Fault location on high voltage transmission line by applying support vector regression with fault signal amplitudes

Abstract

This paper proposes a novel high voltage transmission line fault location scheme based on application of support vector regression (SVR). The proposed scheme just uses the amplitudes of the fault voltage waveforms, measured at a single end of the line. Various types of faults at different locations with different fault impedances and a variety of fault inception angles are studied on a 400kV–300km high-voltage transmission line power system. The fault voltages are obtained from 1/8 cycle post-fault signals after the noise has been eliminated using a low-pass filter. The amplitudes of the fault voltage signals are used as features to train the SVR. After training, the SVR is used in the exact location of the fault on the transmission line. When compared with other fault location schemes, the proposed scheme requires less information and a smaller time data window to estimate the fault locations. However, the proposed scheme provides more accurate estimations, irrespective of the fault types, fault inception angles and fault impedances.