A Three Stages Decision Tree-Based Intelligent Blackout Predictor for Power Systems Using Brittleness Indices

Cascading outage is one of the main mechanisms for propelling power system toward blackout. Predicting the potential and trend of power system toward blackout helps operator to decide what to do as preventive action against blackout. In this paper, for predicting the size of potential blackout, a three stagesbased decision tree predictor is proposed for estimating the size of the possible blackout. For this purpose, the ability for simulating the process of cascading events is embedded within power factory software by modeling distance, load shedding, out-of-step, and under/over frequency relays. During the process of cascading failures and trend of power system toward blackout, for monitoring brittle condition of power systems, system operational variables are obtained from WAMS by which a series of online brittleness indices are continuously evaluated at the consecutive time intervals. The proposed three stages predicting scheme enables one to discriminate the severity of cascading failures and system brittle condition into three sizes of blackout, namely, α, β, and γ (α <; β <; γ) for which the estimation process are conducted individually for each size but coordinated. The proposed intelligent scheme is demonstrated on 39-bus New England test system and Iran 1063-bus power system with promising results.