Prognostics of Lithium-Ion Batteries Based on the Verhulst Model, Particle Swarm Optimization and Particle Filter

Abstract

A novel data-driven prognostic approach for lithium-ion batteries remaining useful life (RUL) based on the Verhulst model, particle swarm optimization (PSO) and particle filter (PF) is proposed. First, the Verhulst model based on the capacity fade trends of lithium-ion batteries is proposed, which is used as the fitting model and predicting model, respectively. Second, the PSO is applied to improve the fitting model. Third, the improved fitting model combined with the Euclidean distance is employed to determine the upper and lower bounds of the predicting model parameters. Fourth, to estimate the predicting model, the PSO is exploited based on the upper and lower bounds of parameters. Then, to compensate the prediction error, the PF is used to update the predicting model. Finally, the RUL prediction can be made by extrapolating the updated predicting model to the acceptable performance threshold. Four case studies are conducted to validate the proposed approach. The experimental results show the following: 1) the proposed prognostic approach has high prediction accuracy and 2) the proposed model needs fewer parameters than the traditional empirical models.