Adaptive multiple fault detection and alarm processing with probabilistic network

Abstract

This paper presents the fault detection and alarm processing with fault detection system (FDS). FDS consists of adaptive architecture with probabilistic neural network (PNN). Training PNN uses the primary/back-up information of protective devices to create the training sets. However when network topology changes, adaptation capability becomes important in neural network application. PNN can be retained and estimated effectively. With a looped system, computer simulations were conducted to show the effectiveness of the proposed system, and PNNs adapt network topology changes.