Design of neural networks for fault diagnosis of a distribution system

Abstract

The application of a hierarchical distributed neural network to fault diagnosis of a distribution system is presented. The artificial neural network (ANN) design has been developed by considering the structure and functions of the distribution system. Three hierarchical distributed ANN modules are considered for the fault diagnosis of the distribution system. The proposed diagnosis system has been designed for a practical Tenaga Nasional Berhad (TNB) distribution system. In the ANN training process, the data sets are generated using the logic operation of relays and circuit breakers corresponding to the respective fault location for cases of single and multiple faults. The simulation results show that the proposed method using ANNs is encouraging and that the method can be in an on-line environment