Thermal power prediction of nuclear power plant using neural network and parity space model

Abstract

A power prediction system was developed using an artificial neural network paradigm that was combined with a parity space signal validation technique. The parity space signal validation algorithm for input preprocessing and a backpropagation network algorithm for network learning are used for the power prediction system. Case studies were performed with emphasis on the applicability of the network in a steady-state high-power level. The studies reveal that these algorithms can precisely predict the thermal power in a nuclear power plant. They also show that the error signals resulting from instrumentation problems can be properly treated even when the signals comprising various patterns are noisy or incomplete. >