WSEAS Transactions on Systems
Print ISSN: 1109-2777, E-ISSN: 2224-2678
Volume 11, 2012
Model Development and Comparative Study of Bayesian and ANFIS Inferences for Uncertain Variables of Production in Tile Industry
Authors: , ,
Abstract: The life cycle of tile products are decreasing especially for customized products. The demand changes also fluctuate from time to time for each product type. This phenomena created crucial issue in meeting customers’ demands within required due date. The occurrences of uncertain conditions caused the production line performance not able to meet the requirement because they faced uncertain changes in setup time, machinery breakdown time, lead time of manufacturing, and scraps. Hence, an accurate estimation on the production line in the presence of these uncertainties is required. Robust decision making on production line could be made when an accurate estimation of uncertain variables is modeled. Two approaches based on Bayesian inference and adaptive neuro-Fuzzy inference system (ANFIS) were utilized in this study for models development to estimate the effect of uncertain variables of production line in the tile industry. The models were validated and tested based on data obtained from a tile factory in Iran. The strength of our developed models is that the coefficients of decision variables are nonconstant. The best model was judged according to the mean absolute percentage error (MAPE) criterion. The results demonstrated that the ANFIS model generates the lower MAPE by 0.022 and higher correlation by 0.991 compared to the Bayesian model. Consequently, better decisions are generated due to easier identification of uncertainty data and the elaboration made the production planning process better understood.