A Three-Phase Unbalance Adjustment Method Based on Improved K-means Clustering

Abstract

In the distribution network system, the three-phase unbalance is commonly existed, and the harm caused by it is still very serious. The intelligent phase-change switches mainly regulate the three-phase unbalance by transferring loads in real time, but it will lead to frequently operation of the switches as to the rapid change of loads, which may bring about risks of impact and short circuit of the distribution system. Therefore, based on the adjustment mode of existing intelligent phase-change switches, the paper proposed a three-phase unbalance adjustment method based on improved k-means clustering, which firstly employed the Frenchet distance to measure the similarities between different load curves, then the k-means clustering was adopted to cluster the loads with similar change rules, after that, every category of loads was balancedly allocated into the three phases so as to achieve robust three-phase balance. The simulation results in the distribution network showed that the proposed method can effectively reduce the three phase unbalance and achieve a robust three-phase balance state.