Data-driven Based Uncertainty Set Modeling Method for Microgrid Robust Optimization with Correlated Wind Power

Microgrid is considered an important part of the future zero carbon energy systems.However,the uncertainty caused by renewable energy source brings huge challenges to the scheduling of MG and restricts its ability of carbon emission reduction.In this paper,a novel improved multi-ellipsoidal uncertainty set modeling method is proposed to better depict the uncertainty of wind power and reduce the conservativeness of traditional robust optimization.Probabilistic information from historical data is utilized to capture the temporal correlation of forecast error of wind power,as well as the conditional correlation of forecast error with forecast value,making the uncertainty set more data-adaptive to variation of forecast results and more accurate for uncertainty description.A two-stage robust optimization model of a grid-connected microgrid is established based on the proposed uncertainty set and solved by column and constraint generation algorithm.Simulation results based on actual data illustrate the average unbalanced power of microgrid between day-ahead trading and real-time power exchange with utility grid is dropped by nearly 11.16%compared with a deterministic optimization method,11.86%with traditional box uncertainty set-based robust optimization method,and 2.89%with stochastic optimization method.