Research on Data-Driven Optimal Scheduling of Power System

The uncertainty of output makes it difficult to effectively solve the economic security dispatching problem power grid when a high proportion renewable energy generating units are integrated into grid. Based on proximal policy optimization (PPO) algorithm, safe and economical scheduling method is designed. First, constraints operation systems defined. Then, quintuple Markov decision process defined under framework deep reinforcement learning, transformed process. To low sample data utilization in online learning strategies, PPO algorithm based Kullback–Leibler (KL) divergence penalty factor importance sampling technique proposed, which transforms on-policy off-policy improves utilization. Finally, simulation analysis example shows that system with connected grid, proposed strategy can meet load demand different trends. In dispatch cycle generation rates, be absorbed maximum extent ensure