Health Monitoring of Li-Ion Battery Systems: A Median Expectation Diagnosis Approach (MEDA)

The operations of Li-ion battery management system (BMS) are highly dependent on installed sensors. Malfunctions in sensors could lead to a deterioration in battery performance. This paper proposed an effective health monitoring scheme using a median expectation-based diagnosis approach (MEDA). MEDA calculates the median of a possible set of values, rather than taking their weighted average as in the case of a standard expected mean operator. Furthermore, a smoother was developed to capture important patterns in the estimation. The resulting filter was first derived using an one-dimensional (1-D) system example, where the iterative convergence of median-based proposed filter was proved. Performance evaluations were subsequently conducted by analyzing real-time measurements collected from Li-ion battery cells used in hybrid electric vehicles (HEV) and plug-in HEVs (PHEV) duty cycles. Results showed that the proposed filter was more effective and less sensitive to small sample size and curves with outliers.