留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

可靠性灵敏度分析的一种偏倚蒙特卡罗方法

李静辉 康锐

李静辉, 康锐. 可靠性灵敏度分析的一种偏倚蒙特卡罗方法[J]. 北京航空航天大学学报, 2011, 37(6): 705-710,716.
引用本文: 李静辉, 康锐. 可靠性灵敏度分析的一种偏倚蒙特卡罗方法[J]. 北京航空航天大学学报, 2011, 37(6): 705-710,716.
Li Jinghui, Kang Rui. Biased Monte Carlo method for reliability sensitivity analysis[J]. Journal of Beijing University of Aeronautics and Astronautics, 2011, 37(6): 705-710,716. (in Chinese)
Citation: Li Jinghui, Kang Rui. Biased Monte Carlo method for reliability sensitivity analysis[J]. Journal of Beijing University of Aeronautics and Astronautics, 2011, 37(6): 705-710,716. (in Chinese)

可靠性灵敏度分析的一种偏倚蒙特卡罗方法

详细信息
    作者简介:

    李静辉(1984-),男,湖南娄底人,博士生,sophialjh@dse.buaa.edu.cn.

  • 中图分类号: TB 114.3; TP 391.9

Biased Monte Carlo method for reliability sensitivity analysis

  • 摘要: 根据可靠性灵敏度分析的需求和特点,选择似然比方法作为基础的导数/梯度估计方法.立足经典可靠性系统和基于元件的蒙特卡罗方法,推导了原始蒙特卡罗仿真环境下的似然比导数估计方法.为加速仿真,进一步提出了一种偏倚技巧,该技巧在系统结构函数的基础上定义一个应用重要抽样的无偏估计量,并通过最小化该估计量的方差来获得最优偏倚参数值.该估计量拥有的一个重要优势是其方差优化任务可以分解到单个元件的层次上进行,从而避免了高维优化的难题.通过一个可解析求解的简单实例验证了似然比导数估计方法应用于可靠性灵敏度分析的有效性,以及所提出的偏倚技巧对于降低导数估计方差的有效性.对于所考虑的实例系统,提出的偏倚蒙特卡罗方法对各个感兴趣的量都给出了很好的估计,与原始蒙特卡罗方法相比,各个估计量的方差均降低了至少6个数量级.

     

  • [1] Asmussen S,Glynn P W.Stochastic simulation:algorithms and analysis[M].New York:Springer,2007 [2] Marseguerra M,Zio E.Monte Carlo estimation of the differential importance measure:application to the protection system of a nuclear reactor[J].Reliability Engineering and System Safety,2004,86(1):11-24 [3] Nakayama M K,Goyal A,Glynn P W.Likelihood ratio sensitivity analysis for Markovian models of highly dependable systems[J].Operations Research,1994,42(1):137-157 [4] Nakayama M K.Asymptotics of likelihood ratio derivative estimators in simulations of highly reliable Markovian systems[J].Management Science,1995,41(3):524-554 [5] Nakayama M K.On derivative estimation of the mean time to failure in simulations of highly reliable Markovian systems[J].Operations Research,1998,46(2):285-290 [6] L-Ecuyer P.An overview of derivative estimation[C]//Nelson B L,Kelton W D,Clark G M.Proceedings of the 1991 Winter Simulation Conference.Phoenix:IEEE,1991:207-217 [7] Fu M C.Stochastic gradient estimation[C]//Henderson S G,Nelson B L.Handbooks in Operations Research and Management Science:Simulation.Boston:Elsevier,2006:575-616 [8] Fu M C.What you should know about simulation and derivatives[J].Naval Research Logistics,2008,55(8):723-736 [9] Ho Y C.Performance evaluation and perturbation analysis of discrete event dynamic systems[J].IEEE Transactions on Automatic Control,1987,32(7):563-572 [10] Gong W B,Ho Y C.Smoothed (conditional) perturbation analysis of discrete event dynamical systems[J].IEEE Transactions on Automatic Control,1987,32(10):858-866 [11] Rubinstein R Y.Sensitivity analysis and performance extrapolation for computer simulation models[J].Operations Research,1989,37(1):72-81 [12] Rubinstein R Y,Kroese D.Simulation and the Monte Carlo method[M].Hoboken:Wiley,2007 [13] Pflug G C.Sampling derivatives of probabilities[J].Computing,1989,42(4):315-328 [14] Heidergott B,Vázquez-Abad F J.Measure-valued differentiation for Markov chains[J].Journal of Optimization Theory and Applications,2008,136(2):187-209 [15] Heidergott B,Vázquez-Abad F J,Pflug G,et al.Gradient estimation for discrete-event systems by measure-valued differentiation[J].ACM Transactions on Modeling and Computer Simulation,2010,20(1):1-28 [16] Labeau P E,Zio E.Procedures of Monte Carlo transport simulation for applications in system engineering[J].Reliability Engineering and System Safety,2002,77(3):217-228 [17] Campioni L,Vestrucci P.Monte Carlo importance sampling optimization for system reliability applications[J].Annals of Nuclear Energy,2004,31(9):1005-1025 [18] Campioni L,Scardovelli R,Vestrucci P.Biased Monte Carlo optimization:the basic approach[J].Reliability Engineering and System Safety,2005,87(3):387-394
  • 加载中
计量
  • 文章访问数:  3168
  • HTML全文浏览量:  80
  • PDF下载量:  979
  • 被引次数: 0
出版历程
  • 收稿日期:  2010-12-21
  • 网络出版日期:  2011-06-30

目录

    /

    返回文章
    返回
    常见问答