Advances in Robust Parameter Design: From Taguchi's Inner‐Outer Arrays to Combined Arrays

Abstract

Abstract Robust parameter design, originally proposed by Taguchi, is an off‐line technique for maintaining variance reduction and improving quality. According to Taguchi, product variation from the desirable target is affected by two types of factors: control factors and noise factors. The basic idea in robust parameter design is to identify the levels of the control factors so that the product's quality characteristic becomes insensitive to changes in the levels of the noise factors. As a consequence, the effect of uncontrollable variations on the response will be reduced. Taguchi suggested the use of inner‐outer arrays with the control factors being assigned to an inner array and the noise factors assigned to an outer array. After the collection of the data, a performance measure called signal‐to‐noise ratio is calculated and analyzed in order to identify the optimal levels of the control factors. After the introduction of Taguchi's philosophy in the field of experimental design and data analysis, several new designs and techniques have been explored and adopted in this field. This chapter overviews well established approaches and analysis techniques that have been exploited to the robust parameter design field.