Abstract:
The short circuit fault in dc power system typically generate fast-growing transient current which may cause serious damage for dc grid and power electronic equipment. Besides, the protection of dc system is not as mature as the ac system. Thus, dc grid requires a high speed and robust fault detection method. In this paper, a frequency-domain fault detection method is proposed based on S transform (ST), which is able to detect faults within 0.3~ms. It consists of high-frequency detection, which is responsible for fast response due to high time resolution, and low-frequency detection which is used to differentiate faults from other transient conditions. The high-frequency spectrum is obtained by introducing a correction factor into a Gaussian function in the calculation process of ST. This theoretical background of the ST-based fault detection method is presented in details. It is applied to a multi-terminal dc (MTDC) system on MATLAB/Simulink. The simulation results are verified on a point-to-point (P2P) experimental setup with scaled-down voltage. Comparative analysis of the detection time with other dc fault detection methods, namely, based on wavelet transform (WT) and short-time Fourier transform (STFT) substantiates the effectiveness of this method.