A physics-informed and data-driven method for state of charge estimation of lithium-ion batteries in electric vertical takeoff and landing aircraft

Accurately estimating the state of charge (SOC) of lithium-ion batteries is critical for the safety and energy management of electric vertical takeoff and landing (eVTOL) aircraft; however, their unique high-stress operating conditions pose significant challenges to accurate estimation. To address these challenges, a unified physics-informed and data-driven framework for SOC estimation is proposed. A hierarchical Convolutional Neural Network – Long Short-Term Memory – Multi-Layer Perceptron (CNN-LSTM-MLP) architecture is developed to capture the spatiotemporal characteristics of battery behavior under high-stress operating conditions. A novel physics-based indicator, termed the Discharge Power Utilization Index (DPUI), is introduced to characterize the impact of rapid power pulse discharge on SOC identifiability. In addition, a transient-aware dual Kalman filtering strategy is proposed to enable robust online co-estimation of model parameters and polarization states under mission–phase transitions. Comprehensive evaluations on a public eVTOL battery dataset demonstrate robust SOC estimation over the battery life cycle and across different flight phases, with an overall root mean square error (RMSE) of 0.47% and mean absolute error (MAE) of 0.33%, confirming the effectiveness of the proposed method under high-stress eVTOL mission profiles.