Fault Diagnosis of UHVDC Transmission Line Based on Deep Neural Network

Abstract

Existing ultra high voltage direct current (UHVDC) fault detection methods have low sensitivity and are difficult to identify high resistance ground faults. A fault diagnosis method for UHVDC transmission system based on the combination of Convolutional Neural Network (CNN) and Gated Recurrent Unit (GRU) is proposed. In this method, 16 electrical signals such as DC line voltage and DC line current are taken as the input of the fault diagnosis model, and then the features are extracted adaptively through the convolutional neural network, and the original features are dimensionally reduced. Finally, the new features are classified by the gated cyclic element network. The ±800kV UHVDC transmission line model was built by MATLAB/Simulink simulation software, and simulation experiments were conducted on different fault areas and fault types. The test results show that the proposed fault diagnosis method can reliably and accurately identify various internal and external faults of UHVDC transmission lines, and has a strong ability to withstand transition resistance.