Research On The State Of Health Estimation Of Lithium-Ion Battery Based On Unscented Kalman Filter

Lithium-ion battery has a good application prospect because it is the core component of electric vehicle.Battery health status（State Of Health,SOH）characterizes the battery life cycle,which directly affects the reliability and safety of electric vehicle.Accurate estimation of SOH is helpful to improve the accuracy of battery management system（Battery Management System,BMS）and SOC estimation,and contribute to battery fault diagnosis and safety early warning.In order to estimate battery SOH accurately,the work done in this paper is as follows:（1）Firstly,the definition of SOH and the research status of domestic and foreign scholars on SOH estimation are introduced,which are mainly divided into direct measurement method,model-based method and data-driven method.Model-based method is used to estimate SOH in this paper.Then the structure and principle of lithium battery are analyzed,and the degradation principle of lithium battery is described in detail,the reasons for the degradation of three-point lithium battery are summarized,and five factors that affect the SOH of lithium-ion battery are analyzed systematically,which provides theoretical support for the subsequent establishment of equivalent circuit model.（2）Secondly,three commonly used battery simulation models are compared and analyzed,and the equivalent circuit model is selected as the research model according to the internal characteristics of lithium battery.Based on the second-order RC equivalent circuit model,due to the"data overflow phenomenon"caused by the recursive least square method,a recursive least square algorithm（Forgetting Factor Recursive Least Square,FFRLS）with forgetting factor is proposed to identify four parameters in the second-order RC equivalent circuit model.The simulation results show that the FFRLS algorithm proposed in this paper has high accuracy,which lays a foundation for subsequent SOH estimation.（3）Finally,the principles of classical Kalman and unscented Kalman filtering（Unscented Kalman Filter,UKF）are described.Because the variance of state noise and measurement noise of UKF is always constant,its estimation accuracy is low.Based on the improvement of UKF algorithm,an adaptive unscented Kalman filter algorithm is proposed（Adaptive Unscented Kalman Filter,AUKF）,AUKF algorithm enables the system to update state noise and measurement noise in real time by introducing adaptive factors,so as to improve the overall estimation accuracy of the system.Then the open circuit voltage and constant current discharge charging experiments are carried out,and the polynomial curves of OCV-SOC and R₀-SOC are fitted.Based on the above curves,the internal resistance is tracked and estimated.The experimental results show that the root mean square and average absolute error of SOH estimated by AUKF are lower than that of UKF algorithm,which reflects the superiority of AUKF algorithm.