Abstract:

Abstract In order to develop electric vehicles, it is vital to be able to accurately estimate the state charge (SOC) of a lithium battery. To address the problem that the Extended Kalman Filter (EKF) algorithm leads to the Taylor expansion truncation of the higher-order sys-tem. In this paper, a system of state-space equations is established based on the sec-ond-order equivalent circuit model, and a simplified-sphere sample approach is used to improve the Unscented Kalman Filter (UKF) algorithm. The SOC estimation performance of the three algorithms is tested under constant current discharge, pulse dis-charge con-ditions, and UDC conditions, respectively. The simulation results show that Simpli-fied-spherical Unscented Kalman Filtering (SUKF) has smaller errors between SOC esti-mation and theoretical reference values than EKF and UKF. The SUKF is less computa-tionally intensive than UKF and has better timeliness in the onboard battery manage-ment system.

Keywords: