摘要
在大部分的实际工程优化问题中,产品质量特性与不确定性因素的关系是未知的或复杂的。为此提出了一种基于混合响应面技术的多目标稳健优化设计。该方法首先利用多项式响应面以及神经网络构造产品质量特性的显性表达式。在考虑目标函数以及约束稳健性的前提下,结合多目标技术中模糊偏好的方法设计权重,从而得到稳健设计的精确解。该方法解决了基于近似技术的稳健设计中计算精度与计算效率之间的矛盾。通过对双杆衍架结构进行算例分析,证明了该方法的可行性。
For most practical engineering optimal＂ problems,the relationships between quality characteristics and uncertain factors are unknown or complicated. A hybrid response surface method is presented to solve this problem in robust design. By considering the robustness of objective functions and constraint functions the relationship be- tween quality characteristics and uncertain factors can be determined by polynomial response surface method; if polynomial response surface method is not precise enough then artificial neural network is adopted to obtain a pre- cise solution of robust design using multi-objective fuzzy preferences method to determine the weights. The proposed method achieves a good balance between computing precision and cost. As a demonstration, a two-bar structure system is analyzed with our method. Results show that our method is feasible for robust design.
出处
《机械科学与技术》
CSCD
北大核心
2008年第5期628-632,共5页
Mechanical Science and Technology for Aerospace Engineering
基金
国家重点基础研究发展计划973项目(2007CB714000)
辽宁省科学技术计划工业攻关项目(2006219012)资助
关键词
多项式响应面
神经网络
质量特性
稳健设计
polynomial response surface
artificial neural network
quality characteristics
robust design
