Comparative Study of Fuzzy Inference System and Design of Experiment for Clean Room Equipment Factory

Abstract

The clean room's performance can extremely improve by eliminating surface particle from materials or parts before entering the clean area. Due to the inconsistency of air velocity of nozzles, so a method that can deal with uncertainty should be implemented. In this study, the comparison of fuzzy inference system (FIS) and design of experiment (DOE) was purposed with the air shower pass box (ASPB) production process of the clean room company in Thailand. The objective was to find out the appropriate factors to increase air velocity of nozzles. FIS was determined and displayed by linguistic terms. Then, the created fuzzy rules were used to extract the fuzzy inputs which consisted of air box thickness, nozzle orifice diameter, blower's company, and angle of welding. 2 k factorial DOE was utilized with these four input factors. Analysis of variance was used to analyze of 3 replications of air velocity of nozzles. The results showed that the FIS model obtained better performance than DOE model for this application. The most appropriate levels of setting achieved by the FIS model by setting the thickness of the air box at 300 mm, nozzle orifice diameter of 27 mm, blower from company A, and angle of welding at 62 degrees. These levels of setting are applied to use in the production design process of the company.