Series-Connected Lithium-Ion Battery Packs’ Self-Adaptive SOH Estimation via Inconsistency Representation Optimization

State of health (SOH) estimation for lithium-ion battery packs under different operating conditions is a practical research topic. The series-connected lithium-ion battery pack serves as the fundamental structure for meeting higher power requirements. However, the inconsistency among battery cells and the various operating conditions make establishing the SOH estimation model difficult. To address these issues, this paper proposes the SOH estimation method via inconsistency representation optimization. The method performs a multi-curve correlation analysis using monitorable voltages to reduce the effect of battery pack operating conditions variations. Considering the influence of inconsistency, multi-dimensional correlation information is fused to obtain the health indicator (HI) and improve the robustness. On this basis, an explicit model for SOH estimation is established. Adding working condition information into the model enables self-adaptive adjustments without needing capacity degradation information, all while maintaining low model complexity. Experimental tests conducted under various operating conditions indicate that the developed HI possesses a representation capability exceeding 0.94, with a difference in representation capability of less than 0.03. Furthermore, the mean absolute error, root mean square error, and mean squared error of the SOH estimation method are less than 2.63%, 3.05%, and 0.09%, respectively. The proposed method provides technical support for self-adaptive estimation of SOH in battery packs.