Estimation of voltage instability inception time by employing k‐nearest neighbour learning algorithm

Abstract

In this study, the proposed algorithm is a help for utilities to find out when voltage instability will happen, as soon as they realise that a disturbance may/will result in voltage instability. In this regard, the data gathered, on-line, by a wide area monitoring system (WAMS) is supposed to be processed in an energy management system, where, finally, voltage instability inception time will be determined. The proposed algorithm employs k-nearest neighbour algorithm as an instance-based learning algorithm. This estimation uses only pre-disturbance information and exact disturbance data, so it is fast enough for real-time applications. A comprehensive investigation of the input features has resulted in introducing the most appropriate feature set and the most effective search procedure in estimating the instability time. The simulation results show an acceptable accuracy in estimation of voltage instability time.