NEURAL NETWORKS BASED OPERATOR SUPPORT SYSTEM FOR EVENT IDENTIFICATION IN NPPs

A nuclear power plant experiences a number of transients during its operations. These transients may be due to equipment failure or malfunctioning of process support systems. In such a situation, the plant may land in an abnormal state which is undesired. In case of an undesired plant condition, generally known as an initiating event, the operator has to carry out diagnostic and corrective actions. The objective of the plant diagnostic system is to identify potentially unsafe scenarios and to give the plant operators appropriate inputs to perform the corrective actions. When an event occurs starting from the steady state operation, thermal hydraulic parameters develop a time dependent pattern and these patterns are unique with respect to the type of event. Therefore, by properly selecting the process parameters and their value ranges, the Initiating Events (IEs) can be distinguished. To tackle this problem, a number of linear and nonlinear pattern recognition techniques can be utilised [1-5]. For this work, artificial neural networks have been utilised for event identification.