M. Pandit, L. Srivastava, and J. Sharma
Hybrid fuzzy neural network, feature selection, filter module, imbalanced training set, membership values, contingency ranking
This article proposes a method based on hybrid fuzzy neural network (HFNN) comprised of a filter module and ranking module for online voltage contingency screening and ranking, incorporating load uncertainty and flexibility in ranking. A four-stage multi-output Parallel Self-organizing Hierarchical Neural Network (PSHNN) is developed to serve as the ranking module; it is found to be superior to conventional back-propagation (BP) neural network. The filter module reduces the burden on the ranking module by filtering out the noncritical contingencies. Compared to the deterministic crisp ranking, this method offers a more practical and flexible ranking, and is capable of handling contingencies lying on the boundary between two severity classes. Angular-distance-based clustering is used to reduce the dimension of the HFNN. It is difficult and time consuming to deduce the membership rules from complex input- output data. Therefore, in this work a trained artificial neural network is used as a fuzzy inference engine for if-then mapping. The proposed method is tested on IEEE 30-bus test system and a practical 75-bus Indian system and is found to produce accurate results for previously unseen load patterns almost instantaneously.
Important Links:
Go Back
