State of Health Estimation for Power Battery Based on Support Vector Regression and Particle Swarm Optimization Method

This paper proposes a PSO-SVR algorithm of state of health (SOH) estimation for lithium-ion batteries based on incremental capacity analysis (ICA) method. In this study, the effectiveness and robustness of ICA method have been determined by analyzing the relation between peak intensity curve and real SOH curve. In order to fit the peak intensity curve and real SOH curve, support vector regression (SVR) is used in battery SOH estimation. Particle swarm optimization (PSO) is utilized to optimize the variable parameters in SVR, and the PSO-SVR model is established to estimate SOH. With the PSO-SVR model, the estimated SOH was basically consistent with the actual global trend of SOH, and the local fluctuations barely imprinted the research results. In addition, SVR, Radial Basis Function (RBF) and PSO-RBF algorithms are compared to prove the advantages of PSO-SVR algorithm. The results show that the PSO-SVR estimation algorithm is most accurate, and RMSEs of the PSO-SVR estimation algorithm of two batteries are 1.59% and 0.56%, respectively.