偏好函数在稳健设计中的应用

工程设计中存在两种不确定性,随机不确定性和主观不确定性.经典的稳健设计可以很好地解决设计中的随机不确定性因素,但是这些方法没有考虑设计中出现的主观因素,因此设计结果可能对某些主观因素非常敏感.当前稳健设计中很少在同一模型中同时考虑这两种不确定因素.为了在工程设计中获得全面的稳健设计,在汽车转向横拉杆的设计中,同时考虑设计中的随机不确定性因素和主观因素,并应用偏好函数集成主观因素.横拉杆的设计是多目标优化设计,目标分别为变形最小、质量最轻.文中应用整体性能偏好函数补偿计算这两个矛盾的目标,寻找稳健的最优解.在个体性能偏好函数定义中,采用线性函数定义;在整体性能偏好函数中存在的权重、补偿系数等主观因素,采用区间法求解.这种方法不仅对设计中的随机因素有稳健性能,并且可以科学地为权重及补偿系数等主观因素变量赋值,提高工程设计的稳健性能.

PREFERENCE FUNCTION IN ROBUST DESIGN

There are two sorts of uncertainty in engineering design, random uncertainty and epistemic uncertainty. There are fewer methods those consider epistemic uncertainty in a classic robust design, while there are many methods to incorporate random uncertainty, so design results may be sensitive to some epistemic factors. There are fewer models those attempt to incorporate both sorts of uncertainty. In order to obtain comprehensive robust design, epistemic uncertainty and random uncertainty are considered synchronously in a design of tie-rod of steering mechanism, and preference function is applied to incorporate epistemic uncertainty. The tie-rod design is multi- objective that includes minimum distortion and mass. The two objectives are conflicted, so aggregation preference function is applied to trade-off between two individual preferences. The definition on individual preference is linear; and the epistemic uncertainty of importance weights and trade-off coefficient between different preferences in aggregation preference function is calculated applying a range method. The aggregated preference method is not only robust to random uncertainty, but also scientific to evaluate on epistemic uncertainty such as importance weight and trade-off coefficient, etc. The robust performance of engineering design is improved.