Efficient On-Board Health Monitoring for Multicell Lithium-Ion Battery Systems Using Gaussian Process Clustering

Accurately monitoring health condition of individual lithium-ion (Li-ion) cells such as state of charge (SOC) and state of health (SOH) in multicell batteries is crucial to build high-performance and safety-critical battery systems. However, this involves considerable computational burden to embedded battery management systems (BMSs), as number of battery cells increases. This paper proposes an efficient on-board health monitoring method for multicell Li-ion battery systems consisting of large number of cells using the proposed Gaussian process clustering. The Gaussian process clustering method first preprocesses all cell voltages using Z-score standardization and classifies a normal cell cluster and an abnormal cell cluster using outlier scoring vector. Then, a representative of the normal cell cluster and abnormal cells are carefully monitored using the threads of a condition monitoring algorithm. Instead of using cell-based method for all cells in a pack, the proposed method can significantly reduce the computation cost for onboard implementation. We show the usefulness of the proposed health monitoring algorithm through simulation results.