Real-time Energy Management in Microgrids: An Envelope-based MPC Approach

Microgrids play a pivotal role in enhancing the flexibility, resilience, and operational autonomy of modern power systems. This paper proposes an envelope-based model predictive control (MPC) approach to support the real-time energy management in microgrids under renewable uncertainties. By leveraging the flexibility envelope and action envelopes derived from day-ahead scheduling, the coarse day-ahead forecast is effectively incorporated into dispatch strategies to avoid myopic actions. In addition, the incorporation of the rolling-horizon framework enables the adjustments in real-time based on short-term forecasts, which further improves the solution optimality. Moreover, the dispatch feasibility can be guaranteed through a quantile-based framework, and the proof is given in this paper. Case studies shows that the proposed method achieves an optimality gap of 4.45% and 4.98% under short-term prediction with maximum error of 15% and 30%. Compared with the classical MPC and basic envelope method, the renewable spillage rate and electricity procurement are much lower.