modifications made according to the principles and design ideas disclosed in the present invention fall within the protection scope of the present invention.

Claims (5)

1. a vehicle-mounted lithium battery state estimation method based on improved genetic unscented Kalman filtering, is characterized in that: it comprises the following steps: S1. Under the condition of constant current discharge of the lithium battery, sample the open circuit voltage and current of the lithium battery; S2. According to the second-order RC equivalent circuit model, establish the state space equation of the lithium battery, and use the current sampled by S1 to calculate the estimated value of the open circuit voltage by using the state space equation; compare the sampled open circuit voltage and the calculated open circuit voltage estimate value; S3, using the Kalman filter algorithm to estimate the second-order RC equivalent circuit model and update it in real time through the estimated value and measured value of the open circuit voltage calculated by S2; S4. Through constant current charging and discharging of lithium batteries with different SOC values, the open circuit voltage is measured after a period of rest, and the relationship between the open circuit voltage and SOC of the lithium battery is obtained by fitting; S5, using the improved genetic algorithm to seek the optimal noise covariance matrix; S6. Using the optimal noise covariance matrix obtained in S5, using the unscented Kalman filter algorithm, the battery state is estimated by the open circuit voltage estimated value obtained by S2, and the current open circuit voltage estimated value is obtained; then the lithium battery is open circuit obtained through S4. The relationship between voltage and SOC, and output the estimated SOC value corresponding to the estimated value of the current open-circuit voltage; The S5 is specifically: 5-1. Select the chromosome encoding method, and randomly generate the initial population according to the selected encoding method; 5-2. Determine whether the convergence conditions are met, and whether the number of iterations is reached. If it is satisfied, execute 5-6, but not execute step 5-3; 5-3. Use inverse dichotomy to select crossover individuals: The n individuals in the chromosome set are randomly divided into n/2 sets; the chromosomes in the set are crossed to generate daughter chromosomes; the two daughter chromosome sets are randomly selected for combination to generate n/4 combinations, and Mark which set each chromosome comes from; perform paired crossover operations on chromosomes from different sets within the set; randomly select two progeny chromosome sets to combine to generate n/8 combinations, and mark which set each chromosome comes from; continue to follow the aforementioned operations , until finally merged into a set; 5-4. Use the improved crossover probability algorithm to perform the crossover operation on the selected crossover individuals:

Among them, P _e0 is the reference crossover probability, which is between 0.85 and 0.95 according to the actual situation; F _best is the optimal individual fitness value in the current population;

is the average fitness value of the current population; F is the fitness value of the individual entering the cross-pairing operation; 5-5. Select mutation and return to 5-2; 5-6. Output the optimal solution. 2. the vehicle-mounted lithium battery state estimation method based on improved genetic unscented Kalman filtering according to claim 1, is characterized in that: described S1 specifically comprises: Select vehicle-mounted lithium battery packs with the same consistency to perform charge and discharge experiments at room temperature, and select the average voltage value of lithium batteries as valid data; To prevent over-discharge of lithium batteries, set the discharge cut-off voltage; Set the constant current discharge rate of the lithium battery, and sample the open circuit voltage and current of the lithium battery with a fixed sampling period. 3. the vehicle-mounted lithium battery state estimation method based on improved genetic unscented Kalman filtering according to claim 1, is characterized in that: described S3 is specifically: 3-1. According to the second-order RC equivalent circuit model of the lithium battery, the system transfer function is obtained by Kirchhoff's law; 3-2. Perform bilinear transformation on the transfer function, and select the state equation and the observation equation under the condition of a fixed sampling period; 3-3. The state variables are estimated online by the Kalman filter algorithm, and then the inverse transformation of z is used to obtain the identification values of each parameter in the second-order RC equivalent circuit model. 4. the vehicle-mounted lithium battery state estimation method based on improved genetic unscented Kalman filtering according to claim 1, is characterized in that: described S4 is specifically: 4-1. Fully discharge/charge the lithium battery and let it stand for a period of time; 4-2. Then carry out the constant current pulse charge/discharge experiment at equal intervals on the lithium battery, let it stand after each pulse experiment, and measure the terminal voltage of the lithium battery; 4-3. Obtain the average voltage of the lithium battery pack at each stage through multiple experiments; 4-4. Obtain the average voltage of lithium battery packs with the same consistency through experiments as the open circuit voltage of a single lithium battery, and fit the function. 5. the vehicle-mounted lithium battery state estimation method based on improved genetic unscented Kalman filtering according to claim 1, is characterized in that: described S6 specifically comprises: 6-1. Analyze the electrochemical process of the battery, and list the state equation and measurement equation of the system; 6-2. Calculation of initial value of each state quantity; 6-3. Establish Sigma point; 6-4. Update the state equation; 6-5. Update the measurement equation; 6-6. Repeat steps 6-2 to 6-5 above.

Images