基于贝叶斯信息融合的复杂系统可靠性增长分阶段评价方法
发布时间：2019-03-17 12:19
【摘要】：大型复杂系统的质量关系到地区或国家在某一个大型项目上的成败,甚至是在某一个领域的国际地位,同时关系到生命财产安全。可靠性是质量的固有属性之一,可靠性贯穿产品或系统的研制、定型到投入使用的整个过程,因此必须重视可靠性管理。研制阶段通过可靠性增长试验实现可靠性增长,进行可靠性评估时常用的有Duane模型,AMSAA模型和Bayes可靠性评估方法等。大型复杂系统,有成千上万的不同类型的器件组成,本文研究了复杂系统增长试验连续进行,相似产品较少,系统的失效机理也很难掌握情况下,在可靠性增长过程中出现突变点时,突变点的辨识和系统可靠性增长评估问题,重点对突变点导致的增长速度减缓情况开展研究。通过增长趋势图建立分段模型辨识突变点,在此基础上,基于最大熵方法确定Bayes先验分布,通过某大型装置安装集成阶段的数据进行验证,证明了方法的有效性和可用性。首先介绍了研究背景,研究目的和意义等内容。在第二章介绍了可靠性的相关概念和指标,以及可靠性增长管理的模型和方法的综述。第三章分析了可靠性增长突变的原因,建立了基于增长趋势的可靠性增长分段模型;能够体现纠正措施对增长特性的影响,具有较为广泛的应用范围。第四章在多阶段系统可靠性增长评估研究的基础上,建立了基于最大熵方法的Bayes可靠性评估模型;通过案例证明了模型的有效性。通过分析认为,辨识突变点有利于本文研究的开展;最大熵方法在进行Bayes模型的先验参数求解时的有效性和方便性;Bayes模型能够有效对多阶段数据信息进行融合。本文主要得到如下结论:(1)建立的分段模型适用于突变点导致的可靠性增长速度加快,减缓和多突变点下的增长速度的不确定变化。(2)建立的分段模型能更明确可靠性增长速度变化特点,可以更好地了解系统可靠性增长的变化规律;(3)建立的系统可靠性增长评估方法用于融合多阶段的故障信息,得到更准确的评估结果。
[Abstract]:The quality of large-scale complex system is related to the success or failure of a region or country in a large-scale project, even to the international status in a certain field, and also to the safety of life and property. Reliability is one of the inherent attributes of quality. Reliability runs through the whole process of product or system development, setting up to put into use, so it is necessary to pay attention to reliability management. In the development stage, reliability growth is realized by reliability growth test. Duane model, AMSAA model and Bayes reliability evaluation method are commonly used in reliability evaluation. There are thousands of different types of devices in large-scale complex systems. In this paper, the growth tests of complex systems are carried out continuously, the similar products are few, and the failure mechanism of the system is difficult to grasp. In the process of reliability growth, the identification of catastrophe points and the evaluation of system reliability growth are discussed, and the research on the deceleration of the growth rate caused by the mutation points is emphasized. A piecewise model based on the growth trend graph is used to identify the mutation points. On the basis of this, the prior distribution of Bayes is determined based on the maximum entropy method. The validity and usability of the method are verified by the data of the installation and integration stage of a large device. Firstly, the background, purpose and significance of the research are introduced. In the second chapter, the related concepts and indicators of reliability are introduced, and the models and methods of reliability growth management are summarized. In the third chapter, the reason of the sudden change of reliability growth is analyzed, and the subsection model of reliability growth based on growth trend is established, which can reflect the influence of corrective measures on the growth characteristics and has a wide range of applications. In chapter 4, based on the research of multi-stage system reliability growth evaluation, the Bayes reliability evaluation model based on the maximum entropy method is established, and the validity of the model is proved by a case. Through the analysis, it is concluded that identifying the mutation points is beneficial to the research in this paper; the maximum entropy method is effective and convenient in solving the prior parameters of the Bayes model; and the Bayes model can effectively fuse the multi-stage data information. The main conclusions of this paper are as follows: (1) the proposed piecewise model is suitable for the acceleration of reliability growth caused by mutation points. (2) the piecewise model can clarify the change characteristics of reliability growth rate more clearly, and can better understand the change rule of system reliability growth rate; (2) the variable law of system reliability growth can be better understood by the piecewise model which can alleviate the uncertain change of growth speed under multi-mutation points; (3) the proposed reliability growth assessment method is used to fuse multi-stage fault information and obtain more accurate evaluation results.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F124;F224
本文编号：2442295
[Abstract]:The quality of large-scale complex system is related to the success or failure of a region or country in a large-scale project, even to the international status in a certain field, and also to the safety of life and property. Reliability is one of the inherent attributes of quality. Reliability runs through the whole process of product or system development, setting up to put into use, so it is necessary to pay attention to reliability management. In the development stage, reliability growth is realized by reliability growth test. Duane model, AMSAA model and Bayes reliability evaluation method are commonly used in reliability evaluation. There are thousands of different types of devices in large-scale complex systems. In this paper, the growth tests of complex systems are carried out continuously, the similar products are few, and the failure mechanism of the system is difficult to grasp. In the process of reliability growth, the identification of catastrophe points and the evaluation of system reliability growth are discussed, and the research on the deceleration of the growth rate caused by the mutation points is emphasized. A piecewise model based on the growth trend graph is used to identify the mutation points. On the basis of this, the prior distribution of Bayes is determined based on the maximum entropy method. The validity and usability of the method are verified by the data of the installation and integration stage of a large device. Firstly, the background, purpose and significance of the research are introduced. In the second chapter, the related concepts and indicators of reliability are introduced, and the models and methods of reliability growth management are summarized. In the third chapter, the reason of the sudden change of reliability growth is analyzed, and the subsection model of reliability growth based on growth trend is established, which can reflect the influence of corrective measures on the growth characteristics and has a wide range of applications. In chapter 4, based on the research of multi-stage system reliability growth evaluation, the Bayes reliability evaluation model based on the maximum entropy method is established, and the validity of the model is proved by a case. Through the analysis, it is concluded that identifying the mutation points is beneficial to the research in this paper; the maximum entropy method is effective and convenient in solving the prior parameters of the Bayes model; and the Bayes model can effectively fuse the multi-stage data information. The main conclusions of this paper are as follows: (1) the proposed piecewise model is suitable for the acceleration of reliability growth caused by mutation points. (2) the piecewise model can clarify the change characteristics of reliability growth rate more clearly, and can better understand the change rule of system reliability growth rate; (2) the variable law of system reliability growth can be better understood by the piecewise model which can alleviate the uncertain change of growth speed under multi-mutation points; (3) the proposed reliability growth assessment method is used to fuse multi-stage fault information and obtain more accurate evaluation results.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F124;F224
【参考文献】
相关期刊论文 前1条
1 陈希孺;变点统计分析简介 (Ⅲ)极大似然法、累计次数法、Bayes法[J];数理统计与管理;1991年03期
相关硕士学位论文 前2条
1 廖小波;机床故障率浴盆曲线定量化建模及应用研究[D];重庆大学;2010年
2 孙志平;复杂系统可靠性增长管理与评价方法研究[D];电子科技大学;2013年
，本文编号：2442295
本文链接：https://www.wllwen.com/jingjilunwen/jingjiguanlilunwen/2442295.html
