Abstract: A method is established to solve robust design problems for static and dynamic characteristics by grey relational analysis (GRA). The core of the method is to obtain a relational grade ranking with a larger-the-better characteristic, which can reflect product quality or system performance, through standardization of orthogonal experiment data and computation of relational grade. Compared with Taguchi method, it shows that this method not only attains higher discrimination and adaptability, but also can easily solve dynamic robust design problems with complex functional relationship, which Taguchi method is very hard to solve. Furthermore, this method doesn’t involve complicated process of mathematical statistics, such as parameter estimation, hypothesis testing, et al, which is undoubtedly more practical and easier to understand and apply. To verify this method, two practical examples are considered.

Key words: Dynamic characteristics, Grey relational analysis, Orthogonal experiment, Robust design, Static characteristics