Privacy-Preserving Operational Decision Making for Networked Autonomous Microgrids Based on Bilevel Mixed-Integer Optimization

This paper presents a privacy-preserving operational decision-making approach for autonomous microgrids networked via a power distribution system, where the distribution system operator (DSO) and microgrid master controllers (MMCs) make independent decisions as different stakeholders. First, the scheme of bilevel optimization (i.e., the Stackelberg leader-followers game) is applied to investigate the sequential interactions between DSO and MMCs. Then, given the coexistence of continuous and binary variables in the lower-level problems, an exact strong-duality-based reformulation and decomposition framework is customized to cope with the nonconvex nature of the bilevel mixed-integer optimization. Meanwhile, a fast-enhanced Benders decomposition algorithm is proposed to realize local privacy-preserving decision-making, where multiple unified Benders cuts are generated once and the strongest one is selected back to Benders mater problem. Mathematically, the proposed algorithm will not change the equilibrium point and corresponding optimal solution after finite iterations. Finally, through numerical experiments on a simplified two-bus test system and a modified IEEE 123-bus system, we demonstrate the effectiveness of privacy preservation, as well as the robustness and scalability of the proposed approach.