Fault Diagnosis of Lithium-ion Battery System Based on Hybrid System and Recursive Least Squares-Extended Kalman Filter

Abstract

A diagnosis method of relay faults and sensor faults based on hybrid system for lithium-ion battery system is proposed in the paper. A battery system, which contains not only discrete dynamics but also continuous dynamics, can be classified into a hybrid system in which hybrid automaton is established to simultaneously handle these two dynamics. For battery fault diagnosis, besides the observation of discrete events, the distinguishability analysis of continuous dynamics is also needed. The recursive least squares and extended Kalman filter algorithm is used to track the continuous dynamics of the battery and generate voltage residual in this paper. The distinguishability analysis is performed based on the voltage residual and current. Federal Urban Driving Schedule test on a battery pack is conducted for evaluating the proposed method. The results indicate that the fault diagnosis method can perform well for the battery system.