Research On Method Of The Small Current Grounding Fault Selection Based On Wavelet Neural Networks

In China the power supply system of6~35kV usually use non-ground neutral systemsor grounding system by the arc suppression coil, commonly called the small currentgrounding system. The probability of the occurrence of single-phase transmission line toground fault is very high. When the single-phase ground fault occurs in the small currentgrounding system, the three-phase line voltage system is still symmetrical, so the operatingprocedures provisions allowing single-phase continue to run one to two hours. Becausesingle-phase ground fault current is small, and many factors, making single-phase groundfault detection accuracy in the small current grounding system is often relatively low.Therefore, the study to effectively improve the selectivity of ground fault protection is verynecessary. So, how to achieve fast and accurate detection of single-phase ground fault hasbecome an important issue in improving the reliability of small current grounding system.This paper summarizes and analyzes the variety of existing line selection method andlists the advantages and disadvantages of each method of line selection on the basis ofmany domestic and foreign literatures for the fault line selection problem of the smallcurrent single-phase grounding system. The characteristics of steady signal and thetransient signal in the non-grounding neutral system and the neutral the arc suppressioncoil grounding system are analyzed in this paper. It provides the theoretical basis for thestudy on some methods of fault line selection and it has proposed a new method of faultline selection using transient signal.This paper presents a fault line selection example in the small current system usingwavelet packet decomposition in the technique of the wavelet analysis and gives the flowof the line selection method based on the theory of wavelet transform and wavelet packetdecomposition.In addition, the paper studies the neural network theory and presents thefault detection method based on the wavelet neural network and also gives the faultdiagnosis process. It can improve the correct action rate of single-phase fault line selection forthe small current grounding system. Through theoretical analysis and simulation verifi-cation, these two methods of the small current grounding fault line selection can improvethe selectivity of protection for single-phase fault line selection. Finally, this paper builds up amodel of the small current system by the arc suppression coil grounding way with fourlines using Simulink simulation platform in MATLAB, to simulate the fault line selectionbased on the wavelet analysis technology and the wavelet neural network technology. Inconclusion, the simulation confirmedthe method of the line fault detection using the wavelet analysis technology and the wavelet neural network technology for the small currentgrounding system has high selectivity of protection. At last, the paper may has somereference value for researches on methods of the small current grounding fault lineselection.