Automotive Battery State-of-Health Monitoring: a Parity Relation Based Approach

Effective vehicular power management requires accurate knowledge of battery state, including state-of-charge (SOC) and state-of-health (SOH). A well-known approach to battery SOH monitoring is to infer SOH from battery impedance or resistance. However, to provide accurate and robust battery SOH information, an integrated algorithm is needed. In this paper, we present a parity relation based integrated method for battery SOH monitoring. A parity relation is designed to characterize the behaviors of good batteries during vehicle cranking. A residual, defined as the discrepancy between the actual battery voltage and its estimation obtained from the trained parity relation, is used to infer battery SOH. Through analysis based on the presented battery model describing the battery dynamics during cranking, it is shown that the residual integrates the SOH information provided by both battery resistance and voltage loss, hence enhancing diagnostic/prognostic performance. Extensive performance evaluation results using real vehicle cranking data have shown the effectiveness of the algorithm.