Resilient adversarial–evolutionary multi-agent intelligence for real-time EV charging and energy trading in renewable-integrated smart gridsOA

The rapid expansion of electric vehicles(EVs)and renewable energy resources introduces new operational stresses in modern power networks such as peak-load surges,voltage fluctuations,and quality degradation.This work presents a hybrid intelligence-based multi-agent framework that combines Adversarial Reinforcement Learning(ARL)with Dynamic Grey Wolf Optimization(DGWO)to coordinate EV charging,renewable usage,and energy trading.Each entity—EVs,charging stations,renewable units,and the grid operator—acts as an adaptive agent capable of self-learning and cooperative decision-making under uncertainty.The ARL component strengthens learning under variable demand,while DGWO continuously refines control parameters to ensure fast and stable convergence.Simulation studies on a renewable-supported microgrid show a 21%reduction in peak demand,18%higher renewable energy utilization,22%less EV waiting time,and 15%greater profitability than conventional GA,PSO,GWO,and RL methods.Voltage deviation stayed within±3%,power factor exceeded 0.97,and THD remained below 4%,meeting IEEE 519/1547 standards.These results confirm that the proposed ARL–DGWO framework offers a scalable and reliable solution for next-generation EV-grid coordination.