A lithium-ion battery remaining useful life prediction method based on unscented particle filter and optimal combination strategy

Abstract

Remaining useful life (RUL) prediction is crucial for the lithium-ion battery prognosis and health management. To track the battery degradation process of the highly nonlinear characteristic and predict accurately its RUL, Particle filter (PF) methods are widely used. However, in the state estimation of PF, the degeneracy and impoverishment of particles make the prediction results unreliable and inaccurate. To solve this problem, this paper proposes an integrated lithium-ion battery RUL prediction method based on a particle resampling strategy, namely optimal combination strategy (OCS), and unscented particle filter (UPF). First, the unscented Kalman filter is used to generate the proposal distribution of particles for the calculation of the particle weights in PF. Then, the OCS is employed for the resampling process to improve particles distribution and keep their diversity. Finally, two lithium-ion battery RUL prediction experiments have been conducted, in which the mainstream methods are compared. The results show that the proposed method effectively predicts the battery RUL and validates its superiority and robustness.