%0 Journal Article
%T 基于Kriging模型的小子样失效数据可靠性分析<br>Reliability Analysis for Failure Data of Small Samples Based on Kriging Model
%A 任博
%A 吕震宙
%A 刘超
%A 段飞蛟
%J 机械科学与技术
%D 2015
%X 针对高可靠性航空柱塞泵现场故障数据稀少,难以准确确定数据分布类型的问题,提出采用Kriging方法建立部件故障不确定性的精确描述模型,挖掘部件可靠性信息,并基于不确定性描述的隐式Kriging模型进行蒙特卡洛抽样,产生与原始数据统计规律一致的大容量样本,为基于蒙特卡洛方法的结构系统可靠性分析奠定基础。此外,为衡量所提模型与原始数据统计分布特性的差异,提出基于累积分布函数的单位面积期望误差测度。测度分析结果表明,Kriging方法所建的部件故障不确定性模型比传统方法更准确。以航空柱塞泵为例,进行可靠性数据分析,验证所提方法的正确与高效。<br>For the sparse failure data of high reliability products used in the aircraft design,it is difficult to determining the distribution model of data,which can reflect the statistic characteristics of components' failure information. A novel method is proposed to describe the uncertainty of failure data based on Kriging model in order to mine the reliability information of component. Then, a large number of samples based on the Kriging model can be drawn, and the larger sample completely reflects the properties of the original small sample, which is convenient for system reliability analysis based on Monte Carlo sampling. Meanwhile, a new measure index about Cumulative Distribution Function (CDF) is established to measure the statistical characteristic difference between the uncertainty model based on Kriging method and the original failure data. Take the hydraulic pump as an example, the result of measuring analysis illustrates the proposed method is more exact to describe the statistical characteristic of the original failure data than traditional methods
%K 克里金
%K 可靠性
%K 可靠性数据分析
%K 逆变换抽样&lt
%K br&gt
%K aircraft
%K covariance matrix
%K data mining
%K design
%K errors
%K flowcharting
%K functions
%K maximum likelihood estimation
%K measurements
%K normal distribution
%K reliability analysis
%K stochastic models
%K vectors
%K Weibull distribution
%K data analysis
%K Kriging model
%K monte carlo methods
%K reliability
%K sampling
%U http://journals.nwpu.edu.cn/jxkxyjs/CN/abstract/abstract6225.shtml