Structural and Multidisciplinary Optimization
Hi-index | 0.00 |
This paper introduces a new robust optimization technique which performs tolerance and parameter design using a genetic algorithm. It is demonstrated how tolerances for control parameters can be specified while reducing the product's sensitivity to noise factors. As generations of solutions undergo standard genetic operations, new designs evolve, which exhibit several important characteristics. First, all control parameters in an evolved design are within a set of allowed tolerances; second, the resulting product response meets the target performance; and finally, the product response variance is minimal.