系统工程与电子技术 ›› 2020, Vol. 42 ›› Issue (11): 2661-2668.doi: 10.3969/j.issn.1001-506X.2020.11.31

• 可靠性 • 上一篇    

二元相关退化系统可靠性分析及剩余寿命预测

杨志远1(), 赵建民1(), 李俐莹2(), 程中华1(), 郭驰名1()   

  1. 1. 陆军工程大学装备指挥与管理系, 河北 石家庄 050003
    2. 河北科技大学信息学院, 河北 石家庄 050000
  • 收稿日期:2019-10-24 出版日期:2020-11-01 发布日期:2020-11-05
  • 作者简介:杨志远(1990-),男,博士研究生,主要研究方向为装备维修保障理论与应用、系统退化建模。E-mail:yzy_sjz90@126.com|赵建民(1962-),男,教授,博士,主要研究方向为装备维修保障理论、故障预测与健康管理。E-mail:jm_zhao@hotmail.com|李俐莹(1990-),女,博士研究生,主要研究方向为智能优化算法。E-mail:liliying@hebust.edu.cn|程中华(1972-),男,教授,博士,主要研究方向为装备保障理论与技术。E-mail:czh7211@yahoo.com.cn|郭驰名(1972-),男,讲师,博士,主要研究方向为视情维修优化、退化分析与寿命预测。E-mail:guochiming@nudt.edu.cn
  • 基金资助:
    国家自然科学基金(71871220);国家自然科学基金(71871219)

Reliability analysis and residual life estimation of bivariate dependent degradation system

Zhiyuan YANG1(), Jianmin ZHAO1(), Liying LI2(), Zhonghua CHENG1(), Chiming GUO1()   

  1. 1. Department of Management Engineering, Army Engineering University, Shijiazhuang 050003, China
    2. School of Information, Hebei University of Science and Technology, Shijiazhuang 050000, China
  • Received:2019-10-24 Online:2020-11-01 Published:2020-11-05

摘要:

退化相关性和个体差异对二元退化系统可靠性有直接影响,针对该问题,在退化过程模型基础上,建立了相应的系统可靠度和剩余寿命预测模型。首先同时考虑个体退化过程和相关性差异,采用随机参数的Gamma过程和Copula函数建立系统二元相关退化模型,为提高模型适用性,随机参数采用非共轭先验分布假设。在此基础上,分析随机参数对系统可靠度影响,提出基于贝叶斯理论的剩余寿命预测方法。利用马尔可夫链蒙特卡罗(Markov Chain Monte Carlo, MCMC)方法对模型未知参数进行估计。案例分析结果说明了在此类系统可靠性估计时考虑个体差异的必要性,也验证了该剩余寿命预测方法的精确性。

关键词: 可靠性, 剩余寿命, 二元相关退化过程, 贝叶斯理论, Copula函数

Abstract:

Degradation dependency and individual difference have a direct influence on the reliability of bivariate degradation system. To solve this problem, based on the degradation process model, the corresponding system reliability and residual life prediction models are developed. Firstly, considering the individual differences of the degradation process and dependency simultaneously, a bivariate dependent degradation model of the system is developed using the Gamma process and the Copula function with random parameters, and to improve the applicability of the model, the assumption of non-conjugate prior distribution is adopted for random parameters. On this basis, the influence of stochastic parameters on system reliability is analyzed, and a residual life prediction method based on the Bayesian theory is proposed. The unknown parameters of the model are estimated by Markov chain Monte Carlo (MCMC) method. The case analysis shows the necessity of considering individual differences in reliability estimation of such systems, and also verifies the accuracy of the residual life prediction method.

Key words: reliability, residual life, bivariate dependent degradation process, Bayesian theory, Copula function

中图分类号: