A Study on the and Robustness of Regional Distributed Energy Active Power Automatic Dispatch

China Southern Power Grid Yunnan Power Grid Company is committed to advancing the construction of clean energy sources dominated by new energy, achieving full integration of distributed power sources, and continuously optimizing the construction layout to promote new energy storage and support the development of a new power system. However, challenges remain, including grid load capacity constraints, distribution network upgrade requirements, safety and stability issues, insufficient system regulation capacity, and difficulties in grid connection and consumption. This paper proposes a two-stage active power automatic scheduling model based on hierarchical deep reinforcement learning, combined with an adaptive privacy protection strategy, to achieve autonomous collaborative control of distributed energy while ensuring information security. Simulation results demonstrate that compared to traditional optimization methods, this approach exhibits strong robustness under non-ideal conditions such as power fluctuations and communication delays, providing a new technical paradigm for exploring intelligent regional energy systems. The research findings offer valuable insights for increasing the consumption of distributed new energy and enhancing grid load capacity.