【作者】 尹晓虎；

【导师】 温熙森；

【作者基本信息】 国防科学技术大学 ， 机械工程， 2008， 博士

【摘要】 装备维修系统是由装备维修所需的物质资源、人力资源、信息资源以及管理手段等要素组成的复杂系统体系。伴随作战理念和作战样式的演变以及装备技术的发展,装备维修系统的构成结构和运行模式都发生了深刻变化,科学维修、精确保障成为指导装备维修工作和评价装备维修系统效能的根本要求。装备维修系统具有的随机性、动态性、相关性和演化性,给装备维修系统的分析、设计、集成和控制带来巨大挑战。过程信息是过程控制的基本手段。理解和掌握装备维修系统的动力学行为,也即装备维修系统的运行规律,是对其进行分析、设计、集成和控制的基础,动力学分析方法为此提供了合适的研究框架和工具。科学维修、精确保障的实现,需要解决如下问题:(1)准确掌握装备退化失效的基本规律,实施必要的维修;(2)优化装备维修策略,在恰当时机执行最佳次数的维修;(3)合理调配维修资源,及时响应维修需求、按需保障;(4)全局规划要素配置,有序集成维修系统形成体系合力。围绕上述问题,本文从动力学角度研究装备维修系统,具体工作如下:(1)从动力学辨识的角度研究了基于状态监控参数反演装备退化失效规律的问题。采用随机Ito方程建立了刻画状态监控参数变化的动力学方程,基于状态监控参数变化率的最大熵概率密度估计确定了装备退化过程的动力学性质并提出迭代递归预测误差算法进行动力学辨识,利用高斯核密度K均值聚类和期望最大化算法实现了装备退化状态估计与退化失效规律的有限位相型分布逼近,从而建立了装备退化失效过程的动力学分析框架,并成功应用于某型发动机的磨损退化过程分析。(2)从装备使用与维修相耦合的角度研究了装备维修时机与维修次数的控制问题。通过在装备退化状态空间上扩展装备维修状态,分析了不同维修活动效果对装备使用与维修过程状态转移的影响,针对单一失效模式和复杂失效模式分别提出了密度演化方法和矩阵解析方法,建立了维修时机和维修次数的优化准则,并成功应用于某型发动机磨损的预防性维修周期和非完美维修次数的优化。(3)采用系统动力学方法研究了由于装备使用过程、维修策略作用、维修资源约束相互影响造成的装备维修系统的动力学行为模式。基于装备使用与维修过程的状态转移关系,分析了维修系统中的基本因果关系和反馈结构,以某型发动机维修系统为例,建立了维修系统的系统动力学模型,探讨了维修系统的行为模式及其相关影响因素。(4)借鉴网络分析方法研究了装备维修系统的集成模式与集成效能评估问题。通过分析装备维修系统的基本构成要素,建立了装备维修系统的元模型及其效能评价指标,从维修信息利用率的角度考察了维修要素增长及其作用关系增长对维修系统体系效能的影响,确认了装备维修系统集成的最佳方式与可能的集成模式。通过装备维修系统的动力学分析技术研究,可望准确掌握装备退化失效的状态演化规律从而提高维修实施的针对性和必要性,能够理解维修活动对装备退化的作用从而更好地控制装备维修时机和次数,能够理解装备维修需求的产生规律从而更好地响应维修资源需求并实现按需保障,能够理解维修要素分布对装备维修系统效能的影响从而更好地配置和部署装备维修要素。因此,装备维修系统的动力学分析技术为优化和控制装备维修过程与维修系统奠定了基础,有关结论可应用于指导装备维修系统的分析、设计、构建和集成。更多还原

【Abstract】 The equipment maintenance system is the complex system of systems, whichcomprises many maintenance elements, such as materials, manpowers, information andladders of management, etc, to carry out maintenance actions. With the evolution ofoperational theories and operational styles and the development of equipmenttechnologies, the structures and functional modes of equipment maintenance systemshave been transforming profoundly, which require that maintenance actions should becarried out scientifically and supported accurately. The equipment maintenance systemin nature, however, is stochastic, dynamic, correlative and evolutive, which brings greatchallengestoanalyze,design,integrateandcontrolit.Process information is the basic means to control it. Comprehending and graspingthe dynamical behaviors of the equipment maintenance systems, i.e. the operationallaws, is the foundation of analyzing, designing, integrating and controlling it. Thus, thedynamical analysis methods naturally become the suitable framework and tools forstudyingdynamicalbehaviorsoftheequipmentmaintenancesystem.To carry out maintenance scientifically and support accurately, the followingissues must be solved: (1) to execute necessary maintenances by grasping theequipment’s degradation and failure mechanism well and truly, (2) to carry outmaintenance at proper occasions and with the optimal number of times by optimizingthe equipment maintenance policy, (3) to respond to the maintenance requirementstimely and support them correspondingly by allocating maintenance resourcesreasonably, (4) to integrate maintenance systems orderly to form systematic effects byplanningthemaintenanceconstituents’deploymentglobally.Centering on these issues, this thesis studies the equipment maintenance systemfromtheviewpointofdynamics,andthemajorworkscanbesummarizedasfollows:(1)The problem of acquiring the laws of degradation and failure of equipmentswiththeirstatemonitoringdataisstudiedfrom theviewpointofdynamicsidentification.A group of dynamical equations based on stochastic Ito equation are put forward tocharacterize the change process of state monitoring data. The maximum entropyestimationalgorithmisderivedtoestimatetheprobabilitydensityfunctionoftherateofincrements of state monitoring data, which is used to determine the dynamical propertyof the equipment deterioration process. The iterative recursive prediction error methodisprovidedfortheparametersidentificationoftheequipmentdeteriorationprocess.Andthe deterioration states of the equipment are estimated with Gaussian kernel basedK-means clustering algorithm and the law of equipment deterioration is approximatedwith finite phase type distribution by using the expectation maximization algorithm.With all these equations and methods above, a dynamical framework for analyzing the equipment degradation process is built, which has been applied successfully to analyzetheweardeteriorationprocessofsomeengine.(2)The problem of controlling maintenance occasions and maintenance numbersis studied from the viewpoint of the coupling in the usage and maintenance of theequipment. Extending the maintenance states over the state space of the equipmentdeterioration process, the effects of various maintenance actions on the usage of theequipment and the state transition of its maintenance process are analyzed. Aiming atthe single and complex failure modes, the densityevolution method and matrix analyticmethod are put forward respectively to build the optimal criteria of maintenanceoccasions and maintenance numbers, both of which have been successfully used tooptimize the preventive maintenance cycle and maintenance numbers for the wear ofsomeengine.(3)The system dynamics method is used to study the dynamical behavior modesof the equipment maintenance system, which result from the interaction among theequipment working process, maintenance polices and the constraints of maintenanceresources.Basedonthestatetransitionoftheequipmentusageandmaintenanceprocess,the basic causality and feedback structure in the equipment maintenance system areanalyzed. The system dynamics model of the maintenance system is established forsome engine, and the behavior modes and the influencing factors of the equipmentmaintenancesystemareexplored.(4) To evaluate the equipment maintenance systems integration performance, thenetwork analysis method is adopted to analyze the integration modes and measure thesystematic performance of maintenance systems. By way of analyzing the basicelements of the equipment maintenance systems, its elementary model and itsperformancemeasures areset up.Theeffect ofthegrowthofmaintenanceelements andtheir interactions on the maintenance system performance is studied from the viewpointof maintenance information utilization rate, and the best integration mode and possibleintegration patterns of equipment maintenancesystems are determined.Based on the study of dynamical analysis methods of the equipment maintenancesystem, the state evolution rules of the equipment deterioration and failure process canbe mastered and the maintenance can be performed pertinently and necessarily; theeffects of maintenance actions on the degradation of the equipment captured and themaintenance moment and maintenance number of times controlled by rule and line; thegeneration laws of maintenance requirements understood and the maintenance resourcesrequirements answered in time and executed accurately; the impact of the deploymentof maintenance elements on the performance of maintenance system caught on and themaintenance elements configured and deployed best. Thereby, the study of dynamicalanalysis methods lays the foundation of optimizing and controlling the equipmentmaintenance process and maintenance system, the results of which could be used to guide the analysis, design, integration and control of the equipment maintenancesystem.更多还原