维修性先验信息的融合方法

在利用Bayes小子样方法进行维修性指标验证时，针对维修性先验信息与现场维修信息为异总体的可能性较大的问题，分别建立了前期试验阶段维修时间信息向现场试验信息折合的内积模型和相似装备维修时间信息向待评装备维修时间信息折合的线性模型。采用随机加权法将先验信息转换为先验分布，并以现场信息为基准，根据各先验信息与现场信息分布模型中均值参数的差异程度来确定各先验分布的权重，然后通过计算综合先验分布与假设的正态分布的概率密度误差，对其融合后为正态分布的假设进行了检验，同时也提出了准确计算分布模型参数的方法。对某舰船装备维修性先验信息的折合与融合的实例表明了方法的有效性。

Fusion method of prior maintainability information

In the maintainability index verification by the Bayesian small sample method, to solve the problem that the prior maintainability information and on-site repair information are very likely different populations, the inner product model that the repair time information in the prior test stage converts into on-site repair information and the linear model that the similar equipment repair time information converts into that of the equipment which is ready to be evaluated are established. The random weighting method is used to convert a piece of prior information into a prior distribution. With the on-site information being considered as the standard, and according to the discrepancy degree between the mean parameter of each prior information distribution and that of the on site information distribution model, the weight of the prior distribution is determined. Then by calculating the probability density errors between the integrated prior distribution and the hypothetic normal distribution, the hypothesis that the fused integrated prior distribution is normal distribution is tested, and the method to accurately calculate the distribution parameters is also put forward. Prior maintainability information conversion and fusion of some ship equipment shows that the method is effective.