Neural network based fault diagnosis in an HVDC system

Abstract

This paper presents a neural network (NN) based method for fault classification in a power system. A new method of generating the training data is proposed which has the advantage that the total number of fault simulations needed to generate the training patterns is less than that required by the conventional training method. This is obtained at the cost of a time delay in the NN output response. The performance of the proposed method is investigated using the Matlab simulation model of a simple HVDC system.