Robust State of Health Estimation of Lithium-ion Batteries Using

The State of Health (SOH) of lithium-ion batteries is directly related to their safety and efficiency, yet effective assessment of SOH remains challenging for real-world applications. This paper investigates the estimation of SOH, specifically capacity fading, under partial discharge with varying State of Charge (SOC) levels. The challenge arises as partial discharge truncates the data available for SOH estimation, leading to potential loss or distortion of common indicators. To address this, we explore a convolutional neural network (CNN) to extract indicators for both SOH and changes in SOH (∆SOH) across two successive charge/discharge cycles. The random forest algorithm is then employed to produce the final SOH estimate by leveraging the indicators from the CNNs. Performance evaluation is conducted with partial discharge data across different SOC ranges derived from a fast-discharging dataset. The proposed approach is compared against a differential analysis-based approach and two CNN-based methods using only SOH and ∆SOH indicators, respectively. Results indicate that the proposed method exhibits enhanced estimation accuracy and robustness, with sensitivity analysis validating its superior utilization of available partial discharge data for SOH estimation.