An Approach for Computer Experiment Based Robust Design D. J. Li; X. Liu; N. Soderborg; A. Sudjianto; M. Vora; S. C. Wang

摘要

It has become imperative to avoid depending solely upon hardware experiments as much as possible and to count on experiments with computer models to bring quality new products to the market cost efficiently and quickly. An analytical robust design based on computer aided model is a necessity. Beside the benefit of a much faster and cheaper design iterations at the early product development phase, analytical robust design can offer some additional benefits. Though the hardware testing and analytical modeling approaches for robust design are trying to accomplish the same task, i.e., robust product performance thus higher reliability through efficient experimentation, the detail methodology is somewhat different. One of the basic differences of the computer experimentation from the physical experimentation is that the computer models are often deterministic models, therefore some techniques used in the physical experimentation to handle random errors, such as blocking, replication, randomization, etc., are irrelevant in the computer experimentation, which leads to the calling for different techniques for computer experimentation.