Small Current Grounding Fault Line Selection Based On Two-dimensional Wavelet Analysis Method

Neutral non-grounded and Petersen-coil grounded systems are also called neutral indirectly grounded system, and they are widely used in power distribution network in China. The single-phase-to-ground fault happens frequently in such a system. When single-phase-to-ground fault occurs, it is very important that how to detect the fault line accurately and quickly and eliminate it. And this is one of the important subjects in electric power system protection. Due to the small fault current and unobvious fault features after single phase-to-ground fault, the problem of fault line selection has been studied for years. In recent years as the thriving development on automation of power distribution, fault line detection is urged more and more.Firstly, through reviewing of the existing theories on fault line selection of the neutral indirectly grounded system systematically, this paper points out the defects and drawbacks of those, secondly, stable and transient features of single phase-to-ground fault in neutral indirectly grounded system are deeply analyzed and lots of simulation based on MATLAB have been made. Both theory analysis and experiment results show that there is a clear transient course, transient fault signal is much larger than steady signal, and it represents the characters of fault more adequately than steady signal, the method using transient signal to identify the fault line has higher reliability and sensitivity. So a novel fault line selection method extracting transient fault features in the fault currents based on two-dimensional wavelet transform in distribution systems is proposed in the paper.However, the present methods focus on using transient signal amplitude or phase only. These methods are insufficient for extracting more transient signal details. When the single-phase-to-ground fault occurs, zero sequence is most easily obtained and includes most fault information. So zero sequence current of each line is analyzed.Amplitude and phase of the signal, which vary with time, can be better described by analytical expression of the signal. The analytical signal possess the same phase spectrum as the original real signal, but its amplitude spectrum is double sized than the original real signal. When the analytical signal is decomposed by wavelet, it is still an analytical signal. The two-dimensional wavelet transform, which is widely used as an excellent tool to detect the edges of the image, has the capacity for detecting local signal mutation and can extract information using multi-scale character. By adding an imaginary part of Hilbert transformed of real signal, zero sequence current signal as a real signal is first expressed with its analytical form in this paper. So real and imaginary part are treated as two variables. After the suitable wavelet function is selected, the analytical signal of zero sequence current is decomposed by two-dimensional wavelet transform. Through comparing the magnitudes and polarities of wavelet coefficients in each line, the fault line is determined. Real part of transient signal reflects amplitude, and imaginary part reflects phase of the transient information. By structuring analytic signals of zero sequence current, the analysis of combined signal of amplitude and phase is implemented. Compared with the using single amplitude or single phase, applying combined signal can obtain more details of transient signal.Theoretical analysis and simulation results show that the presented method can exactly and effectively choose the fault line in single-phase-to-ground fault and is suitable for various distribution systems with different neural grounding modes. Two-dimensional wavelet transform, as a new kind of tools, is introduced, and an entirely new way is given for power system research.