Accurate prediction of electric vehicles mileage for proactive battery management in varied conditions using multi-criteria approach

Abstract

This research addresses several critical challenges in managing battery aging within Electric Vehicles (EVs). Key challenges include minimizing capacity degradation while maximizing vehicle performance and accurately predicting State of Charge (SoC) under diverse operational conditions. Additionally, integrating factors such as vehicle and battery age, driving cycles, environmental conditions, and regional climate variations posed significant hurdles in achieving comprehensive battery management. To tackle these challenges, the research presents a novel framework integrating an advanced Multihead cross attention-based optimizer with a bidirectional long short-term memory network, further enhanced by the coati optimization algorithm. This innovative approach aims to improve prediction accuracy for critical battery performance metrics, such as SoC and battery lifetime. Implemented and rigorously evaluated using Python, the framework demonstrates a substantial 20% increase in battery performance through detailed SoC and temperature analysis, coupled with an impressive 40% enhancement in battery lifetime. These results highlight the ability to overcome previous challenges and provide robust solutions for effective battery management in real-world EV applications.