Leveraging Generative and Explainable AI for Electric Vehicle Energy Toward Sustainable, Consumer-Centric Transportation

Abstract

In Industry 5.0, predicting electric vehicle energy usage enhances efficiency and sustainability, optimizes charging infrastructure, and meets consumer demands. Leveraging IoT, AI, and analytics enables smart charging that aligns with renewable energy goals, addresses infrastructure limitations, and promotes sustainable transportation by improving user experience, cutting costs, and boosting trust in electric vehicles. This research aims to create a synthetic data set for electric vehicles using an enhanced Generative adversarial network model and, from that, predict the energy for charging electric vehicles using ensemble Machine Learning algorithms. The importance of more detailed features for the best-performing machine learning model has been done utilizing Explainable Artificial Intelligence, specifically, the Shapley Additive Explanations approach, to provide more understanding and derive the inter-dependency among features. The enhanced TemporalCharge Generative adversarial network model provides Skewness and Kurtosis values as -0.51 and -0.182, respectively, for synthetically generated data for the city of Berhampur, Odisha, India, which is very close to four-wheeler electric vehicle charging data of the city. Several plots illustrate the influence of key features on electric vehicle energy consumption during charging, which enhances user optimization, owner empowerment, and ecosystem sustainability.