MPGCN-OPF: A Message Passing Graph Convolution Approach for Optimal Power Flow for Distribution Network

Optimal power flow (OPF) i s t he nonlinear, non convex optimization setpoint problem of power and voltage due to the complex relation between system operating status and the OPF solutions with satisfying load demand. The recent advancement in machine learning computation models with unforeseen data availability witnessed a significant transition into data-driven approaches due to information and a feature mapping property. In this paper, we proposed the Message passing graph convolution (MPGCN) model into a unified framework for OPF solutions. The proposed methodology is based on graph convolution property and message passing interface to take advantage of both techniques. The proposed model is an effective alternative to the existing popular DNN technique in terms of model loss function & performance evalu-ation indices with the IEEE-33 bus power distribution network. The simulation results validate that the proposed MPGCN-OPF model outperforms the DNN model. The performance evaluation indices of the proposed model include MSE, RMSE, and MAE are 0.0664, 0.2576, and 0.0719 respectively.