Abstract
In recent years, the Design of Experiments (hereafter, DOE) have been widely used to decide optimum processing conditions. However, when large interactions between several control factors are present, since they behave as confounding variables, the estimation accuracy is significantly reduced and making the practical use of the DOE extremely difficult in some cases. As a common countermeasure, calculation accuracy is confirmed by comparing, through the final results, the best and worst results. This can be of great harm in terms of time and labor and, if the difference between the best and worst results is large, could result in the DOE estimations being ignored. Therefore, in previous studies, a usable tool for the easy determination of control factor interactions in the DOE was developed; here, said tool was able to determine control factor interactions in the DOE through several mathematical models. This research presented an improvement to the previous tool through an improved algorithm and more detailed mathematical models to evaluate complex control factor interactions. It was concluded that, (1) an improved tool for the determination of control factor interactions in the DOE and the Taguchi Methods was developed, (2) the tool was able to detect previously indistinguishable complex control factor interactions in the DOE or the Taguchi Methods, (3) a new algorithm was able to determine complex control factor interactions in models between control factors and functions.