APPLICATION OF NEURAL NETWORKS TO A CONNECTIONIST EXPERT SYSTEM FOR TRANSIENT IDENTIFICATION IN NUCLEAR-POWER-PLANTS

Abstract

Expert systems that have neural networks for their knowledge bases are called connectionist expert systems. Several powerful advantages of connectionist expert systems over conventional rule-based expert systems are discussed. The backpropagation network (BPN) algorithm is applied to the connectionist expert system for the identification of transients in nuclear powerplants. In this approach, the transient is identified by mapping or associating patterns of symptom input vectors to patterns representing transient conditions. The general mapping capability of the neural network allows one to identify a transient easily. A number of case studies are performed with emphasis on the applicability of the neural network to the classification problems. Based on the case studies, the BPN algorithm can identify the transient well, although untrained, incomplete, sensor-failed, or time-varying symptoms are given. Also, multiple transients are easily identified with a given symptom input vector.