人造卫星系统可靠性建模与分析研究

发布时间：2018-04-09 02:27

本文选题：Johnson平均故障秩法　切入点：Kaplan-Meier方法　出处：《南京航空航天大学》2016年硕士论文

【摘要】：人造卫星均属于长寿命、高可靠性产品,因此能够搜集整理到系统的失效样本非常有限。为了利用有限的人造卫星失效样本,对人造卫星系统进行可靠性建模与分析,以此为人造卫星系统的可靠性设计及试验给予参考和指导,就必须寻求在小样本的情况下,有效适用的可靠性建模分析的统计方法。非参数方法能够很真实的反映原始样本的情况,本文通过对现有的小样本复杂系统的非参数可靠性分析方法的比较研究,确定分别使用Johnson平均故障秩法和Kaplan-Meier方法对人造卫星系统进行可靠性建模,并且利用重非参数Bootstrap的方法进行可靠性评估分析。参数方法以其灵活性,在可靠性分析研究方面受到了广泛的应用。本文对人造卫星系统进行参数可靠性建模分析的时候,采用了威布尔分段模型,该模型能够拟合各种形式的数据样本。具体建模分析过程为:首先,对非参数方法得到可靠度的失效率的分析,确定了需要使用的分段模型;随后,将可靠度通过威布尔变换在WPP图上拟合,并且使用最小二乘法对模型的参数进行估计。最后分析拟合结果,利用重非参数Bootstrap的方法评估模型的适用性。
[Abstract]:Satellites are long-life, high-reliability products, so the ability to collect and collate system failure samples is very limited.In order to use a limited number of satellite failure samples to model and analyze the reliability of Sputnik systems, and to provide reference and guidance for the reliability design and test of Sputnik systems, it is necessary to seek for small samples.Effective and applicable statistical method for reliability modeling and analysis.The nonparametric method can reflect the situation of the original sample very truthfully. In this paper, we compare the existing non-parametric reliability analysis methods of the complex system with small sample.The reliability model of satellite system is established by using Johnson average fault rank method and Kaplan-Meier method respectively, and the reliability evaluation analysis is carried out by using the method of heavy nonparametric Bootstrap.Parameter method is widely used in reliability analysis for its flexibility.In this paper, the Weibull piecewise model is used to model and analyze the parameter reliability of the satellite system, which can fit all kinds of data samples.The concrete modeling and analysis process is as follows: firstly, the failure rate of the non-parametric method to obtain reliability is analyzed, and the piecewise model that needs to be used is determined. Then, the reliability is fitted on the WPP diagram by Weibull transform.And the least square method is used to estimate the parameters of the model.Finally, the fitting results are analyzed and the applicability of the model is evaluated by using the method of heavy nonparametric Bootstrap.
【学位授予单位】：南京航空航天大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：V474

【参考文献】