Generalized Regression Neural Networks Based HVDC Transmission Line Fault Localization
Fault location remains a challenging problem in order to restore power supply rapidly. In this paper, a novel single-terminal traveling wave fault location procedure is proposed in a radial distribution network. Firstly, the propagating speed of zero mode traveling wave is determined by BP neutral network method. Then the preliminary estimation of fault distance is calculated based on the difference of velocities by discriminating the accurate timestamps of initial zero and aerial mode wave fronts. With this preliminary fault distance, the exact arrival timestamps of reflected wave fronts from fault position and far terminal end can be successful yielded. Finally, depended on mathematical relation of selected pairstimestamps, the fault distance can be calculated precisely. Simulations have performed in PSCAD/EMTDC and the results show the accuracy and validity of procedures proposed above. 2014 Elsevier Ltd. All rights reserved.