PRODUCT ROBUST DESIGN WITH A MIXTURE OF RANDOM AND INTERVAL FACTORS

摘要

To solve the product robust design problems involving a mixture of random and interval factors, a novel approach is proposed which combines an integrated metamodeling method with a percentile-based robust optimization model. The integrated metamodeling method embodies the integration of Response Surface Methodology (RSM) and Support Vector Regression (SVR), which is used to construct the metamodels of product performance efficiently. The percentile-based robust optimization model could bring both the design objective robustness and the feasibility robustness of the design constraint into account, which assures a product's reliability and quality robustness to the noise. A case study of a diaphragm spring in automobile clutch is described to show the effectiveness and practicability of the proposed approach.