【作者】 李翔宇；

【导师】 黄洪钟；

【作者基本信息】 电子科技大学 ， 机械工程， 2018， 博士

【摘要】 随着现代科学技术的发展,航天系统越来越朝着复杂化发展,绝大多数的航天系统,例如卫星或者航天器等,在运行过程中存在明显的阶段性特征,其系统结构、功能等随着时间而不断地发生改变,这类系统被称为多阶段任务系统(Phased Mission System,PMS)。相较于传统的单阶段任务系统可靠性建模,PMS可靠性建模需要考虑阶段之间的相关性,因此更为复杂,同时也可以对系统可靠度进行更加精确的评估。目前,随着航天事业的不断发展,PMS可靠性的研究逐渐成为了可靠性研究的热点之一。但是目前在PMS可靠性建模与优化研究中,存在着一个较大的问题,即在PMS建模与优化研究中,绝大多数的研究都假设系统中的元件服从指数分布,从而利用Markov链等方法对PMS中的动态行为进行建模,但是这一假设不符合大多数的工程实际。针对这一问题,本文对系统中元件服从非指数分布的复杂PMS可靠性建模与优化方法进行研究,主要内容和成果如下:(1)有限维修下的多阶段任务系统可靠性建模方法研究针对航天PMS中存在着冷备份及有限维修的动态行为,元件失效/维修时间服从非指数分布,难以用传统的Markov过程进行建模的情况,提出应用半马尔科夫过程(Semi-Markov Process,SMP)来对系统的动态行为进行建模,应用模块化方法减小系统模型的规模,降低建模的复杂程度。同时应用数值近似算法对SMP模型中出现的复杂积分进行高精度近似计算,并将该方法与蒙特卡洛仿真算法进行对比分析,说明了该方法具有较高的计算精度与计算效率。实现对有限维修下的非指数多阶段任务系统的可靠性建模。(2)考虑随机冲击失效的多阶段任务系统可靠性建模方法研究绝大多数的航天PMS,例如卫星/航天器等,长期工作于环境恶劣的外太空,随机出现的宇宙射线所携带的高能带电粒子、重粒子等会对各种电子元器件造成严重影响,导致元件加速退化甚至直接失效,针对这些随机冲击的影响,提出应用马尔科夫更新过程(Markov Regenerative Process,MRGP)对冲击影响下的动态PMS进行建模。考虑随机冲击造成的总剂量效应与单粒子效应等影响,通过引入元件的随机冲击模型,利用近似计算的方法来得到冲击影响下的元件失效分布函数,然后将元件的随机冲击模型嵌入系统MRGP模型中,得到冲击影响下的动态系统模型。最终,通过数值近似计算与MC仿真方法验证,实现对随机冲击影响下的多阶段任务动态系统的可靠性建模,并对随机冲击的参数进行灵敏度分析。(3)多态多阶段任务系统可靠性建模与评估方法研究针对不可修PMS中的元件存在多态的特性,即元件在正常工作与完全失效之外还存在着多种中间状态,本论文提出多态元件的阶段代数与PMS-MMDD模型,通过应用阶段代数,对传统的MMDD逻辑运算规则进行改进,得到多态元件阶段依赖性运算规则。通过对比分析,相较于传统的MMDD模型与PMS-BDD模型,PMS-MMDD模型对不可修多态PMS的建模更加高效。同时,考虑多态元件的状态转移时间服从非指数分布的情况,提出一种基于Markov更新方程的不交和路径概率评估方法,并应用迭代数值方法进行高精度计算分析,实现对不可修多态多阶段任务系统的可靠性建模与评估,并与MC仿真方法进行对比分析。(4)混合备份下的多阶段任务系统可靠性优化提出混合备份下的PMS冗余优化方法。在航天系统中,为了保证较高的可靠性,经常采用大量冗余备份,而现有的PMS冗余优化方法研究中,只对冷备份或者热备份的单一冗余备份方式进行优化分析,而忽略了混合备份的策略。针对这一问题,在本文的研究中,综合考虑三种备份策略,应用SMP与模块化方法,建立混合备份下的PMS优化模型,同时对元件的类型、备份数量以及备份方式进行优化,实现对混合备份下的PMS的可靠性优化。更多还原

【Abstract】 With the developing of modern science and technology,the aerospace equipment grows more and more complex.Most of these aerospace systems are phased mission systems(PMSs).The whole lifetime of the PMSs can be divided into several time durations according to the different system configurations and functions at different time,like the satellites or the spacecraft.Compared the traditional single-phased system,the reliability modelling of the PMS is more complex due to the consideration of the dependency among phases while the system reliability can be assessed more accurately.Nowadays,with the development of the aerospace industry,the reliability research of the PMS has become an emerging research topic.But in the existing research of the PMS modelling and optimization methods,there exists one critical problem.Most of the researches assume that the failure time of the components follow the exponential distributions,so that the system dynamic behaviors can be solved with the Markov chains,but this assumption is not practical.So this dissertation is focused on the modelling and optimization methods of PMS consisting of non-exponential components.The research contributions and innovative outcomes are summarized as follows:(1)Reliability analysis of phased mission system with non-exponential and partially repairable componentsIn practice,many real PMSs,like the satellites,consist of mechanical components or mechatronics whose lifetime follow non-exponential distributions like the Weibull distribution.In this type of system,the Markov process is not capable of modeling the system dynamic behaviors.In this paper,the Semi-Markov Process(SMP)is applied to solve the problem that the components’ lifetime in dynamic systems follows nonexponential distributions.And the modular method is applied to simplify the system model.A numerical approximation algorithm for the SMP is proposed to assess the reliability of the PMSs consisting of non-exponential components.Furthermore,the accuracy and calculation efficiency of the approximation algorithm are explored compared to the Monte Carlo simulation method.(2)Reliability analysis of phased mission system under random shocksMany PMSs,like the satellite or the spacecraft,work in the outer space for a long time.And the energetic charged particle and heavy particle within the randomly occurred cosmic rays have a significant impact on the electronics.These random shocks could lead to the degradation of components,even failure.In this paper,A Markov regenerative process(MRGP)based method is proposed to modelling the PMS under random shocks.Firstly,different shocks models for components are analyzed.Then,the components shocks models are integrated into the system MRGP model.At last,the PMSs under random shocks are analyzed along with the modular method and numerical method.And a sensitivity analysis is also provided for completion(3)Reliability assessment of multi-state phased mission system with non-repairable multi-state componentsIn practical,many systems with multistate components cannot be repaired,such as satellites or the spacecraft working in the outer space.To deal with complex dependency among phases,the phase algebra for multi-state components is proposed.Then,by applying the proposed phase algebra,the MMDD manipulation rules are improved into the phase dependency manipulation rules.Compared to the PMS-BDD model and traditional MMDD method,can do the system modelling process with less nodes and the self-conflicting paths can be automatically cancelled and the system MMDD model can be generated without additional steps.On the other hand,a Markov-renewal equation based method is proposed to deal with the non-exponential multistate components in the evaluation of the path probability.And by the iterative numerical algorithm,the system reliability is assessed by the proposed method and verified by comparing to the MC simulation method.(4)System optimization of phased mission system with mix redundancy strategy.A redundancy optimization method for the PMS with mix redundancy strategy.In the areospace industry,redundancy strategy is usually used to ensure the high reliability.In the existing researches of the PMS redundancy optimization research,only cold standby or the hot standby strategy is considered.In this paper,a mixed redundancy strategy,which is a general case of the hot standby and cold standby,is proposed to optimize the PMS.By applying the SMP and modular method,the optimization model of the PMS with mix redundancy strategy is established.And the component type,redundancy quantity and type of redundancy can be optimized at the same time by an improved genetic algorithm.更多还原