A Data-Driven Adjustable Robust Optimization Model for Energy Management of Networked Microgrids

Environmental pollution, rapid depletion of fossil fuels, and high energy losses during transmission-distribution are the main problems of traditional power grids. This motivates the development of microgrids (MGs), which are a localized network of fossil fuel and renewable generators, energy storage systems, and electrical loads. Due to the limited operational capacity of individual MGs, multiple adjacent MGs can be networked to form a cluster of interconnected MGs. This paper develops a robust energy management and scheduling model for the co-optimization of internal network operation inside MGs and external energy sharing between MGs. The uncertainty of renewable energy sources is handled by proposing a data-driven robust optimization model with a self-adaptive uncertainty set. This set is constructed by the kernel density estimation method based on the distributional information extracted from uncertainty data. To account for the multi-level and sequential decision-making process of scheduling, the energy management model is formulated as an adjustable robust optimization problem by incorporating wait-and-see decision variables. The results show that compared to conventional robust optimization models, the proposed model is more effective in dealing with uncertainty and can ensure the robustness of scheduling decisions at a lower cost.