An IoT-Enabled Battery Capacity Monitoring System for Solar Panels in Electric Vehicles with Self-Attention and Bidirectional LSTM Models

The expansion of automobiles significantly contributes to deteriorating air quality by augmenting the demand for non-renewable fossil fuels. Solar panels incorporated within EV provide a viable energy solution that may address this issue. Nonetheless, there are challenges to address during this transition, particularly with energy efficiency and battery management. This study concentrates on efficient energy usage, analyzing the components of electric vehicles' battery management systems and their interrelations. A significant issue in charging electric vehicle batteries is the inefficiency that arises when cells with a high SOC charge more rapidly than those with a low SOC. Particle swarm optimization was employed to minimize features, while z-score normalization was utilized to preprocess data for optimal performance. To enhance analytical efficiency, the proposed SA-BiLSTM model integrates a self-attention mechanism with multi-channel features, enabling it to replicate the capabilities of solar panels in electric vehicles without necessitating manual panel arrangement. Achieving a classification accuracy of 97.37%, SA-BiLSTM existing methodologies, as per performance assessment. The results indicate that solar panels in electric vehicles can address battery optimization issues and enhance sustainability. Future research should focus on enhancing these strategies for improved integration with RES in real-world applications.