基于增强降维原理的可靠性优化机械设计  

Mechanical Design of Reliability Optimization Based on Enhanced Dimension Reduction Principle

作　　者：翁秀奇[1] 陈加国[1] 

机构地区：[1]南京工业职业技术学院机械学院,南京210023

出　　处：《科技通报》2016年第6期63-66,71,共5页Bulletin of Science and Technology

摘　　要：针对目前可靠性优化设计中可靠性分析存在的难以得到优化抽样点问题,本文提出了一种基于变量抽样点的增强降维(enhanced dimension reduction,e DR)方法。该方法针对每个随机设计变量采用不同的轴向抽样点。通过距离准则确定是否增加额外抽样点。这种引入额外抽样点的概念可以在保证传统e DR精度的情况下还可以进一步提高分析效率。通过具体的数学和工程案例对所提出的抽样法进行分析性能研究。对比于传统的基于固定抽样点的e DR方法和功能度量法,所提出的改进抽样法在精度和效率方面都要更优。In order to overcome the optimum sampling problem in reliability analysis of reliability-baseddesign optimum （RBDO）, a new enhanced dimension reduction （eDR） method based on variablesampling points is proposed. It employs different axial sampling points for each random design variable.The concept of importing extra sampling points can increase the efficiency compared with conventionaleDR methods without losing accuracy. In addition, the performance of proposed method is evaluated byapplying in specific mathematic and engineering RBDO problems. The results demonstrated that whencompared with conventional eDR methods with fixed sampling points and performance measureapproach, the proposed methods perform superiorly in both aspects of accuracy and efficiency.

关 键 词：可靠性分析 增强降维方法 轴向抽样 额外抽样点 

分 类 号：TH122[机械工程—机械设计及理论]