Fast DC Optimal Power Flow Based on Deep Convolutional Neural Network

Abstract

The optimal power flow is the cornerstone of the operation and management of electric power systems. However, the stochastic and intermittent uncertainty due to the proliferation of renewable energy resources (RES) poses a non-trivial challenge to timely obtain the optimal operation point of the power system. To address the computational burden issue, a deep convolutional neural network (DCNN) model is proposed to learn the mapping from the injections to the optimal objective. The DCNN reduces the training parameters as well as improves the approximation accuracy. IEEE 14/118/300 bus power systems are conducted, and the optimal power flow model is solved by Gurobi/Python. Simulation results show that DCNN speeds up the calculation time by up to 100 times in comparison to the state-of-the-art solver and simultaneously maintains the required accuracy.