Distributed Power Grid Partition Collaborative Voltage Control Method Based on Deep Learning and Federated Learning

This paper proposes a distributed power grid partition collaborative power control method based on deep learning(DL) and federated learning(FL) to address issues such as strong voltage fluctuations and voltage exceeding limits in power grid systems containing distributed power sources. Firstly, based on the characteristics of distributed power grids, the features of DL models and federated control models are summarized, and their applications in the field of voltage control are analyzed;Then, based on this, a collaborative voltage control method was proposed, which combines the process characteristics of DL and FL to further improve the voltage control level and reduce network losses;Several voltage control methods were simulated and tested on a distributed power grid system. The simulation results showed that the collaborative voltage control method proposed in this paper had the smallest node voltage deviation and network loss in the distributed power grid system, with a reduction of 17.6% and 5.8%, respectively, improving the overall operating efficiency of the power grid.