ESTIMATION OF ENERGY CONSUMPTION IN REAL-TIME EV SENSOR DATA THROUGH EXPLAINABLE AI AND MACHINE LEARNING ALGORITHM

Abstract

Electric Vehicles (EVs) represent a transformative shift in transportation, promising reduced dependence on fossil fuels and lower emissions. EVs rely on sensors to gather vast amounts of real-time data on parameters such as speed, acceleration, battery charge, and environmental factors, which all play a crucial role in determining energy efficiency. This research focuses on real-time estimation of energy consumption using machine learning and explainable AI (XAI) to interpret sensor data effectively. While previous studies primarily focused on evaluating energy usage through traditional methods or basic machine learning algorithms, this work leverages advanced models like Random Forest and Neural Networks, trained on extensive real-time data from Battery Electric Vehicles (BEVs) across varied driving conditions. Additionally, the SHapley Additive exPlanations (SHAP) method is utilized to enhance model interpretability, offering insights into how different parameters—such as vehicle speed and battery current—affect energy consumption. This explainability not only enables more accurate predictions of energy use but also assists in identifying key factors influencing energy inefficiency in real-time scenarios. The proposed approach enhances prior work by improving prediction accuracy and adaptability through XAI, supporting more precise energy management strategies. Ultimately, this research contributes to optimizing EV performance, extending battery life, and reducing range anxiety, which are critical for accelerating EV adoption and guiding future policies on transportation electrification.