Reducing the conservatism of robust optimization in an efficient way has always been a challenging problem for microgrid day-ahead dispatch with wind power integrated. In this paper, we address this problem by formulating a multi-ellipsoidal uncertainty set (MEUS) that is able to capture the strong temporal correlation of forecast error of wind power (WPFE) as well as the conditional correlation of WPFE with the forecast power, and combining it with a box uncertainty set. The dimension of each ellipsoid is optimized based on a comprehensive evaluation index to reduce the invalid region, so as to improve the conservatism of the model. A two-stage robust optimization model of microgrid is built based on the improved MEUS, which is cast into a mixed-integer second-order cone programming problem and solved by column and constraint generation algorithm. The effectiveness of the proposed method is verified by numerous simulations with actual data.
Keywords microgrid, temporal and conditional correlation, improved multi-ellipsoidal uncertainty set, two-stage robust optimization