State Estimation Of Lithium Battery Based On H-Infinity Filter

Based on the gradual deterioration of environmental pollution and energy crisis,people gradually began to pay attention to the research and use of new energy vehicles,and lithium battery pack is the power core of new energy vehicles.In order to make more effective and safe use,battery management system(BMS)also came into being.The core of the battery management system is to estimate the battery state in real time,among which the estimation of battery SOC is the most important,so it has become the main problem of the current research.The state of charge(SOC)of the battery refers to the proportion of the remaining power of the battery to the total power,which is equivalent to the fuel gauge in the fuel vehicle and is an important basis for the remaining driving time of the electric vehicle.In order to improve the accuracy of state of charge estimation of lithium battery,two aspects of parameter identification and estimation algorithm improvement are discussed in this paper.Among several commonly used SOC estimation algorithm,Kalman filter algorithm is put into use on a large scale because of its simplicity and high efficiency.This method is to obtain accurate estimation results in white noise environment by determining the spatial state model of the battery.However,most of the noise in the nature is colored,especially in the driving process of electric vehicles,which makes it difficult for a single Kalman filter algorithm to maintain excellent performance.H_∞ filtering algorithm,as a strong anti-interference algorithm,is an improvement on the theory of Kalman filtering algorithm,which can make it still have better performance in the colored noise environment.By adding adaptive factor and H_∞ filter algorithm to the adaptive robust unscented Kalman filter(ARUKF)algorithm proposed in the unscented Kalman filter(UKF)to estimate the SOC of the lithium battery,and then the parameters and capacity of the battery equivalent circuit model are identified online by the extended Kalman filter(EKF).The accuracy of SOC estimation is improved by continuously adjusting and updating the parameter accuracy.The dual Kalman estimation algorithm composed of ARUKF algorithm and EKF does not need to know the distribution of noise in advance,and the noise can be non Gaussian,so it has little restriction in practical application.This method uses the constant change of battery model parameters and the covariance adaptive adjustment of process noise and measurement noise to reduce the impact of noise.The equivalent circuit model of the battery is established before the process starts,and the relevant parameters and working condition data are further determined through experiments;Then the basic flow of the ARUKF algorithm and the estimation flow of the EKF algorithm are given.By comparing the model error and SOC estimation accuracy of on-line parameter identification and off-line parameter identification,it is verified that the accuracy of double Kalman estimation algorithm is much higher than that of UKF algorithm.Then,by adding colored noise and white noise in the cycle,it is verified that the double Kalman estimation algorithm can effectively reduce the influence of colored noise on SOC estimation,thereby increasing the SOC estimation accuracy.Even in the environment of colored noise,the double Kalman estimation algorithm can keep the SOC estimation error at about 0.5%,and the ARUKF algorithm can also keep the SOC estimation error at 1%,which is more accurate than the UKF algorithm.