Classification method based on Thevenin model and partial discharge voltage profiles for battery state of health estimation

Accurate estimation of the State of Health (SoH) of lithium-ion batteries is essential for ensuring reliable operation, optimizing performance, and extending service life in applications ranging from electric vehicles to aerospace and portable electronics. This paper proposes a classification-based method for SoH estimation that combines partial discharge voltage profiles with a Thevenin-equivalent circuit model and a statistical similarity metric. A reference data library was generated through MATLAB/Simulink simulations of a 3.3 V, 2.3 Ah LiFePO₄ cell subjected to consecutive constant-current charge–discharge cycles, starting from an initial SoH of 100% and continuing until 80% capacity was reached. Each simulated discharge was performed at 2.5 A and each charge at 1.75 A, producing paired voltage and current profiles along with corresponding SoH values. The proposed approach was validated using both the NASA Prognostics Center of Excellence battery dataset and experimental data from two commercial Samsung 18650 Li-ion cells. Results show that the method achieves high accuracy, demonstrating its suitability for systems when only partial discharge data are available.