东北大学学报:自然科学版 ›› 2020, Vol. 41 ›› Issue (7): 943-948.DOI: 10.12068/j.issn.1005-3026.2020.07.006

• 机械工程 • 上一篇    下一篇

一种改进的全局可靠性分析方法

刘博林, 谢里阳   

  1. (东北大学 机械工程与自动化学院, 辽宁 沈阳110819)
  • 收稿日期:2019-09-05 修回日期:2019-09-05 出版日期:2020-07-15 发布日期:2020-07-15
  • 通讯作者: 刘博林
  • 作者简介:刘博林(1980-),男,辽宁朝阳人,东北大学博士研究生; 谢里阳(1962-),男,安徽岳西人,东北大学教授,博士生导师.
  • 基金资助:
    NSFC-辽宁联合基金资助项目(U1708255).

An Improved Method of Global Reliability Analysis

LIU Bo-lin, XIE Li-yang   

  1. School of Mechanical Engineering & Automation, Northeastern University, Shenyang 110819, China.
  • Received:2019-09-05 Revised:2019-09-05 Online:2020-07-15 Published:2020-07-15
  • Contact: LIU Bo-lin
  • About author:-
  • Supported by:
    -

摘要: 基于Kriging模型的高效全局可靠性分析(EGRA)方法在每次循环迭代过程中只能增加一个样本点.为了提高EGRA方法的效率,提出了一种基于多代理模型的改进全局可靠性分析算法.通过引入Kriging模型的预测误差,在每次循环迭代过程中,计算多个代理模型的期望可行函数(EFF)获得多个最佳样本点,将这些样本点同时加入到样本库中并更新所有代理模型,直到满足给定的精度为止.在建立极限状态函数高质量的近似模型后,采用蒙特卡罗方法进行可靠性分析.两个数值算例的分析结果表明,该方法具有较高的效率和准确性.

关键词: 可靠性分析, 期望可行函数, 多代理模型, 不确定性估计, Kriging模型

Abstract: In the efficient global reliability analysis(EGRA) method based on the Kriging model, only one sample point can be added in each iteration. In order to improve the efficiency of EGRA, an improved method of global reliability analysis based on the multi-agent model was proposed. By importing the prediction errors of the Kriging model, the expected feasibility function(EFF) of the multi-agent model was calculated to obtain multiple sample points in each iteration, then these points were added to the design of experiment(DoE) at the same time and all the agents were updated until the given precision was satisfied. After a high-quality model of the implicit limit state function was established, the Monte Carlo method was employed to perform the reliability analysis. The results of two relevant numerical examples verify the high efficiency and accuracy of the proposed method.

Key words: reliability analysis, expected feasibility function, multi-agent model, uncertainty estimation, Kriging model

中图分类号: