加速退化试验多关键性能参数相关性失效的产品可靠性评估模型

发布时间：2018-09-14 14:21

【摘要】：日新月异的科技发展推动越来越多的产品有着高可靠和长寿命的特征,由此对高长寿命产品的可靠性评估成了一个重要的课题。然而,在相对短试验期内高长寿命产品几乎不可能失效,甚至会出现"零失效"现象,由此基于失效数据很难对产品可靠性进行评估。加速退化试验(Accelerated Degradation Test,ADT)有效的使用了产品的性能退化数据中所包含的信息并对产品的可靠性进行评估和寿命进行预测。但目前关于加速退化试验的研究大多局限于单个关键性能参数。然而,产品可能会有多个失效机制,即就是说产品的多个性能参数会同时退化,并且多个性能参数退化数据相互间具有一定的相关性。因此,本文考虑产品的多元性能退化数据及其相关性,结合数理统计学和Copula函数相关理论对加速退化试验中产品的可靠性评估模型进行研究。首先,简述了论文选题的研究背景和意义,可靠性试验方法的发展进程;加速退化试验可靠性统计模型国内外研究现状;以及加速退化试验中退化数据的统计分析方法的研究现状。其次,对Copula函数的相关理论(如Copula函数的定义、Sklar定理、Copula函数选择准则以及参数估计等)进行了介绍;为多元加速退化试验可靠性模型的建立奠定了一定的理论基础。考虑多元性能参数的产品可靠性评估问题,基于退化量服从不同分布的一元性能参数的产品退化量分布模型,建立了几种多元加速退化试验加速应力可靠性评估模型,包括了独立失效下、相关性失效下和非竞争关系下多元加速退化试验加速应力可靠性评估模型。基于Copula函数考虑多性能参数相关性失效在正常应力水平下的产品可靠性评估问题,结合数理统计学和Copula函数相关理论,建立了加速退化试验多关键性能参数相关性失效在常应力下的产品可靠性评估模型,包括加速退化试验各性能参数退化轨迹的拟合;多元加速退化试验性能参数退化量分布的推断;多性能参数退化量间的相关结构推断;产品常应力水平下的当量化可靠性评估,并通过算例验证了可靠性评估模型的可行性。基于性能数据非竞争关系下多元加速退化试验,探讨了多元退化量在非竞争关系下产品常应力当量化可靠性评估模型,并通过算例验证了所给出的模型的可行性。
[Abstract]:With the rapid development of science and technology, more and more products have the characteristics of high reliability and long life. However, in the relatively short test period, it is almost impossible to fail and even "zero failure" will occur, so it is difficult to evaluate the product reliability based on the failure data. The accelerated degradation Test (Accelerated Degradation Test,ADT) effectively uses the information contained in the product performance degradation data and evaluates the reliability and lifetime of the product. However, most of the current studies on accelerated degradation tests are limited to a single key performance parameter. However, the product may have multiple failure mechanisms, that is to say, the multiple performance parameters of the product will degenerate at the same time, and there is a certain correlation between the degradation data of multiple performance parameters. Therefore, in this paper, considering the multivariate performance degradation data and their correlation, combined with mathematical statistics and Copula function correlation theory, the reliability evaluation model of accelerated degradation test is studied. Firstly, the research background and significance of the selected topic, the development process of reliability test method, the research status of reliability statistical model of accelerated degradation test at home and abroad are briefly introduced. And the research status of statistical analysis methods of degradation data in accelerated degradation test. Secondly, the related theories of Copula function (such as the definition of Copula function and the criterion of parameter estimation) are introduced, which lays a theoretical foundation for the reliability model of multivariate accelerated degradation test. Considering the problem of product reliability evaluation with multivariate performance parameters, based on the model of product degradation quantity distribution from different uniform performance parameters, several kinds of accelerated stress reliability assessment models for multivariate accelerated degradation tests are established. The reliability evaluation model of accelerated stress in multivariate accelerated degradation test under independent failure, correlation failure and non-competitive relationship is included. Based on Copula function, the reliability evaluation problem of multi-performance parameter correlation failure under normal stress level is considered, combined with mathematical statistics and Copula function correlation theory. A product reliability evaluation model for accelerated degradation test under constant stress is established, which includes fitting the degradation trajectory of each performance parameter of accelerated degradation test. The estimation of the distribution of the degradation quantity of the performance parameters, the correlation structure inference between the degradation quantities of the multi-performance parameters, the time-quantization reliability evaluation under the constant stress level of the product, and the feasibility of the reliability evaluation model are verified by an example. Based on the multivariate accelerated degradation test under the non-competitive relation of performance data, the quantitative reliability evaluation model of product constant stress under non-competitive relation is discussed, and the feasibility of the proposed model is verified by an example.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F224

【相似文献】