复杂系统可靠度置信区间估计

Estimation Reliability Confidence-intervals for Complex-systems

针对复杂系统可靠性试验非常少甚至没有做的情况 ,提出了基于单元信息进行可靠性综合的方法。该方法不需假设系统或单元产品的寿命服从某一分布 ,减少了因寿命分布选择不当所造成的可靠性和方差的误差。在获得单元可靠性的均值和方差的基础上 ,利用系统可以分解为单一的串联或并联关系 ,通过逐步综合获得复杂系统的可靠性均值和方差估计值。利用系统信息熵原理 ,将部件的试验数据折合为系统的试验数据 ,获得系统的子样数。由此提出了小子样下的系统可靠性置信区间估计新方法 ,该方法只假设系统可靠性估计服从正态或对数正态分布。新方法使用限制少 ,计算简单 。

Reliability comprehensive methods based on components information are presented when there is no complex system reliability information.The methods do not necessary to assume the components time-to-failure distribution.Thus,error of reliability and variance decreases.Complex systems reliability mean and variance can be acquired based on components reliability information,at the same time,the only restriction is that it must be possible to decompose the system into series or parallel elements. The test data of components are converted into the test data of the complex system based on the information enthropy theory. Suppose the system reliability estimation follows normal or log-normal distribution,the system reliability confidence intervals can be evaluated according to samples.The application of this approach to small sample complex system will result in few restrictions and easily to conpate.