State of health estimation for lithium-ion battery based on charging process and improved extreme learning machine

Accurately estimating the state of health (SOH) for lithium-ion batteries is critical for ensuring efficient and reliable battery management. In this study, we propose a novel SOH estimation approach by integrating an improved Extreme Learning Machine (ELM) model optimized using a genetic algorithm (GA). Firstly, five health features strongly correlated with battery capacity degradation were extracted from the charging voltage curve and the incremental capacity (IC) curve, which were effectively denoised using a Gaussian filtering method. Then, a correlation analysis based on the Spearman method was conducted to verify the relevance of these extracted features. Subsequently, these selected health indicators were utilized as inputs to construct the ELM-based SOH estimation model, where the GA was innovatively adopted to automatically optimize key model parameters, significantly enhancing the accuracy and robustness of SOH prediction. Experimental results demonstrate that the proposed GA-ELM model outperforms traditional estimation approaches, achieving a mean absolute percentage error (MAPE) of less than 1%.