北京航空航天大学学报 ›› 2017, Vol. 43 ›› Issue (5): 951-960.doi: 10.13700/j.bh.1001-5965.2016.0386

• 论文 • 上一篇    下一篇

非完美特性下的多状态系统检测与维修优化

李志栋1,2, 张涛1   

  1. 1. 中国科学院 空间应用工程与技术中心, 北京 100094;
    2. 中国科学院大学 , 北京 100049
  • 收稿日期:2016-05-10 出版日期:2017-05-20 发布日期:2017-05-27
  • 通讯作者: 张涛,E-mail:zt@csu.ac.cn E-mail:zt@csu.ac.cn
  • 作者简介:李志栋,男,博士研究生。主要研究方向:计算机仿真、系统优化、人工智能、高性能计算等;张涛,男,博士,研究员,博士生导师。主要研究方向:高可靠软件测试验证技术、高可靠电子信息系统分析设计方法、复杂系统仿真、虚拟现实技术等。
  • 基金资助:
    KJZ总体设计1(Y31405210N)

Optimization of inspection and repair of multi-state system under imperfect characteristics

LI Zhidong1,2, ZHANG Tao1   

  1. 1. Technology and Engineering Center for Space Utilization, Chinese Academy of Sciences, Beijing 100094, China;
    2. University of Chinese Academy of Sciences, Beijing 100049, China
  • Received:2016-05-10 Online:2017-05-20 Published:2017-05-27
  • Supported by:
    KJZ Overall Design 1 (Y31405210N)

摘要: 对于带有周期检测的系统进行维修优化时,不仅需要考虑系统自身可靠性信息,还应该充分利用检测数据并优化检测周期。以多状态并联可修系统为研究对象,考虑非完美检测和非完美维修,以降低系统运行成本率为目标实现系统检测和维修优化。利用非齐次马尔可夫链建立系统可靠性模型,对系统退化、检测和维修进行蒙特卡罗仿真。利用粒子滤波融合系统模型与检测数据并估计系统剩余寿命。设置寿命相关门限触发系统维修,以成本率期望仿真结果为目标函数,使用遗传算法优化检测周期和维修阈值。通过算例证明该方法可有效克服检测误差并实现检测和维修优化。

关键词: 多状态系统, 维修优化, 非完美检测, 粒子滤波, 遗传算法

Abstract: For repair optimization of system with periodic inspection, we need to not only consider the system reliability information, but also make full use of the inspection data and optimize the inspection period. A multi-state parallel repairable system was taken as the research object, and the system inspection and repair optimization were realized considering the imperfect inspection and imperfect repair with the objective of reducing operation cost rate. System reliability model was established by non-homogeneous Markov chains, and Monte Carlo simulation was carried out for system degradation, detection and repair. Particle filter was used for fusing the system model and inspection data, and the residual life of the system was estimated. Life related thresholds used for triggering repairs were set, and simulation results of the expecting cost rate were used as the objective function for a genetic algorithm to achieve optimization of the inspection period and the thresholds. It is proved that this method can effectively overcome the inspection error and achieve the optimization of system repair and inspection.

Key words: multi-state system, repair optimization, imperfect inspection, particle filter, genetic algorithm

中图分类号: 


版权所有 © 《北京航空航天大学学报》编辑部
通讯地址:北京市海淀区学院路37号 北京航空航天大学学报编辑部 邮编:100191 E-mail:jbuaa@buaa.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发