• 论文 •    

基于性能测量法的序列化多学科可靠性分析

刘少华,李连升,刘继红   

  1. 北京航空航天大学 机械工程及自动化学院,北京100191
  • 出版日期:2010-11-15 发布日期:2010-11-25

Performance measure approach based sequential reliability analysis of multidisciplinary systems

LIU Shao-hua, LI Lian-sheng, LIU Ji-hong   

  1. School of Mechanical Engineering & Automation, Beihang University, Beijing 100191, China
  • Online:2010-11-15 Published:2010-11-25

摘要: 针对现存的多学科可靠性分析方法计算效率低的问题,提出了基于性能测量法的序列化多学科可靠性分析方法。该方法将系统分析、系统灵敏度分析和可靠性分析序列化并组成一个递归循环,消除了每次循环迭代中系统级优化器对可靠性分析模型完整的优化计算。系统分析和系统灵敏度分析分别为可靠性分析提供了极限状态函数的值和梯度信息,可靠性分析采用基于改良的先进均值法减少了子系统分析次数。同时,系统灵敏度分析过程中采用全局灵敏度方程执行并行子空间灵敏度分析,提高了可靠性分析的效率。通过两个算例验证了该法具有较高的计算效率。

关键词: 性能测量法, 多学科设计优化, 可靠性分析, 改良的先进均值法

Abstract: To resolve the low computing efficiency of the existing reliability analysis methods of multidisciplinary systems, a performance measure approach based sequential approach to reliability analysis of multidisciplinary systems was proposed. In this approach, system analysis, system sensitivity analysis and reliability analysis were arranged in a sequential manner to form a recursive loop and thus eliminated the repeated total computation of the optimization model in the optimizer. System analysis and system sensitivity analysis provided the value and gradient information of limit-state function for reliability analysis respectively. The modified advanced mean value method was used in reliability analysis, which reduced the number of subsystem analysis. Meanwhile, the global sensitivity equation was adopted during the process of system sensitivity analysis to implement subspace sensitivity analysis concurrently, therefore, the efficiency of reliability analysis was improved. The high computational efficiency of proposed method was verified with two examples.

Key words: performance measure approach, multidisciplinary design optimization, reliability analysis, modified advanced mean value method, system sensitivity analysis

中图分类号: