Research On High Impedance Grounding Fault Detection Method Of Distribution Network

Distribution network is close to users,and its operation environment is complex.It is easy to contact branches,buildings or the ground when running,and the high impedance grounding fault occurs.For this kind of fault,because of the large transition impedance,the change of electrical quantity is not obvious,and the conventional protection is difficult to operate,so the fault can not be handled in time.Although most high impedance grounding faults do not have serious impact on distribution systems,they may evolve into more serious accidents after sustained development.High impedance fault identification is one of the most challenging and difficult problems in the field of distribution network relay protection.In this paper,the characteristics of high impedance faults in distribution lines and the methods of fault identification are studied,and the problems of high impedance faults identification in resonance grounding system and small impedance grounding system are solved.For the resonant grounding system,the equivalent model of single-phase high-impedance grounding fault is established,and the transient characteristics of under-damped and over-damped high-impedance grounding fault are analyzed in detail.Theoretic deduction and field data show that most high impedance faults are under-damped and beat frequency phenomenon occurs in the transient process of arc suppression coil zero sequence current after fault.The self-similarity of arc suppression coil zero sequence current waveform is low.Accordingly,a high impedance fault identification method based on multifractal detrended fluctuation analysis(MFDFA)is proposed.The method uses MFDFA to determine the fractal parameters,and analyses the influence of different fault conditions and noise on the fractal parameters.The multi-fractal characteristic parameter curve H(q)of zero-sequence current satisfies the arc tangent distribution in the case of under-damped resonant system,and the zero-sequence current H(q)curve tends to be constant when the system runs with unbalanced load or operates with switching.According to the difference of the fractal characteristics,the fault identification criterion is constituted.The reliability and sensitivity of the criterion are verified by the simulation results of MATLAB and RTDS simulation platform.For low resistance grounding system,the zero sequence equivalent network is analyzed.The characteristics of zero sequence current in time domain,frequency domain,steady state and transient state after high impedance fault are studied.The distortion characteristics of unstable arc near zero-crossing point of zero sequence current at fault are focused on.When the arc fault occurs,there is nonlinear distortion near the zero crossing point of zero sequence current.The increment of odd harmonics such as 3,5,7,9 is obvious,and the even harmonic increment is obviously lower than the odd harmonic increment..There is no obvious difference between odd harmonic and even harmonic increment in zero sequence current when the system is running with unbalanced load or switching on and off load.Based on these characteristics,a high impedance fault identification method based on generalized S transform and extreme learning machine is proposed.Neutral zero sequence current is transformed by generalized S-transform to form the complex time-frequency matrix,of which extract the elements to form the characteristic matrix,which are related to the low frequency band near the zero-crossing point of zero-sequence current.Then,based on the eigenvalue matrix energy spectrum,the extreme learning machine is used to identify the similar disturbances between the high impedance fault and the normal operation of the system.The proposed method only needs to collect zero sequence current information.The reliability of the proposed method is verified by the simulation results of MATLAB and RTDS simulation platform.