KNN Algorithm based Fault Type Classification in Radial Distribution Network
This paper presents a K-Nearest Neighbors (KNN) based fault type classification in radial distribution network. Karrenbauer Transform is used for modal current signal from three-phase transient. The time delay values of the modal current signals with db6 detail wavelet coefficient levels (D1, D1+D2, and D1+D2+D3) are used in the KNN algorithm. In KNN algorithm, the faulted detail coefficients are normalized and partitioned by the holdout method. The Euclidean distance is used to guess the k-nearest point. The results show that the proposed method identifies all kinds of fault types inradial distribution network.
Keywords - Radial Distribution, Karrenbauer Transform, Traveling Wave, Db6 Wavelet, KNN Algorithm