Fault Line Selection Method Of Distribution Network Based On K - Means Clustering Analysis

When a fault occurs in the Resonant grounding system the transient process usually ends up after 1 to 2 frequency cycle then enter the steady-state process. Arc suppression coil mainly compensates steady-state zero mode current of faulted feeder, the impact on electromagnetic transient process of zero-sequence transient current caused by a fault is small. When a line fault occurs, the zero-sequence transient current of fault feeder and healthy feeder are obviously different on the waveform. With the advent of the era of big data, pattern recognition cluster analysis method has been applied to many fields. Using cluster analysis method to handle large disorganized disturbance recorder data can satisfy the information requirements of the fault line selection,, Now, divide the data from the end of the measurement of "fault line" and "healthy line " into two distinct categories, use the pattern recognition methods to discriminate them in order to achieve fault line selection in resonance grounding system.This paper uses db10 wavelet to decompose the fault transient zero sequence current sampled from matlab electromagnetic transient simulation into 6 wavelet layers, the sample lasts 5ms after fault, calculate the total energy of transient zero sequence current under full band, and the Integrated wavelet relative entropy; the two dimensions:the transient zero sequence current energy and relatively comprehensive wavelet energy entropy are characterized as a measure of fault features, and maps it to the two-dimensional plane. Use k-means clustering algorithm to calculate the cluster center data. In the clustering space, "fault line" forms a cluster center, "healthy line" form a cluster center. Apply Euclidean distance to measure current fault data belongs to which category, so as to realize a fault line.