Adaptive Graph Convolutional Network-Based Distribution System State Estimation

Abstract

The management and control of the power systems rely on reliable and timely distribution system state estimation, which is present to be challenging due to significant voltage variations caused by high renewables. To tackle this problem, a graph convolutional network (AGCN) is proposed for the distribution system state estimation (DSSE) by considering highly volatile renewable generation. In particular, the AGCN can enable prompt state estimation for viable system states. In the proposed model, the graph convolutional layer can capture the correlations of the nodal power injections so that enhanced estimation accuracy can be achieved. Moreover, the node-embedding technique is employed in the graph convolutional layer to represent the nonlinear correlation nature, through which the proposed model is allowed to cover general scenarios in the application. The simulation results have been provided to verify the accuracy and effectiveness of the proposed model through IEEE 33-node and the 118-node distribution systems.