State of health estimation for lithium-ion batteries using impedance-based simplified timescale information

Accurate estimation of the state of health (SOH) of lithium-ion batteries is critical for enhancing battery safety and operational reliability. The distribution of relaxation times (DRT) provides information of battery electrochemical impedance spectroscopy (EIS) on the timescales, reflecting internal kinetic processes and showing strong correlations with SOH. However, the extraction and application of this timescale information within battery management systems (BMS) are impeded by the need for broadband EIS data and intricate mathematical processes for DRT method. To address these challenges, a simplified timescale information (STI) method based on impedance is proposed, which delineates different battery states without requiring complex calculations. A data-driven SOH estimation model is developed using a gradient boosting decision tree algorithm with STI. Results from the test set indicate that the model, using selected STI (SSTI) features, achieves an average error of only 1.36 %, outperforming existing impedance feature extraction methods. Even excluding battery usage history (such as degradation temperature and state of charge), the model employing SSTI maintains an average error of 2.4 %. Moreover, the proposed SSTI method imposes minimal computational demands and does not require broadband EIS data. As SSTI features can be rapidly obtained through EIS chip, this method shows promise for online, real-time applications, paving a new path for data-driven BMS.