【作者】 黄云峰；

【导师】 王时龙；

【作者基本信息】 重庆大学 ， 机械电子工程， 2010， 博士

【摘要】 车间调度是制造业生产管理和自动化最为重要的内容之一,也是半个多世纪历久弥新的研究挑战。然而,在针对特定类型问题取得一定成果的同时,调度问题的复杂性和系统求解思想的缺失,却往往为研究者所忽视。基于雷舍尔的复杂性模型从本体论和认识论两个方面对调度复杂性进行了分析,指出车间调度应作为复杂系统进行研究。认为调度范式的核心是智能范式,从问题、方法、手段、资源等因素入手,结合智能范式对调度范式的演进过程和趋势进行了分析。认为新范式应以务实、广泛集成、简洁、分治和人机系统等五类研究视角,对车间调度范式的各研究要素进行考察。在此过程中,以和谐这一普适概念为中心,以多Agent系统为载体,从个体和谐、群体和谐以及环境和谐三个方面,提出自我驱动、通信驱动和环境驱动三种动力机制。给出了调度系统的总体结构,结合所提出的调度问题求解的三域结构和MadKit开发平台,对控制台软件和多Agent软件系统进行了详细分析和设计,初步形成了和谐调度范式的基础框架和支撑体系。认为个体和谐是调度和谐的基石,而个体和谐的实质是Agent的心理和谐,其核心必然包括Agent认知与情感的和谐。在对情感与理性的相容性分析以及情感在集成、适应、动力和交互等方面功能研究的基础上,结合情感功能与调度需求的适配性分析,指出围绕情感所展开的情感计算在车间问题的求解上潜力巨大,具备有效性和一定的必要性。提出了情感计算在调度研究中的本源性、相关性和可操作性策略。给出了基于情感的车间调度系统的总体结构,基于顺序检查理论(SCT理论)研究了情感调度中情感发生器的设计;对情感效应方式和情感行为模式进行了分析。在此基础上,研究了Agent的情智协同机制。针对以作业车间调度和柔性作业车间调度为代表的静态调度问题,采用基于情感的系统进行了多Agent求解过程适应的分析。针对动态调度问题,提出了一种情感分组合同网算法。实验结果验证了情感机制的可行性和有效性,表明情感机制作为和谐调度范式的重要组成部分,在车间调度领域具有进一步深入研究的价值。更多还原

【Abstract】 Shop scheduling is one of the most important fields of production management and automation in manufacturing, and has undergone a timeless challenge for more than half a century. However, the results obtained so far are mainly for specific problems, and the complexities of scheduling problems, together with the lack of systematic thinking, are often neglected by the researchers.Based on Rescher’s complexity model, the complexities of shop scheduling were analyzed through both ontological view and epistemological view. And it was pointed out that shop scheduling problems should be studied as complex systems. Intelligence paradigm was observed as the core of scheduling paradigm. And together with intelligence paradigms, the evolving processes and trends of scheduling paradigms were analyzed from the factors of problems, methods, means and resources. It was pointed that the new paradigm should bear research perspectives of being pragmatic, broad integration, simplicity, dividing to conquer and man-machine systems to study the factors of the scheduling paradigm. In the process, with the universal concept of harmony as the center, and Multi-Agent System(MAS) as the carrier, the mechanism combining driving forces of self, communication and environment was proposed from the aspects of individual harmony, group harmony, and environment harmony. Furthermore, the overall structure of the scheduling system was given. And with the support of the three domain architecture and MadKit development platform, the console software and multi-agent software of the scheduling system were analyzed and designed in detail. Thus, the basis of harmony paradigm for scheduling and its supporting system was initially formed.Individual harmony is the cornerstone of scheduling harmony, and in essence means the psychological harmony of the agent, with cognition and emotion harmony as its core. With the compatibility analysis of emotion and rationality, the emotion functions in integration, adaptation, motivity and interaction were studied, and it was pointed out that emotion is an effective mechanism in dealing with scheduling problems. As the emotion functions match well with the scheduling requirements, affective computing, with emotion as its core, is of effectiveness and certain necessity, and has great potential in solving scheduling problems. The functions and roles of emotions in scheduling were discussed. The strategies of endogenesis, relevance and operation in affective computing were committed. The overall structure of the emotion-based shop scheduling system was designed. Based on Sequential Check Theory (SCT), the emotion elicitor for scheduling was built up. And the mode of emotion effect and behavior were analyzed.On above bases, the integration mechanisms of emotion and intellegence of MAS were studied. As to typical static scheduling problems such as job shop scheduling problems and flexible job shop scheduling problems, emotion-based MAS was designed and applied to adapt to the solution process. For dynamic scheduling problem, the Emotion-Grouping Contract Net algorithm was committed. The results verify the feasibility and effectiveness of the studied emotion mechanisms, which as the important compenents of harmony paradigm for scheduling, are worthy of further research.更多还原