Joint Estimation of State of Health and Remaining Useful Life for Lithium-ion Batteries Using Real-World Vehicle Operation Data

Powe <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$r$</tex> batteries, as complex time-varying nonlinear systems, present considerable difficulties in accurately estimating their State of Health (SOH) due to limited parameter measurability, strong coupling of internal characteristics, and gradual capacity degradation over time. Unlike methods that rely on experimental cycling data from laboratory batteries, this study introduces the Dynamic gated TCN-Transformer network that leverages real-world vehicle operation data to enhance SOH and Remaining Useful Life (RUL) prediction. The model adopts a dual-branch architecture, enabling it to capture both local and global degradation trends. Additionally, a dynamic gating fusion mechanism is introduced to adaptively integrate critical information, improving robustness and predictive performance. Extensive validation using real vehicle operation data from multiple regions demonstrates the model's effectiveness, achieving an average RMSE of 1.5080 and an average MAPE of 1.2869%. The results demonstrate that the proposed model in this study can effectively estimate the SOH of power batteries from real-world vehicle data.