Hybrid Bar-Delta Filter-Based Health Monitoring for Multicell Lithium-Ion Batteries using an Internal Short-Circuit Cell Model

Abstract

This paper introduces a novel health monitoring framework using Gaussian process clustering and the proposed hybrid bar-delta filter for multicell Lithium-ion battery systems. The Gaussian process clustering is applied to screen all cell voltages and cluster cells into a normal cell cluster or an abnormal cell cluster. Then, the proposed hybrid bar-delta filter is designed for battery cell health monitoring using a triple filter (i.e., bar filter) estimating states and parameters of an average cell model built by the averaged voltage and current of the normal cell cluster and hybrid delta filters (i.e., delta filters) for estimation of individual cell states and critical parameters including an internal short circuit resistance and capacity used for fault diagnosis. We validate the proposed health monitoring algorithm by using simulation results.