Power System Network Topology Identification Based on Knowledge Graph and Graph Neural Network

The automatic identification of the topology power networks is important for data-driven and situation-aware operation grids. Traditional methods lack a data-tolerant mechanism, accuracy their performance in terms thus affected by quality data. Topology related to link prediction problem. graph neural network can be used predict state unlabeled nodes (lines) through training on features labeled with fault tolerance. Inspired characteristics network, we applied it this study. We propose method identify based knowledge network. graphs quickly mine possible connections between entities generate structure data, but case errors or informational conflicts they cannot accurately determine whether relationships really exist. use data mining connection obtained eigenvalues, has tolerance mechanism adapt needs as database. combination compensate deficiency single method. tested proposed using IEEE 118-bus system provincial system. results showed that our approach feasible highly tolerant. It even presence conflicting missing measurement-related information.