Research on Fault Location in Distribution Networks Based on Graph Convolutional Neural Networks

In recent years, the development of the national economy allows the power demand of our country to increase, and enables the scale of the distribution network to further expand. How to accurately judge the faults in the complex distribution network is an important research direction in this field. Solving this problem is of practical significance to promote the safe and stable operation of the distribution network. Locating faults in distribution networks aims at finding the specific location of a fault through appropriate data analysis technology, which is the core technology in fault detection of distribution networks. The traditional fault location methods of distribution networks are mostly based on the analysis of the characteristic information of a single node, and so lack the utilization of the spatial structure of the distribution network. Therefore, once the collected data are inconsistent, inaccurate, interference, noise, and so on, the accuracies of fault location may be significantly reduced. Using a novel technology called Graph Convolutional Neural Network, this paper presents a model of fault location in distribution networks based on deep Graph Convolutional Neural Networks. This location technology can make full use of the information aggregation function and topological structure of a distribution network, in order to find and locate faulty nodes by combining the characteristic information of neighborhood nodes. Therefore, it can effectively improve the accuracies of fault location in distribution networks.