中文摘要：

为了解决发动机在现场小子样成败型区间检测数据下贮存寿命预测精度不高的问题,考虑到型号研制的继承性,可收集到大量相似型号发动机的贮存信息,提出了基于相似型号信息融合的贮存寿命预测方法。首先,利用配分布曲线法,将各来源相似型号成败型区间测试信息转变为相应的贮存可靠度函数;然后,根据对数似然准则,确定相似型号产品信息的可信度,并确定融合贮存可靠度函数;再用自助法,得到发动机可靠贮存寿命的区间估计。结果表明,对数似然准则所确定的融合权重可体现型号之间的相似性;通过融合相似型号贮存信息,可提高发动机贮存寿命的预测精度。

英文摘要：

In order to improve the precision of storage life prediction in small samples,the prediction method based on similar products information fusion is proposed.On the one hand,the locale storage information for new engine usually be small samples of binary data gained through interval inspection.On the other hand,considering the inherited characteristics of development,lots of storage information of similar products can be gathered.Firstly,the fit-ldistribution-lmethod was used to change binary data for each similar product into storage reliability function.Secondly,the likelihood rule was proposed to get the fusion weight.Thirdly,the interval estimation of reliable storage life was gained by Bootstrap method.Theory and simulation results showed that the logarithm likelihood rule can validly embody the similarity between products and the prediction precision can be efficiently improved through information fusion.