A data-driven remaining capacity estimation approach for lithium-ion batteries based on charging health feature extraction

Capacity degradation monitoring of lithium batteries is necessary to ensure the reliability and safety of electric vehicles. However, capacity of cell is related to its complex internal physicochemical reactions and thermal effects and cannot be measured directly. A data-driven remaining capacity estimation approach for lithium-ion batteries based on charging health feature extraction is presented in this work. The proposed method utilizes rational analysis and principal component analysis to extract and optimize health features of charging stage which adapt to various working conditions of battery. The remaining capacity estimation is realized by relevance vector machine and validations of different working conditions are made with six battery data sets provided by NASA Prognostics Center of Excellence. The results show high efficiency and robustness of the proposed method.