Modeling for Control Knowledge in Sintering Process Using Neural Network and Fuzzy Inference

Abstract

This paper describes the modeling of two control knowledge in a sintering process which aims to stabilize the state of sintering and yield variation. One control knowledge for return fine ratio which includes nonlinear data evaluation is modeled by fuzzy inference. The other control knowledge which selects the action variables by a two dimensional heat pattern of the sintering process is modeled by using neural network in the extraction of typical heat pattern. The fuzzy control system that uses operator's knowledge of return fine ratio control leads to uniform operation and improvement of yield. The neural network system makes possible the analysis of the relationship between the two dimensional heat pattern and yield and other sensing data.