A novel fusion-based deep learning approach with PSO and explainable AI for batteries State of Charge estimation in Electric Vehicles

One of the main purposes of the Battery Management System (BMS) is to estimate the State of Charge (SoC) of Lithium-ion batteries (LIBs). In this study, we propose a novel fusion model combining Convolutional Neural Networks (CNNs), Long Short-term Memory networks (LSTMs), and Convolutional LSTM (ConvLSTM) architectures to efficiently capture spatio-temporal patterns in battery data efficiently, hence improving the estimate accuracy of SoC. Particle Swarm Optimization (PSO) is employed to optimize hyperparameters and enhance the model's accuracy. The fusion model outperforms the separate CNN, LSTM, and ConvLSTM models regarding performance metrics. Specifically, the fusion model with PSO achieved a Mean Absolute Error (MAE) of 0.01, Root Mean Square Error (RMSE) of 0.094, and a R-2 score of 99% according to experimental assessments. The results confirm the model's effectiveness in estimating SoC, indicating its potential to enhance the dependability and efficiency of BMS. Furthermore, Explainable Artificial Intelligence (XAI) was employed to elucidate the battery SoC model's decision-making process and identify the key elements contributing to the estimate process.