SOH-KLSTM: A hybrid Kolmogorov-Arnold Network and LSTM model for enhanced Lithium-ion battery Health Monitoring

Accurate and reliable State Of Health (SOH) estimation for Lithium (Li) batteries is critical to ensure the longevity, safety, and optimal performance of applications like electric vehicles, unmanned aerial vehicles , consumer electronics, and renewable energy storage systems . Conventional SOH estimation techniques fail to represent the non-linear and temporal aspects of battery degradation effectively. In this study, we propose a novel SOH prediction framework (SOH-KLSTM) using Kolmogorov-Arnold Network (KAN)-Integrated Candidate Cell State in LSTM for Li batteries Health Monitoring. This hybrid approach combines the ability of LSTM to learn long-term dependencies for accurate time series predictions with KAN’s non-linear approximation capabilities to effectively capture complex degradation behaviors in Lithium batteries . KAN addresses LSTM’s limitations in handling non-smooth approximations and memory decay over extended sequences. The combination of LSTM and KAN ensures that the model accurately depicts both the time-dependent changes and the complicated non-linearities of battery degradation. Experimental validation was performed on several subsets from the NASA Prognostics Center of Excellence (PCoE) dataset, which includes Li-ion battery data collected during hundreds of charge–discharge cycles under various operating conditions. The proposed model achieved a Root Mean Square Error (RMSE) of 0.001682 in the NASA B0005 subset, significantly outperforming the LSTM-only model, which achieved an RMSE of 0.058334. This corresponds to a 97.12% reduction in prediction error, reflecting the superior predictive performance of our proposed model, with an accuracy approximately 35 times greater than that of the LSTM model alone. The results of additional NASA PCoE subsets further highlight the superior performance and computational efficiency of the model, positioning it as a promising solution for real-time battery health monitoring and management systems. • Accurate SOH prediction is critical for BMS to ensure optimal battery performance. • A novel SOH-KLSTM model that integrates KAN with LSTM is proposed. • The LSTM is improved by a KAN-enhanced candidate cell state computation. • Extensive experimental validation on several subsets of NASA dataset are conducted. • The SOH-KLSTM model is optimized for real-time battery health monitoring.