Enhancing P-V curve tracking efficiency using deep reinforcement learning

To tackle the dual challenges of real-time performance and accuracy in online voltage stability analysis of power systems, this paper proposes an accelerated method for Power-Voltage(P-V) curve tracking based on deep reinforcement learning. By constructing a Deep Q-Network (DQN) framework and incorporating experience replay, a master-slave network design, an e-greedy strategy, we enhance the effectiveness of P-V curve tracking in power systems. Experimental results demonstrate that the application of this method to the actual Yunnan power system significantly reduces the number of iterations and execution time for P-V curve tracking tasks while utilizing fewer training resources.