Synchronization Estimation Research Of SOC And Parameters Identification For Li-ion Battery Based On Dual Extended Kalman Filter

With the continuous growth of the global economy,the energy crisis is becoming more and more serious.Along with the consumption of fossil energy,air pollution and other problems are becoming more and more obvious.According to extensive research,exhaust emissions are the main culprits of air pollution and greenhouse effect.Therefore,electric cars will become one of the important ways to replace gasoline burning cars.In electric vehicles,it is necessary to conduct necessary research on battery-related state indicators,including battery SOC(battery state of charge)and battery model parameter estimation.The accuracy of SOC prediction can directly reflect the current range of the battery.The high-precision SOC estimation can greatly improve the efficiency of the battery and effectively reduce the use cost of the power battery.The online identification of battery model parameters can make the parameters of the battery model change with the external temperature,current,aging and other factors,so that the battery model can more accurately characterize the electrochemical reaction of the battery under different conditions,and establish a more accurate basis for the subsequent state monitoring.Therefore,this paper focuses on the synchronous estimation of battery SOC and battery model parameters,and carries out the following studies:(1)According to the data obtained in the battery experiment and the research team’s previous in-depth research on the characteristics of lithium batteries,a second-order RC equivalent circuit model was established,and the mathematical equations of the model were established,and the battery model parameters were assumed to be in the unit sampling time.The inside is the discrete expression of the fixed value for the battery model.In addition,combined with the experimental data of the pulse discharge conditions of the battery model at different temperatures by the research team,the improved particle swarm optimization algorithm is used to identify the offline parameters of the established second-order RC equivalent circuit model.The final improved particle swarm optimization algorithm is different.The average error of the terminal voltage calculated from the model parameters identified under temperature is about 5.5m V;the maximum error is less than 70 m V.On the one hand,the experimental results effectively verify that the second-order RC equivalent circuit model can correctly characterize the internal chemical characteristics of the battery.On the other hand,it fully proves the effectiveness and practicability of the improved particle swarm optimization algorithm in identifying battery model parameters offline.(2)Combining with the battery model parameters obtained by the MPSO algorithm off-line identification,the Kalman filter algorithm(KF)was firstly derived theoretically based on the orthogonal projection method,and the feasibility of its application in SOC estimation was verified.However,due to the highly nonlinear characteristics of the battery,the Extended Kalman Filter(EKF)algorithm is introduced to estimate the battery SOC.However,EKF assumes that the process and measurement noise are unchanged,which is heavily dependent on the accuracy of noise prediction information.Therefore,the adaptive filtering algorithm is introduced,and the influence of the size of the moving window in the adaptive filtering algorithm on the adaptive algorithm is also discussed.The size of the moving window determines the sensitivity of the algorithm and the data information contained in the algorithm.Finally,through the simulation of different dynamic conditions,under the condition that the initial SOC value is not accurate,the EKF algorithm is used to obtain the SOC average estimated absolute error is less than 5.00%,the average estimated relative error is less than 7.50%.After the introduction of adaptive filtering algorithm,the average absolute error of SOC estimation is less than 1.80%,and the average relative error is less than 4.00%.The experimental results fully verify the effectiveness and adaptability of the adaptive filtering algorithm and EKF for state estimation.(3)Introduced the multi-time scale adaptive dual Extended Kalman filter algorithm(MultiSoc-ADEKF)based on the variation of Soc.First introduced the dual Extended Kalman filter(DEKF)algorithm in the online parameter identification and state estimation for the battery,the advantages and disadvantages of specific performance for DEKF,good adaptability to engine operating conditions,a moderate amount of calculation and can better meet the needs of battery model parameters real-time updates,but DEKF information is assuming system noise and measurement noise is not changes over statistical properties,so the adaptive algorithm combined with the traditional DEKF called ADEKF algorithm is established;Secondly because the battery is a slowly time-varying parameter model parameters,within each step is used to identify not only causes the parameters algorithm significantly increased amount of calculation can also lead to SOC and parameter identification is easy to appear the phenomenon of divergence and convergence,so this article puts forward the time scale of the SOC variation based on the principle,can be called the principle of adaptive time scale,Multi-SOC-ADEKF algorithm was established.After working condition of the different simulation test,the results show that the weighted average voltage of absolute error is less than 1.50 m V,terminal voltage error estimate relative error less than 0.06% on average,for the average SOC estimation absolute error less than 1.20%,the average estimate relative error less than 3.00%,proves that the proposed algorithm in battery SOC estimation and the validity and accuracy of the model parameter synchronization estimation.