【作者】 柳剑；

【导师】 张根保；

【作者基本信息】 重庆大学 ， 机械工程， 2014， 博士

【摘要】 随着现代化制造系统日益朝着大型化、自动化、柔性化、高效化和精密化方向发展，其组成设备单元的种类与数量迅速增加，结构也更为复杂，再加上复杂多变的使用工况与运行环境，使其发生故障的几率也会明显增加，因此如何保证制造系统的运行可靠性已成为用户制造企业面临的重大问题。现代化制造系统在运行过程中多样化的需求、动态不确定的制造环境以及异常复杂的内在退化机理使得传统的制造系统可靠性研究方法过于局限和单薄。针对现有研究工作中存在的不足之处，本文以制造系统服役阶段由各种广义外在环境随机事件与各组成设备内在退化机理引起的性能状态退化为切入点，将脆性理论和多状态系统理论引入制造系统运行可靠性分析与维修保障策略的研究中，研究了制造系统脆性效应的累积与传播机理，建立起基于“R-E”体系的制造系统运行可靠性分析技术框架与性能状态演变的预测模型，提出了考虑脆性效应的改进区间通用生成函数可靠性评估方法，在此基础上分析并制订了动态预测维修策略以保障制造系统的运行可靠性，并从维修相关性的角度以服役阶段效益产出最大为目标对该维修策略进行优化。本文提出的理论方法体系从微观和宏观两个方面对制造系统的运行可靠性进行评估，从定性和定量两个方面对制造系统维修策略进行分析，在一定程度上克服了传统制造系统可靠性分析方法与维修策略在用户实际应用中的困难与缺陷，并为制造系统运行可靠性分析与维修策略制订提供了一条新的思路。本文具体内容包括以下几个方面：（1）介绍了制造系统及其运行可靠性的相关理论知识，论述了制造系统可靠性与维修策略的发展历程及国内外的研究现状，说明了本论文的课题来源、主要研究内容及总体结构。针对制造系统层次结构复杂、故障源多、制造任务多变、状态多变等特点，论文将多状态可靠性理论、复杂系统脆性理论以及“人—机—环境”系统工程学引入制造系统服役阶段的运行过程中，探讨了制造系统多态性的形成机理，描述了制造系统故障演化与传递机理，分析了制造系统运行过程中的广义外在环境，并提出了基于“R-E”体系的制造系统运行可靠性分析技术框架，为制造系统运行可靠性的评估与维修保障策略的制订提供了理论支撑。（2）研究了广义外在环境下制造系统脆性效应的累积与传播模型。首先引入因素空间理论来描述广义外在环境中动态因素的物理脆性激发作用，构建了制造系统脆性效应与各动态外在因素之间的映射模型。然后引入脆性度指标来量化脆性激发因素直接与间接对制造系统性能状态退化程度的综合影响，并系统分析了脆性激发因素风险度与耦合度的计算流程。最后根据制造系统物理脆性的激发与过程脆性的激发两类脆性激发机制建立起制造系统设备单元层与制造系统层的脆性效应传播模型，并采用改进蚁群算法去搜寻给定制造任务下的制造系统最大脆性激发路径，为制造系统具体故障演化与传递机理的分析提供了一种新的思路。（3）研究了基于脆性效应的制造系统运行可靠性评估方法。首先引入状态空间模型，描述了脆性激发因素作用下制造系统的状态演变过程，建立起制造系统性能状态演变的预测模型。然后提出一种考虑脆性效应的改进通用生成函数法对制造系统运行可靠性进行分析，在此基础上借鉴区间分析理论以及仿射算术的思想对改进通用生成函数法进行区间扩展，对制造系统区间运行可靠性进行评估，不仅体现了不同制造任务下制造系统异质性与动态不确定性的存在，同时也使得运行可靠性评估结果更加准确可靠。（4）研究了广义外在环境下制造系统维修保障策略的选择算法。首先对制造系统维修保障及维修策略的基本知识进行了概述。其次针对广义外在环境下制造系统及其设备单元存在的异质性以及动态不确定性，给出了制造系统的主要维修决策属性，并归纳了制造系统最佳维修策略的选择流程。然后应用模糊层次分析主观赋权类方法和信息熵客观赋权类方法分别对制造系统维修策略进行排序选择，并采用模糊Borda法综合以上两种方法进行综合提取，有效地克服传统单一方法计算过程存在的主观性和模糊性问题，提高了维修策略排序选择的准确性和合理性。最后运用该方法得到制造系统关键设备单元——加工中心的最佳维修保障策略应为预测维修。（5）研究了因素驱动的制造系统动态预测维修策略。首先针对传统制造系统维修保障策略研究中未全面考虑动态环境事件影响制造系统性能状态退化的问题，并基于制造系统脆性效应的理论分析以及性能状态演变预测模型的建立，分析了广义外在环境下制造系统的预测维修策略，给出了维修策略的成本效益评价模型。然后在给定动态维修决策阈值、有限缓冲区容量以及维修资源等维修相关性约束条件下，以服役阶段效益产出最大为目标建立了基于遗传算法的制造系统维修策略寻优流程。最后以国内某制造企业的一套制造系统为例，通过对比分析证明了本文提出的动态预测维修策略的可用性与有效性，对于提升我国制造装备运行可靠性、经济性以及加工制造能力，具有潜在的借鉴价值。更多还原

【Abstract】 As modern manufacturing systems developed towards larger size, higherautomation, more flexible, more efficient and higher precision continuously, the partsand equipments in systems greatly increase, and the structures are more complicated.As a result the probability of system failures will be increased significantly undercomplex working conditions and circumstances, and therefore how to ensure theoperational reliability of manufacturing systems has become a major problem facingthe users of manufacturing systems. Conventional reliability analysis methods formanufacturing systems are limited and weak, because of diverse manufacturingmissions, dynamic manufacturing environment and extremely complex degradationmechanism during the operation of modern manufacturing systems. For thedeficiencies that exist in the current researches, in this paper, various random events inthe universal external environment and performance degradation of manufacturingsystems caused by intrinsic degradation mechanism are treated as a starting point, thenbrittleness theory of complex system and theory of multi-state system are introduced instudy on operational reliability analysis and maintenance policy for manufacturingsystem, the mechanism of accumulation and propagation of brittleness effect inmanufacturing systems is studied, the technical framework of operational reliabilityanalysis and the predictive model of performance degradation for manufacturingsystems based on R-E system are established, and the improvement interval universalgenerating function for reliability evaluation is proposed considering brittleness effect.According to the operational reliability analysis above, the dynamic predictivemaintenance policy is formulated to ensure the operational reliability of manufacturingsystems, and the maintenance policy is optimized with the largest benefit output in theoperational phase as target in the end. The theoretical method system presented in thispaper evaluates operational reliability of manufacturing systems from micro and macroaspects, analyses maintenance policies of manufacturing systems from qualitative andquantitative aspects, overcomes the difficulties and deficiencies of conventionalreliability analysis methods and maintenance policies in the applications to someextent, and provides a new approach for analysis of operational reliability andformulation of maintenance strategy for manufacturing systems. Topics covered in thispaper are as follows： (1) Some basic theories about manufacturing system and its operational reliabilityare introduced, the development and the status of researches on reliability andmaintenance policy of manufacturing system are discussed, and the source, the mainworks and frame of this paper are illustrated here. For the characteristics of morecomplicated structure, more failure sources, varied manufacturing missions, and variedperformance, multi-state reliability theory, brittleness theory of complex system andsystem engineering of man-machine-environment are introduced to discuss theformation mechanism of multi-state during the operation phase, describe themechanism of failure evolution and propagation, analyze the universal externalenvironment of manufacturing systems, and propose technical framework ofoperational reliability analysis for manufacturing system based on R-E system. Theseworks provide a theoretical support for evaluation of operational reliability andformulation of maintenance strategy for manufacturing system.(2) The accumulation and propagation model of brittleness effect inmanufacturing systems under universal external environment is studied. Firstly, thetheory of factor space is introduced to describe the physical stimulation of dynamicfactors in universal external environment, and the mapping model between brittlenesseffect in manufacturing systems and the dynamic external factors is constructed. Then,brittleness degree is given to express the direct and indirect impact of each brittlenessmotivators on system degradation, and the algorithms for risk degree and couplingdegree of brittleness motivators are analyzed systematically. Finally, according to thebrittleness mechanism of physical stimulation and flow stimulation, the brittlenesspropagation models of equipment layer and manufacturing system layer are established,and the improved ant colony algorithm is used to search propagation path of brittlenesseffect with maximum intensity. These works provide a new method for analysis ofevolution and propagation mechanism of concrete failures in manufacturing systems.(3) Operational reliability evaluation of manufacturing system based onbrittleness effect is studied. Firstly, state space model is introduced to describe the stateevolution of manufacturing system under the impact of brittleness motivators,following the establishment of state evolution predictive model of manufacturingsystem. Moreover, the improved universal generating function for reliability evaluationof manufacturing is proposed considering brittleness effect, and the interval extensionof reliability value is made on the basis of the interval analysis theory and affinearithmetic to evaluate the interval-valued operational reliability. The result shows the heterogeneity and dynamic uncertainty of manufacturing systems under differentmanufacturing missions, and makes reliability evaluation more accurate and reliable.(4) The selection algorithm for maintenance policies of manufacturing systemunder universal external environment is studied. Firstly, the basic theory ofmaintenance and its policies of manufacturing system are summarized. Secondly, forthe heterogeneity and dynamic uncertainty existed in manufacturing systems underuniversal external environment, decision-making attributes for maintenance are given,and the selection process of optimum maintenance policy is summarized. Then, themaintenance policies of manufacturing system are ranked by fuzzy AHP (subjectiveweighting) and information entropy method (objective weighting), and the optimummaintenance policy is selected by the fuzzy Borda method combining with fuzzy AHPand information entropy. This method effectively overcomes the problems aboutsubjectivity and ambiguity existed in conventional selection algorithms, and improvesthe accuracy and rationality of ranking selection of maintenance policies. Finally, usingthis method, the optimum maintenance policy for machining center that is keyequipment in manufacturing system should be predictive maintenance.(5) The factor-driven dynamic predictive maintenance policy for manufacturingsystem is studied. Firstly, in view of incomplete consideration to the influence ofdynamic environment events on manufacturing system degradation in conventionalmaintenance policies, the predictive maintenance policy under universal externalenvironment is analyzed and the cost-effective evaluation model of maintenance policyis established based on brittleness theory and state evolution predictive model ofmanufacturing system. In addition, as the dynamic maintenance decision thresholdvalue was given and the buffer contents and the maintenance sources were limited, themaintenance policy optimization flow of the manufacturing system based on GeneticAlgorithm is built with the largest benefit output in the operational phase as target.Finally, an example of manufacturing system in one domestic manufacture enterprise istaken to prove the usability and validity of the method mentioned after thebenchmarking analysis. This study of maintenance policy has potential reference valueto improve the operational reliability, economical efficiency and capability ofmanufacturing equipments in China.更多还原