A Method for Battery SoH Estimation Based on K-means and LightGBM algorithm

The accurate prediction of the State of Health (SOH) of lithium-ion batteries is critical for the health monitoring and cost-effective operation of electric vehicles (EV) and energy storage systems. Current methods for SOH estimation, such as the ohmic resistance and capacity comparison methods, face limitations in real-time applications due to varying operating conditions and extensive computational requirements. To address the above research gaps, this research aims to develop a method that can more accurately classify battery health features and more reliably estimate SOH in a real-time manner. The proposed method utilizes K-means clustering to group similar battery health features, which serve as the foundation for the LightGBM model to make accurate and robust estimations. The experimental results demonstrate that the proposed method significantly outperforms traditional algorithms in terms of accuracy in estimating the SOH. By offering a more precise and reliable tool for battery health monitoring, this research contributes to more effective battery management systems, ultimately enhancing the efficiency and lifespan of lithium-ion batteries.