Using Kohonen Feature Maps To Monitor The Condition Of Synchronous Generators

Abstract

This paper illustrates the way in which a neural network, employing unsupervised learning, can be used for the automatic surveillance of the operational condition of synchronous generators. Results show that, if the size of the network is chosen judiciously, then it is possible to cosistently, and unambiguously identifi a range of atypical conditions in such machines.