Distributed Dual Horizon Energy Management Framework for Battery Sizing Forecasting and Market Aware Scheduling

The increasing integration of renewable energy, electric vehicles (EVs), and distributed energy resources in modern microgrids creates challenges for efficient, reliable, and cost‐effective energy management. Existing methods often suffer from inaccurate forecasting, suboptimal battery sizing, limited market awareness, and poor distributed coordination. To address these challenges, this research proposes a distributed dual‐horizon stochastic energy management framework that jointly optimizes long‐term battery planning and short‐term operational scheduling. The long‐horizon module employs a Symbiotic mechanism‐based Supernova Explosion Optimization Algorithm (SSEOA) to determine optimal battery size, considering lifecycle costs, degradation, and operational expenses. The short‐horizon module uses a Quaternion‐Enhanced Attention Network (QEAN) for high‐accuracy forecasting of renewable generation, load demand, pricing, and battery states, enabling multi‐objective adaptive optimal power flow. A distributed coordination layer leverages Interpretable Multi‐Agent Reinforcement Learning integrated with Hills Ecology Optimization (IMRL‐HEOA) to manage EV charging, storage, demand response, and peer‐to‐peer energy trading securely and scalably. Simulation results demonstrate daily cost savings up to ₹625, voltage stability within ±2%, renewable utilization above 80%, and smoother load profiles, outperforming conventional machine learning, deep learning, and reinforcement learning methods. The framework offers a holistic, resilient, and market‐aware energy management solution suitable for modern microgrids.