| 引用本文: | 黄秀平,周经伦,程龙.可靠度和费用约束下的加速老炼试验时间优化.[J].国防科技大学学报,2017,39(1):109-114.[点击复制] |
| HUANG Xiuping,ZHOU Jinglun,CHENG Long.Optimization of accelerated burn-in time with reliability and cost constraints[J].Journal of National University of Defense Technology,2017,39(1):109-114[点击复制] |
|
| |
|
|
| 本文已被：浏览 7790次 下载 6680次 |
| 可靠度和费用约束下的加速老炼试验时间优化 |
| 黄秀平, 周经伦, 程龙 |
|
(国防科技大学 信息系统与管理学院， 湖南 长沙 410073)
|
| 摘要: |
| 为了提高产品实际使用可靠度，减少老炼试验和保修费用，基于效用函数研究了具有早期失效的产品在可靠度和费用约束下的加速老炼试验时间优化方法。考虑产品寿命分布参数的不确定性，利用自助法估计参数的先验分布，然后通过仿真方法获得实际使用可靠度和费用的联合效用函数。以联合效用函数值最大为优化目标建立加速老炼试验时间优化模型，得到最优的加速老炼试验时间。结合对数正态分布产品在温度应力下的加速老炼试验数值示例，阐述上述方法的应用，并对参数进行敏感性分析，结果表明提出的加速老炼试验方法不但能缩短产品的试验时间，还可以最大程度同时满足费用和可靠性的要求。 |
| 关键词: 老炼试验 优化 效用函数 加速模型 对数正态分布 |
| DOI：10.11887/j.cn.201701017 |
| 投稿日期：2015-09-08 |
| 基金项目:国家自然科学基金资助项目(61304221) |
|
| Optimization of accelerated burn-in time with reliability and cost constraints |
| HUANG Xiuping, ZHOU Jinglun, CHENG Long |
|
(College of Information System and Management, National University of Defense Technology, Changsha 410073, China)
|
| Abstract: |
| In order to improve the field reliability of products and reduce the total cost of burn in procedure and warranty, an optimization method of accelerated burn-in time for products with infant mortalities within the reliability and cost constrains was developed on the basis of utility function. Considering the uncertainties with respect to the lifetime distribution parameters of the product, the prior distributions of these parameters were derived by using the bootstrap method, and then the joint utility function combining field reliability with cost was constructed through simulation technique. Choosing the maximum of joint utility as objective, the optimal accelerated burn-in time model was established, and the optimal burn-in duration was derived. The application of the above method was illustrated explicitly by a numerical example in which a product with log-normal distribution was burned in under elevated temperature condition, and the parameters′ sensitivity analysis was also conducted. The results show that the proposed method not only can shorten the burn in duration, but also can meet the requirements for cost and reliability simultaneously to the utmost. |
| Keywords: burn-in optimization utility function acceleration model log-normal distribution |
|
|
