Solution of Bi-level Optimization Problem of Virtual Power Plant Participation in Demand Response Based on Genetic Algorithm

The Virtual Power Plant (VPP) aggregates decentralized flexible resources to achieve efficient clean energy use and grid balance. Managing distributed resources and optimizing demand response in a VPP presents a complex bi-level optimization challenge: the upper level designs dispatch and incentive mechanisms, while the lower level focuses on user-side load response. This complex interaction cannot be solved by traditional optimization methods due to the lack of explicit analytical expressions, requiring heuristic approaches like genetic algorithms (GA). GA, inspired by biological evolution, excel in global search and managing constraints, effectively addressing such optimization problems. This paper’s proposed framework designs upper-level strategies, including price signals, subsidy policies, and peak/off-peak distribution, aiming to maximize VPP revenue. The lower level simulates user responses to these incentives, adjusting load and demand accordingly. Feedback between levels refines strategies iteratively. Simulation results confirm that GA efficiently solve the bi-level VPP demand response problem, adapting well to various scenarios.