基于异常值检验的Bayesian方法验前信息可信度计算

由于采用了验前信息,Bayesian方法为小子样条件下的装备试验与评价提供了可行的理论依据,从而在装备试验与评价领域得到了广泛的应用。然而,由于验前信息具有多源性及主观性,当采用其进行Bayesian统计推断时,可能会导致Bayesian统计推断的结果偏差较大,使得决策风险增加。因此,研究Bayesian方法中验前信息的可信度显得至关重要。在现场试验样本信息已知的前提下,给出了验前信息可信度的定义,提出一种基于异常值检验的验前信息可信度计算方法,并基于Matlab数学软件,通过仿真思路实现了算法求解。最后,以某雷达探测距离试验为例,证明方法具有科学合理性。

Prior Information Creditability Calculation of Bayesian Method Based on

Using the prior information, Bayesian method provided a feasible theoretic basis for small-sample teat and was widely used in the area of equipment test. Nevertheless, using prior information to make Bayesian statistical inference may lead to large deviation and high risk due to the multi-source and subjectivity of prior information. Therefore, it is important to do the research in prior information creditability of Bayesian method. Under the known field testing information, this paper proposes a simulation method of prior information creditability based on outliers test and realizes the algorithm by Matlab. Finally, the rationality of the method is proved by an equipment testing instance.