Identification of Emergency in Power Systems by Using Combinations of Neural Networks

V. Pavlovsky and O. Butkevych (Ukraine)


neural networks, clustering, identification, failure operation, pattern recognation, alarm processing.


This paper discusses an artificial neural network (ANN) approach to alarm processing. Basis on this approach the ANN-based identification system’s structure was proposed. Using alarm information of protection relays and circuit breakers this system identifies malfunction and/or wrong actions of protective devices as well as malfunction of circuit breakers. The proposed approach takes into account different power system (PS) configurations, incompleteness of alarm information and includes clustering of ANN output for separating a wrong answers. The testing results clearly show the benefit that can be obtained by using combination different ANNs for the improving of quality of identification.

Important Links:

Go Back