Transductive Transfer Learning-Assisted Hybrid Deep Learning Model for Accurate State of Charge Estimation of Li-Ion Batteries in Electric Vehicles

Accurate estimation of the State of Charge (SoC) of Li-Ion batteries is crucial for secure and efficient energy consumption in electric vehicles (EVs). Traditional SoC estimation methods often require expert knowledge of battery chemistry and suffer from limited accuracy due to complex non-linear battery behaviour. Owing to the model-free nature and enhanced ability of non-linear regression in deep learning (DL), this paper proposes a hybrid DL model trained by a novel metaheuristic technique, namely the Hybrid Sine Cosine Firehawk Algorithm (HSCFHA). The proposed method utilises the Transductive Transfer Learning (TTL) technique to leverage the intrinsic relationship between different real-world datasets to estimate the SoC of batteries accurately. The evaluation analysis includes three diverse datasets of EV charging drive cycles: the Highway Fuel Economy Test Cycle (HWFET), Highway Driving Schedule (US06) and Urban Dynamometer Driving Schedule (UDDS), at various temperatures of 0°C, 10°C, and 25°C. The considered evaluation metrics, i.e., Normal Mean Squared Error (NMSE), Root Mean Squared Error (RMSE), and $R^{2}$ , achieve values of 0.091%, 0.087%, and 99.51%, respectively. The TTL-HSCFHA-DNN effectively produces higher accuracy with a time-efficient convergence rate, compared to existing methods. The approach enables EV systems to operate more efficiently with improved battery life.