Fast online decision tree-based scheme for predicting transient and short-term voltage stability status and determining driving force of instability

Abstract

This paper presents a novel fast machine learning-based scheme for predicting the transient and short-term voltage (STV) stability status and determining the driving force of instability, without any need to measure post-fault-clearance data. In the proposed scheme, by employing decision tree (DT) classifier and using the measurements of the installed phasor measurement unit (PMU) at the generator and the induction motor (IM) buses, first, the stability status of either transient or STV (including the sustained low voltage without recovery and the voltage collapse) is predicted just after disturbance in Wide Area Measurement System (WAMS) framework. Then, for unstable cases, based on transient and STV stability mechanisms, the driving force of the instability is determined between transient and STV instability. In order to evaluate the performance of the proposed scheme, the method has been tested on IEEE 39-bus test system and IEEE Nordic test system. The simulation results show the proposed scheme is highly accurate and timely.