【作者】 王立超；

【导师】 邹云； 于永利；

【作者基本信息】 南京理工大学 ， 控制科学与工程， 2009， 博士

【摘要】 新装备系统投入使用时往往由于子系统间的相互作用而产生瞬时可用度的波动现象,这体现为新装备在投入初期需要磨合,不能快速形成战斗力。对于装备瞬时可用度这种波动现象的研究目前还没有建立合理的指标评价体系。目前对于装备可用度的研究基本围绕在稳态可用度等稳态指标上,它反映了装备瞬时可用度当时间t趋于无穷时的相关性态。随着科技的快速发展,装备的更新和淘汰速度加快,装备的服役期现在可能只有几年或者更短,对装备服役期间的装备瞬时可用度的研究变得更加有意义。本文以工程需求为背景,以当前研究工作存在的问题为主要突破口,从前人已有研究工作出发,初步建立了系统瞬时可用度在一定区间内波动问题研究的基本方法和框架。论文的主要内容如下:1.分析了已有的单部件可修系统、修理有延迟的可修系统和考虑预防性维修的可修系统的瞬时可用度模型,建立了有限时间约束下的系统瞬时可用度模型,该模型具有形式简单,计算方便的特性。并且利用矩阵论的相关理论和方法,证明了系统瞬时可用度的稳定性,即系统稳态可用度的存在性;2.第一部分关于有限时间约束下系统瞬时可用度的稳定性证明,使得系统瞬时可用度的特性很多都体现在它稳定之前的变化情况,即有限时间段内的系统瞬时用度波动(变化)特性。通过分析,本文提出一套刻画系统瞬时可用度在有限时间内波动特征的波动参数体系,该参数体系在一定程度上能很好地反映系统瞬时可用度波动的程度。根据工程需要,本文建立了基于波动参数的最优控制模型;3.在截尾离散Weibull分布条件下,对经典可修系统的瞬时可用度模型进行了仿真分析,在系统平均修复时间、平均故障间隔时间和平均后勤延误时间等系统常用稳态指标维持不变的情况下,研究了第二部分提出的波动参数受相关时间分布的特征参数的影响变化情况,并得到了一定的规律,便于系统设计;4.在截尾离散Weibull分布条件下,基于波动参数的最优控制模型退化为有约束的多变量优化模型。本文把装备全寿命周期分为论证、研制和使用三个阶段,分别研究了刻画系统瞬时可用度波动特征参数的约束优化模型,诸如最小可用度振幅模型、系统最优匹配模型和最优预防性维修周期模型等。最后,论文选用效率相对较高的粒子群算法作为优化工具,对模型的有效性进行了仿真说明。更多还原

【Abstract】 The interaction of subsystems often leads to fluctuations of instantaneous availability in the early use of new equipment, which represent that the new equipment cannot form fighting capacity quickly and need necessarily further adjusting. However, the evaluation theory of such fluctuations of instantaneous availability has not been established. The current researches on system availability mainly focus on the steady-state availability that denotes the behavior of the instantaneous availability when the time tends to infinity. With the rapid development of science and technology, equipment updating and out becomes faster and the equipment service period may now be only a few years or less, so the study on the instantaneous availability of new equipment systems during their service becomes more meaningful.Based on the engineering demands and the existing research results, this paper constructed a set of basic research framework and proposed the corresponding methods on the fluctuations of instantaneous availability. The major works are as follows:Firstly, based on the existing instantaneous availability models of one-unit repairable systems, the repairable systems with repair delay and the repairable systems with preventive maintenance, the instantaneous availability models under limited time constraints are built, which has simpler forms and is more convenient for computing. Moreover, the stability of the instantaneous availability models is proved, i.e., the steady-state availability exists, and the expressions of the steady-state availability are obtained.Secondly, the stability of instantaneous availability generates the interests in the fluctuation of instantaneous availability before it enters steady state. Via the analysis on instantaneous availability models, some fluctuating parameters are presented to characterize the fluctuation of the instantaneous availability and optimal control models on the fluctuating parameters of system instantaneous availability are put forward in whole life cycle.Thirdly, some simulations and analysis under truncated discrete Weibull distributions are given, in which the steady indices such as the mean time between failures, the mean time to repair and mean logistics delay time are all assumed to be fixed. The relationships between the parameters to describe the fluctuations of the instantaneous availability and the parameters of the truncated Weibull distributions are studied, and then some rules are obtained from simulation results.Finally, under truncated discrete Weibull distributions, the optimal control models reduce to constrained optimization models. The whole life cycle are divided into three phases including the demonstration phase, the development phase and the operational support phase.In each phase, the constrained optimization models such as the minimum amplitude model, the minimum adaptive time or the optimal preventive maintenance policy are studied. Moreover, the constrained optimization models are solved by Particle Swarm Optimization algorithm, which is more effective than Genetic Algorithm and Multi-agent Annealing Algorithm. Simulation results show the optimization models are valid and effective.更多还原