Hybrid neural network topology (HNNT) for line outage contingency ranking

Abstract

The line outage contingency was identified as one of the contributors to voltage instability problem. This event has led to significant financial losses in power system resulted from the failure in power operation and energy delivery. This paper presents a hybrid neural network topology (HNNT) for line outage contingency ranking. HNNT is a combination of artificial neural network (ANN) with a loading classifier and fundamental expert system modules. The post-outage severity was predicted by an ANN module trained using the Levenberg-Marquardt modified backpropagation. A line-based voltage stability index termed as fast voltage stability index (FVSI) was utilized as the indicator. Loading classifier distributed the post-outage severity into their respective loading condition. The contingency severities were consequently ranked into four categories using a rule-based module (RBM) that acts as the fundamental expert system. Validation was performed on the IEEE Reliability Test System (RTS) and results indicated that the proposed HNNT can be applied practically.