Deep Reinforcement and Graph-Based Optimization for Sustainable Energy Management in Electronic Vehicles

Abstract The integration of Electric Vehicles (EVs) into modern transportation systems has become central toaddressing global energy challenges and mitigating environmental concerns. These technologiespromise substantial reductions in greenhouse gas emissions and a shift toward sustainable mobility.Yet, their success depends on overcoming critical challenges in energy management and charginginfrastructure optimization within dynamic and uncertain environments.This study introduces a hybridframework that combines Deep Dyna Reinforcement Learning (D2RL) and Genetic Optimization to enhance energy management strategies, thereby improving the operational efficiency and cost-effectiveness ofEVs. In EVs, the proposed approach dynamically adjusts engine parameters to optimize power distribution between the internal combustion engine and the electricmotor.For EVscharging infrastructure, the research employs Graph Convolutional Networks (GCNs) to capturespatial and temporal relationships among vehicles, charging stations, and the power grid. Bymodelling stations as interconnected nodes, GCNs enable efficient scheduling based on vehiclelocation, battery state of charge, station availability, and energy pricing. Coupled with D2RL, thissystem ensures optimal resource allocation, reducing waiting times and operational inefficiencies.Inconclusion, the integration of D2RL, GCNs, within energy management strategies represents asignificant advancement toward efficient, reliable, and eco-friendly transportation solutions. Thisresearch establishes a scalable framework that can be adapted across diverse applications, setting abenchmark for future developments in sustainable mobility infrastructure.