Power Grid Fault Diagnosis Method Based on VGG Network Line Graph Semantic Extraction

LIU Jiang, Wang Yi

Abstract: The first premise of building a smart grid is to achieve the stability of the power system. As the scale of the power grid continues to expand, more complex power grid structures and power grid faults have put forward higher requirements for power grid fault diagnosis. Therefore, it is very important to develop a method that can diagnose faults quickly and accurately. With the widespread application of synchronous phase measurement units (PMUs) in power grids, it is possible to accurately diagnose faulty types by analyzing high-precision data. In order to solve the problems of feature loss and slow convergence in the training process of machine learning, this paper proposes a power grid fault diagnosis method, which converts the PMU data into a line graph as input, and realizes the power grid fault diagnosis method through the excellent neural network model VGG. First, select the appropriate electrical dimension in the PMU data and visualize it, then use VGG to learn image features, output fault diagnosis results, and finally test through the measured PMU data in a certain area. The experimental results show that, compared with the traditional fault diagnosis strategy, the method proposed in this paper can extract data features more effectively, and has the advantages of fast calculation speed, strong generalization ability, and good performance in complex situations.

Keywords: phasor measurement unit, VGG, power grid fault diagnosis, Graph feature extraction