Application of Graph Neural Network for Fault Location in PV Penetrated Distribution Grids

Authors:

Abstract
This paper proposes a Graph Neural Network (GNN) model for fault location in distribution grids to increase the situational awareness in distribution grid monitoring. The proposed model uses the linearized graph convolution filter to enhance the accuracy and convergence of the model. This model incorporates the measured phasor data of voltages and geographical locations of installed measurement devices as input data. The input data is given to graph convolution layers to extract the features of the nodes in different region of the network. Then, the extracted features are used to determine faulty nodes as a classification problem. The proposed method is tested on two modified IEEE 13- and 34-node test systems and the results are compared with baseline methods, which are k-nearest neighbors (kNN) and support vector machines (SVM). The results demonstrate the merit of the proposed algorithm.
Funding Information
  • National Science Foundation (ECCS-2033927)

This publication has 12 references indexed in Scilit: