A new hybrid energy management framework addresses a growing challenge in modern power systems: efficiently coordinating electric vehicle charging while maximizing renewable energy integration and minimizing operational costs.
The proposed system combines two advanced computational approaches. A Pyramidal Dilation Attention Convolutional Neural Network (PDACNN) forecasts renewable generation and electricity demand using multiscale pattern recognition and attention mechanisms—techniques that prioritize the most relevant data signals. This forecast feeds into a Green Anaconda Optimization algorithm that schedules EV charging and discharging cycles while respecting grid constraints such as transformer capacity and voltage limits.
Simulation results demonstrate substantial performance gains. The hybrid framework reduced annual operating costs to approximately $499,626—roughly 35% lower than conventional strategies and 14% lower than genetic algorithms. Carbon dioxide emissions fell by 53.8% compared to baseline approaches. The system achieved a load coverage accuracy of 92.4%, ensuring reliable power delivery even with high renewable penetration.
Crucially, the framework converged to optimal solutions in just 62 iterations, indicating computational efficiency suitable for real-time grid operations. Tests compared two scenarios: one with centralized IoT coordination and another without, revealing that connected smart devices substantially improved performance.
This research addresses a practical gap in virtual power plant (VPP) operations. As electric vehicles proliferate and renewable sources dominate generation mixes, utilities must coordinate thousands of distributed assets in real time. The framework's ability to simultaneously forecast variable supply, optimize flexible loads, and respect physical grid limits makes it applicable to emerging smart grid architectures.
Deploy potential is significant. Utilities managing high-penetration renewable portfolios and growing EV infrastructure could implement this approach to reduce peak demand charges, minimize battery cycling costs, and improve grid stability. The model's stability and rapid convergence suggest feasibility for integration into existing VPP control centers and microgrid operators.
