航空学报 > 2014, Vol. 35 Issue (12): 3350-3357   doi: 10.7527/S1000-6893.2014.0010

融合加速退化和现场实测退化数据的剩余寿命预测方法

王浩伟1, 徐廷学2, 赵建忠2   

  1. 1. 海军航空工程学院 研究生管理大队, 山东 烟台 264000;
    2. 海军航空工程学院 兵器科学与技术系, 山东 烟台 264000
  • 收稿日期:2013-12-09 修回日期:2014-03-03 出版日期:2014-12-25 发布日期:2014-03-10
  • 通讯作者: 徐廷学 男, 博士, 教授, 博士生导师.主要研究方向: 装备综合保障理论与技术. Tel: 0535-6635061 E-mail: xtx-yt@163.com E-mail:xtx-yt@163.com
  • 作者简介:王浩伟 男, 博士研究生.主要研究方向: 装备可靠性评估与寿命预测. Tel: 0535-6635061 E-mail: wyg2010123@126.com;赵建忠 男, 博士, 讲师.主要研究方向: 备件消耗预测与优化配置技术. Tel: 0535-6635061 E-mail: zjznavy@163.com
  • 基金资助:

    国家自然科学基金(61273058)

Residual Life Prediction Method Fusing Accelerated Degradation and Field Degradation Data

WANG Haowei1, XU Tingxue2, ZHAO Jianzhong2   

  1. 1. Graduate Students' Brigade, Naval Aeronautical and Astronautical University, Yantai 264000, China;
    2. Department of Ordnance Science and Technology, Naval Aeronautical and Astronautical University, Yantai 264000, China
  • Received:2013-12-09 Revised:2014-03-03 Online:2014-12-25 Published:2014-03-10
  • Supported by:

    National Natural Science Foundation of China (61273058)

摘要:

针对先验信息为加速退化数据的情况,提出了利用非共轭先验分布进行Bayesian统计推断的剩余寿命预测方法.不预先假定Wiener过程参数值的分布类型,利用加速系数将加速应力下的参数值折算到工作应力水平下,进而使用Anderson-Darling方法确定参数值的最优拟合分布类型.在对参数值进行折算时,根据周源泉提出的理论对Wiener过程参数与加速应力之间的关系进行了推导.参数估计时,通过极大似然法得到超参数的估计值,利用WinBUGS软件实现Markov Chain Monte Carlo仿真得到参数的后验均值.通过某型军用电连接器寿命预测实例验证了所提方法的实用价值和研究意义,结果表明本方法可有效解决先验信息为加速退化数据时进行剩余寿命预测的难题.

关键词: 剩余寿命, 加速退化数据, 非共轭先验分布, Wiener过程, 加速系数

Abstract:

To address the situation that prior information is accelerated degradation data, a residual life prediction method based on Bayesian inference with non-conjugate prior distribution is proposed. Without presupposing the parameters' distribution types of a Wiener process, the parameters at accelerated stress levels are transformed into normal use stress level through acceleration factors and then the optimally fitted distribution types are determined using Anderson-Darling method. When transforming the parameter values, the relationships between parameters of the Wiener process and accelerated stress are deduced according to the theory proposed by Zhou Yuanquan. The estimates of hyper-parameters can be obtained by maximum likelihood methods, and posterior means of parameters can be inferred by Markov Chain Monte Carlo using WinBUGS software. The practical value and research significance of this study are demonstrated through a certain missile electrical connector lifetime prediction example. The results show that the proposed method can effectively solve the problem of residual life prediction when prior information is accelerated degradation data.

Key words: residual life, accelerated degradation data, non-conjugate prior distribution, Wiener process, acceleration factor

中图分类号: