Graph reinforcement learning reactive voltage control method considering global attention mechanism

Abstract

The invention discloses a graph reinforcement learning reactive voltage control method considering a global attention mechanism, and the method comprises the steps: (I) enabling a reactive voltage control problem to be built into a partially observable Markov decision process, and enabling an intelligent agent corresponding to each inverter to only observe the local state quantity of a limited number of nodes around; (2) constructing a full-connection space-time diagram; (3) performing global attention-based graph reconstruction of the multi-agent by using a graph attention neural network; and (IV) performing multi-inverter action decision making based on a GAMARL method and the like. According to the method, the voltage control performance of the power distribution system is effectively improved, more efficient and stable reactive voltage regulation is achieved, the method adapts to complex dynamic changes of the power distribution system after the high-proportion distributed power supply is connected, and safe and reliable operation of the power system is ensured.

Images