A Research Of Single-phase-to Ground Fault Line Selection In Distribution System Based On Waveform Strectching Tranform And Fuzzy K-means Clustering

In China, small current neutral grounding system is widely used in distribution network, including the ungrounding system and the resonant grounding system. Single-phase-to-ground fault occurs frequently, which is more than 80% of the total faults in small current neutral grounding system. However, it would improve the reliability of the power system when single-phase-to-ground fault occurs in resonant grounding system because of the small grounding current and the symmetrical fault phase voltage, which doesn’t affect the continue supply to load. Due to the complex situation of the single-phase-to-ground fault in small current grounding system, weak fault current with the intermittent arc and the influence of the structural parameters of lines, transformers’nonlinear characteristics and electromagnetic interference, the spectral characteristics, energy distribution and attenuation characteristics of fault line zero sequence currents have great differences, which leads to the lack of reliable grounding fault line section method, the problem is still not solved.This paper summarizes the method of single-phase-to-ground fault line selection in distribution network, analyzes the disadvantage of steady-state line section and describes the advantage of transient based in line section. By analyzing the transient and steady zero-sequence currents in resonant grounding system, pointing out that the similarity of transient zero-sequence current waveforms between two healthy feeders is higher than that between fault and healthy feeders, therefore, a novel fault line selection method based on waveform stretching transformation and fuzzy K-means clustering algorithm is proposed.The first half period of transient zero sequence current after fault is selected to make segmenting phase plane transformation based on phase plane theory using Euclidean distance which can reflect the information of amplitude and polarity synthetically to construct characteristic quantity of each line and form characteristic matrix, so that to represent the transient zero sequence current waveforms in another form. According to the characteristics that lines in system can be divided into fault line and healthy lines, fuzzy K-means clustering is applied to divide feature matrix into two categories while the alone one is the fault line. In the signal processing of transient zero sequence currents, this paper presents the transient zero-sequence currents stretching transformation theory for the first time. The method of stretching transformation theory solves the impact of different numerical attributes of zero sequence currents between non-fault lines in some certain, which improves the similarity of transient zero-sequence current waves of healthy lines and the margin of line selection.The simulation model of lOkV resonant grounded system is built in ATP electromagnetic transient simulation software, and lots of simulations of single-phase-to-ground faults in resonant grounding system, such as different grounding resistance, different phase angle, different types, different fault location, grounding fault line, et al, are carried out. Results show that the proposed method can select fault line correctly in all cases. Considering the factors that may exist in engineering application, the paper also simulates some possible influence factors, like performance of zero-sequence current transformer, degree of arc suppression coil compensation, electromagnetic noise, unsynchronized sampling, et al. The results of simulations show that the proposed method can select line correctly under the condition above. A large number of simulation results show that the line selection method of single-phase-t-ground fault in this paper has the characteristic of high reliability, high line selection margin and strong adaptability. It also has practical engineering application value.