Battery remaining useful life prediction using improved mutated particle filter
Junxia Li, Miao Zhang, Hui Zheng, Jing Jie
Abstract
Abstract Accurate prediction of lithium‐ion battery remaining useful life (RUL) is of great significance for battery health management. Particle filter (PF) is used to predict the RUL effectively, but it suffers from particle degeneracy and impoverishment during the estimation. To solve this problem, this paper proposes a new RUL prediction model based on improved mutated particle filter (IMPF) approach. In the IMPF algorithm, the mutant particles are generated from the prior particles, which not only represent the high probability region but also accelerate the convergence speed of particles. Then an improved polynomial resampling algorithm is proposed to avoid the loss of particle diversity during the resampling process. The simulation results on benchmark function demonstrate the effectiveness of the proposed IMPF algorithm. The method is also applied successfully to a battery capacity degradation data set provided by NASA Ames Research Center. Research results show that the proposed RUL prediction approach has better performance than other related PF techniques.