Distribution Network Electrical Topology Identification Algorithm Based on Deep Learning

Abstract

Abstract With the rapid development of smart grids, power data can be collected online and increase in a large amount. Data problems such as missing data, abnormal data have become more and more prominent. Electrical topology is the basis for the implementation of the distribution network. The further development of the smart grid makes new requirements for the accuracy of the electrical topology identification algorithm. Deep learning relies on a large number of data inputs for feature extraction, and shows great tolerance for data missing and fluctuations. This paper proposes a two-channel 1DCNN (One-Dimensional Convolutional Neural Network) model for the electrical topology identification. In this model, voltage and current data are respectively input in each channel for feature extraction with a two-layer stacked CNN. And BN (Batch Normalization) layer and ReLU (Rectified Linear Unit) are added behind each CNN to help convergence. The experiment results show that the proposed model has good accuracy in electrical topology identification.