摘要
小子样产品可靠性的Bayes评估中通常需要用到主观经验信息,如可靠度均值或可信区间等,这些信息属于不完全先验信息,利用这些信息通常无法确定可靠性分布函数。基于Bayes理论,以贝塔分布作为先验分布类型,利用最大熵原理将不完全信息转化为完全型先验信息,得到产品可靠性的先验分布,再结合观测数据,利用Bayes公式得到产品可靠性后验分布。从仿真算例可以看出,给出的方法能够有效地处理不完全先验信息,提高产品可靠性评估的效率。
In Bayesian reliability evaluation of small-sample product, subjective experience information is often used, such as mean of reliabihty and credibility interval, which belong to incomplete prior information. It＇s difficult to obtain the reliability distribution function by using this information only. Taking Beta distribution prior distribution, we changed the incomplete information into complete prior information by using the principle of maximum entropy and Bayes theory. The prior distribution of reliability is obtained. Then the posterior distribution is available via Bayesian formula. The simulation shows that the method is effective in processing incomplete prior information, and the efficiency of product reliability evaluation can be improved.
出处
《电光与控制》
北大核心
2008年第3期94-96,共3页
Electronics Optics & Control
关键词
小样本产品
可靠性评估
不完全先验信息
最大信息熵
small-sample product
reliability evaluation
partial prior information
maximum information entropy
