Analysis of the coolant condition with the help of artificial neural networks

Abstract

Questions of the behavior prognosis for the water-chemistry (WC) parameters with the help of artificial neural networks (ANNs) are considered. Using the data of a real thermal power plant, two types of ANNs are developed for the prognosis of the electric conductivity change of an H-cationated feedwater sample. Some variants of algorithms for evaluation of the influence of WC parameters on the predicted regime are considered. Some selection algorithms for the initial ANN parameters of are suggested and their comparative analysis is conducted.