노이즈가 포함된 과열증기표의 모델링에 신경회로망 적용 타당성에 대한 연구

Abstract

The thermodynamic state variables in superheated region of steam table are not wholy obtained bymeasurements. This means that steam table contains a little error. In this study small error was artificiallyadded to superheated variables and modeled using neural networks. The results were compared with theanalysis using quadratic spline interpolation method. By and large the relative errors of variables by neural networks were sufficiently small and similar to orless than those by quadratic spline interpolation method. It was concluded that neural networks could be onegood way of modeling for superheated steam table.