Transmission Line Fault Location Using Deep Learning Techniques

Abstract

Precisely detecting the fault location on transmission lines can significantly save labor effort and accelerate the repairing and restoration process. This paper presents a novel single-ended fault location approach for transmission lines using modern deep learning techniques. A mixed convolutional neural network with long short-term memory (LSTM) structure are trained to predict the fault distance given the single-ended voltage and current measurements. Convolutional function, pooling layers, and the LSTM structure are used to preserve the translation invariance and capture the temporal correlation of the time-series input data. Advanced deep learning techniques such as adaptive moment estimation and dropout are used to efficiently train the neural network and prevent over-fitting. Extensive studies have demonstrated the accuracy and effectiveness of the deep-learning-based, singled-ended fault location approach.