Στατιστικός έλεγχος ποιότητας και παραγοντικοί σχεδιασμοί : ανθεκτικοί παραμετρικοί σχεδιασμοί (robust parameter designs)
Ζιώγας, Γεώργιος Μ.
SubjectΑνάλυση και σχεδίαση συστημάτων ; Ποιοτικός έλεγχος -- Στατιστικές μέθοδοι ; Ποιότητα, Βελτιωσή της
Robust parametric designs, which were originally proposed by G. Taguchi (1987), have been studied and extensively used in industrial experiments in the last 20 years. Their aim is to improve product quality and manufacturing production plans that are robust to the volatility caused by factors that are difficult to control by the experimenter. Taguchi proposes the use of cross designs, which can estimate the interactions between control and noise factors (control noise, C N), which are the most significant in the robust design. However the cost of running such a design sometimes is prohibitive due to the large size of runs. Welch, Yu, Kang and Sacks (1990), suggested the use of combined designs, which are more flexible and economic. These designs can easily calculate not only C N interactions, but also interactions between factors of the same type (C C and N N).