Fault Diagnosis Algorithms of Distribution Network Based on Convolutional Neural Network

Abstract

Fault diagnosis is very important for power restoration. This paper proposed a basic network architecture design, using a simplified residual connection technology, using Focal Loss as the objective function for supervised training, adding a BatchNormalization layer to the network for optimization, reducing parameters based on ShuffleNet network, and improving accuracy based on attention mechanism, a process that can automatically determine the appropriate CNN architecture for fault diagnosis problems. The CNN used in this paper takes the current information sampled from the fault recorder of each node in the distribution network as the input directly, and does not need to use digital signal processing methods to extract frequency domain features and manually select features. The proposed algorithm is evaluated on the IEEE 34-node system, and the fault classification accuracy of more than 99% is achieved on different lines.