Estimation of voltage sags patterns with k-means algorithm and clustering of fault zones in high and medium voltage grids

This paper proposes k-means clustering algorithm to identify voltage sags patterns and group fault zones with similar impact in high and medium voltage electric. The proposed methodology comprises three stages. First, network modeling and faults simulation were performed in order to get information about voltage sags caused by faults in the transmission system. Voltage sags patterns were identified at the second stage by means of a k-means clustering algorithm, allowing the determination of fault zones. Using the power quality measurements data base of the major electricity utility of Bogotá, voltage sags were classified according to the previously determined voltage sags patterns. At the third stage of the methodology a comparison between simulated and measured sags is performed, allowing the identification of sags caused by faults