A Data‐Driven Method based on Discrete Wavelet Transform for online Li‐Ion Battery State‐of‐Health Prediction and Monitoring

Abstract Transportation electrification is accelerating the clean energy transition. Due to high efficiencies and energy density, Li‐ion batteries (LIBs) are used as on‐board energy carrier for battery electric vehicles (BEVs). LIBs are subject to rapid degradation due to fast‐charging, mechanical, electrical and thermal factors. Thus, state‐of‐health (SoH) prediction is required to optimize LIBs exploitation over their lifespan. An online accurate and easy‐of‐implementation battery SoH prediction and monitoring method for BEV applications is here presented. The method implements discrete wavelet transform (DWT) analysis to voltage profiles, measured while driving. Specifically, an extensive cycle aging experimental campaign on NCR 18650 cells was performed, applying two typical US test drives (urban and extra‐urban drive cycle, respectively) to the cells at different SoH. Moreover, tests carried out on LIBs at different temperatures demonstrated that temperature effect on the implemented DWT‐based method can be distinguished and separated from cycle aging effect. The proposed method allows a real‐time SoH estimation showing a good accuracy (MAE, ME and RMSE respectively result in 0.917, 2.897 and 1.32) without requiring high computational efforts. This allows to assess battery SoH during the driving. The method can also be extended to other chemistries requiring a dedicated experimental activity for the parameters tuning.