Detection of operating conditions of turbo alternators using neural network

The application of neural network for study and control of turbo-alternators have been attempted by several authors in the past, using the conventional and well known types of neural networks and algorithms. The attempts have not proven to be very effective mainly due the inherent drawbacks associated with the networks and algorithms used for the purpose. These include large training time, and preestimation of the network structure and size for best performance and utilization. In this paper, the author uses a new paradigm, in which each neuron has a centre surround characteristics, along with conventional perceptron for determination of the operating condition of turbo alternators. It is shown that by the proposed method, most of the problems associated with the conventional techniques are overcome