The fault line selection of neutral non-effective grounding system in single phase fault is still a problem which isn't resolved properly. The criterions that make use of steady signal are unreliable because of faint permanent grounding current and unstable arc. Recently the hotspot comes to utilize the transient signal construct criterion. Complex wavelet transform is dedicated to compose of criterion for fault line selection.The basic theory of complex wavelet transform is introduced in the thesis. Complex wavelet can reduce two or three fault of real wavelet simultaneously, such as shift sensitivity, poor directionality and absence phase information. Phase information is significant for power system.Steady fault and transient fault are analyzed about single phase grounding. There is zero transient sequence current in line after grounding fault. The current direction is opposite between fault line and sound line, and the zero transient current of fault line is greater than any other sound line. The character is immune to resonant grounded system.The time-frequency distribution of zero transient current for different system is analyzed. Research indicates that the special frequency band varying with different system and fault mode, and the frequency included in special frequency band distribute different in time band.An adaptive time-frequency band phase comparison statistic method is provided. The method utilizes the magnitude of complex wavelet coefficients to catch special region in time-frequency plane, and the phase of complex wavelet coefficients to recognize fault line that is different with sound line atπ. The phase comparison between fault line and sound line is calculated with statistic method. The criterion makes use of most of information in transient signal, which guarantee reliability of the proposed algorithm. The detecting process with the new criterion is not sensitive for disturbance and immune to earth mode, path resistance, inception angles. A lot of simulation results have shown the validity and reliability of the proposed algorithm.
|