A Constrained Robust Optimization for Day-ahead Scheduling of Microgrids with Source-demand Uncertainty

This paper develops a day-ahead scheduling model for microgrids to reduce their operation cost. In view of the intermittent and uncertain photovoltaic power and power demand, i.e. the source-demand uncertainty, a novel constrained robust optimization is incorporated into the model, which minimizes the expected operation cost for the historical distribution of the uncertain conditions and restricts the cost in the worst-case scenarios. To obtain the empirical scenario set with the temporal correlation which represents the historical distribution, a Nataf transformation-based Latin hypercube sampling is developed for the scenario generation and the K-means method is adopted for the scenario reduction. To effectively solve the scheduling problem with uncertainty, the column and constraint generation method and dual theory are adopted, such that the problem is derived as an alternate optimization between the master problem and sub problem. Numerical simulations in a testing microgrid validate the superiority of the proposed scheduling model, and the equilibrium ability of the constrained robust optimization between economical efficiency and robustness.