A Quality Forecasting System in Glass Melting Processes using Genetic Algorithms

유전 알고리즘을 이용한 유리 용해 공정에서의 불량예측 시스템

This paper presents a computerized quality forecasting system for glass manufacturing. In forecasting the molten glass quality, we are concerned with three major issues : (1) to find the reasonable time lags between a set of process conditions and the quality measurement of glass products, (2) to find the most significant process variables affecting the quality, and (3) to construct the appropriate causal forecasting models using genetic algorithms. The experimental results show the proposed model results in better forecasting than linear regression model. The suggested forecasting model was implemented successfully and is being currently used in a real manufacturing line.