可靠性、维修性和可用性是描述复杂装置质量属性的三个重要指标。在某大型装置的现场故障和维修数据的经验分析基础上，运用分段线性模型确定了装置可靠性改进趋势。针对装置现场故障和维修数据的小样本问题，给出了装置现场可靠性和维修性的贝叶斯模型，并用WinBUGS软件对其进行了求解计算。继而分析给出了装置可用性最恰当描述——任务可用度的闭合表达形式。最后，通过与极大似然法进行对比，发现应用贝叶斯方法分析该大型装置现场可靠性更有效。
Reliability, maintainability and availability are three important attributes to describe the quality of a large-scale device. A certain large-scale device is in the prototype testing phase and clearly exhibits reliability growth. However, the high environmental stress and operation cost result in the difficulties to collect field data. Moreover, the downtime constraint should be considered when the availability analysis is carried out. Therefore, it is very difficult to carry out device reliability, maintainability and availability assessments. In this paper, the empirical analysis is first carried out based on the field failure and maintenance data, then the device reliability improvement trend is determined with a piecewise linear model. For the small sample question of the field failure and maintenance data, the Bayesian method is introduced, and the field reliability and maintainability parameters are calculated by the soft WinBUGS. Then the most appropriate description of the system availability—mission availability with closed expression form is presented. In order to illustrate the effectiveness of the method, the field data from June 4, 2007 to May 17, 2010 has been couected. Through the case study, two reliability change-points are determined and thus the device reliability and maintainability are analyzed with Bayesian method. As a result, the average relative error of Bayesian and maximum likelihood methods are 0.42% and 8.78% respectively. Then it can be concluded that using the Bayesian method to carry out field reliability analysis is more effective compared to the maximum likelihood method.
[1] 陈志军, 王前程, 陈云霞. 基于寿命分布和贝叶斯的加速因子确定方法[J]. 系统工程与电子技术, 2015, 37(5): 1225-1228.
[2] Aktekin T, Caglar T. Imperfect debugging in software reliability: A Bayesian approach[J]. European Journal of Operational Research, 2013, 227 (1): 112-121.
[3] 张瑞军, 张路路, 王晓伟,等. 区间三角模糊多态贝叶斯网络可靠性分析方法研究[J]. 中国机械工程, 2015, 26(8): 1092-1097.
[4] 赵炤, 金光, 钟征,等. 激光装置可靠性仿真环境研究[J]. 计算机仿真, 2004, 21(12): 245-248.
[5] 胡明礼, 张彩芬, 朱建军. 基于贝叶斯网络推理的双重不确定信息集结模型[J]. 中国管理科学, 2014, 22(2): 135-141.
[6] Zhu Benjin, Frangopol D M. Reliability assessment of ship structures using Bayesian updating[J]. Engineering Structures, 2013, 56 (11): 1836-1847.
[7] 孙权, 钟征, 赵建印,等. 神光III原型装置系统可靠性评价指标[J]. 激光与光电子学进展, 2004, 41(2): 6-9.
[8] 刘丽文, 郭祥雷. 武器装备保障供应链中基于绩效的整体外包机制[J]. 中国管理科学, 2009, 17(6): 91-97.
[9] 赵建印, 孙权, 周经伦,等. 基于加速退化数据的金属化膜脉冲电容器可靠性分析[J]. 强激光与粒子束, 2006, 18(9): 1495-1498.
[10] 彭宝华, 孙权, 赵建印. 神光III片状放大器系统可靠性仿真研究[J]. 激光与光电子学进展, 2006, 43(1): 19-23.
[11] 应新雅, 张志华. 产品可靠性新型顺序约束及评估算法改进[J]. 系统工程与电子技术, 2015, 7(2): 468-472.
[12] 邢云燕, 武小悦. 成败型系统变总体下的Bayes可靠性试验鉴定方法[J]. 系统工程理论与实践, 2011, 31(2): 323-327.
[13] 彭宝华, 周经伦, 孙权. 神光III激光装置能源模块的可靠性增长评定[J]. 高电压技术, 2008, 34(10): 2150-2154.
[14] Zhou Yu, Kou Gang, Ergu D, et al. Mission availability for bounded cumulative downtime system. Plos One, 2013, 8(7): e65375.
[15] 祝硕, 许保光, 刘世骞. 周转件构成的可修复系统的可靠性分析[J]. 中国管理科学, 2014, 22(7): 67-75.
[16] Zomorrodi A, Maranas C. Coarse-grained optimization-driven design and piecewise linear modeling of synthetic genetic circuits[J]. European Journal of Operational Research, 2014, 237(2): 665-676.
[17] Atwood C, Kelly D. The binomial failure rate common-cause model with WinBUGS[J]. Reliability Engineering & System Safety, 2009, 94(5): 990-999.
[18] Ali S, Aslam M, Kazmi S. A study of the effect of the loss function on Bayes Estimate, posterior risk and hazard function for Lindley distribution[J]. Applied Mathematical Modelling, 2013, 37(8): 6068-6078.
