VSC-HVDC Fault Location Method Based on Effective Feature and Depth Belief Network

Abstract Aiming at the problems that most fault location methods are applied to high voltage direct current (HVDC) transmission system, such as low accuracy and vulnerability noise, a source converter (VSC)-HVDC method based on effective feature deep belief network is proposed. Firstly, Karenbauer phase mode transform used decouple original data eliminate coupling effect between positive negative lines. Then, Hilbert Huang (HH) obtain six features of information from frequency domain time analysis fully characterize system signal. Finally, extracted input into depth for learning, so estimate in line. VSC-HVDC built MATLAB platform experimental analysis. The results show obtained by proposed has smallest overall error, maximum minimum errors divided about 0.52 km -0.95 km. Compared with other methods, it certain advantages location.