Communication-Efficient Multi-Objective Edge Energy Management for a Grid-Connected Microgrid Using a Day-Ahead Strategy Library

To reduce the communication and centralized-computation burden associated with frequent intraday upper-level updates under renewable-energy uncertainty, this paper proposes an edge-autonomous multi-objective energy management strategy with an offline–online architecture. In the day-ahead stage, Gaussian Copula modeling and trajectory screening generate temporally correlated photovoltaic (PV) scenarios. The conventional Strength Pareto Evolutionary Algorithm 2 (SPEA2) first constructs a base library C0. A dual-space coordinated selection mechanism then retains every C0 strategy exactly and adds complementary trajectories according to their objective responses and differences in energy storage system power trajectories. In the intraday stage, the edge controller combines current local measurements with preloaded PV, load, and price profiles, propagates every stored candidate over a receding horizon, and applies the first action of the strict minimum-composite-cost candidate. The case-study results show that, without relying on intraday upper-level communication, the proposed method achieves a composite-objective value close to that of the perfect-information centralized reference.