【作者】 赵博；

【导师】 刘晓冰；

【作者基本信息】 大连理工大学 ， 机械制造及其自动化， 2000， 博士

【摘要】 调度理论与实践的脱节一直是制约企业生产高效率、敏捷化的关键问题之一。正是在这个背景下，本文提出了结构化集成调度系统理论及基于该理论的虚拟车间智能支撑平台的体系结构，目的在于将调度理论同生产实际结合起来，并实现车间生产管理软件的柔性化。 具体地，本文包括如下三大部分： 一．研究背景及相关理论基础 1．以每个十年为一个阶段总结了经典调度理论的发展历程，介绍了调度问题研究中所涉及到的相关技术和领域，阐述了论文研究的背景，并结合我国企业车间生产管理的实际情况，对调度研究的发展方向做出了科学的预测。 2．以经典调度理论为核心，介绍了调度问题的基本内容，包括基本概念、描述方法、调度目标和算法研究等，最后讨论了经典调度理论的局限性。 二．结构化集成调度系统理论的研究 1．强调了调度问题的大系统特点，提出以资源和任务为核心的系统分析思想，初步地建立了结构化集成调度系统理论的基本框架，确定了该理论研究的内容、目的和一般方法。 2．研究了调度系统的建模问题，将该问题划分为五个子问题：系统建模、资源建模、任务建模、活动建模和环境建模。应用了自定义的一种高级Petri网建立了系统模型；应用面向对象技术分别建立了资源、任务和活动的模型；应用人工智能中基于框架的知识表示方法建立了调度系统的环境模型。 3．在比较分析几种著名的生产控制技术的基础上，提出并研究了调度系统控制的几个关键问题：瓶颈分析、误差分析、扰动分析和控制策略。在瓶颈问题的研究中，拓展了OPT技术中关于瓶颈问题的研究，划分了四种瓶颈因素，分析和研究了它们之间的关系及对系统控制的影响；在误差分析研究中，分析了造成调度系统误差的各种随机噪声及其产生原因；在扰动分析研究中，分析了调度系统的几种随机因素及它们对系统运行过程的影响。在控制策略研究中，研究了全局控制和局部控制的基本思想和方法。 4．在分析典型调度算法的基础上，提出了将算法研究从微观走向宏观的思想，提出了算法集成的概念。在算法对象化基础上，运用人工智能方法，初步实现了算．法的知识表示及算法智能组合与动态选择等机制，并且将多代理技术引入到算法集成研究上，提出了基于多代理机制的算法动态集成方案。 三．虚拟车间构造智能支撑平台的体系结构研究 以结构化集成调度系统理论为指导，提出了虚拟车间的概念，建立了虚拟车间智能支撑平台的体系结构。首先，在简要总结传统车间管理软件研究与开发的不足基础上，详细阐述了平台的基本思想；其次，确定了平台的基本功能和结构，划分了平台结构的六个组成部分：平台管理系统、虚拟车间基本构件模型、应用系统模板、多代理分析系统。数据库和用户定制系统：再次，提出了平台研究与厂发的几个关键技术，并初步提出了解决方案；最后，介绍了该思想在一个项目中的部分应用。更多还原

【Abstract】 The gap between scheduling theory and scheduling pracitice is one of crucial problem that limited enterprises to enhance their production efficiency and agility. Under this background, structured and integrated scheduling system theory and its- based structure of intelligent supported platform for virtul workshop is presented in this dissertation. The goal is to conjoin scheduling theory and practice. Then flexibility of workshop management software could be realized. Concretly, this thesis consists of three parts: Part I. Research background and basic knowledge about pertinent theory I .Development stage of research in classical scheduling theory by every ten years is summarized. Pertinent technology and fields concern with study of scheduling problem is presented. Background of the research of this thesis is introduced. And future research directions in scheduling research are reasonably prognosticated according to actuality of productive management of workshop in enterprises of China. 2.Basic content of scheduling problem such as basic concepts, discribing methods and study of algorithm are expounded encompassing classical scheduling theory. Incompleteness of classical scheduling theory is discussed finally. Part II.Research of structured and integrated scheduling system theory 1. The view that scheduling problem to be a large scale systems problem is emphasised. The idea of system analysis on the core of resources and jobs is presented. Framework of structured and integrated scheduling system theory is build elementarih. The content, goal, and general method of this theory are established. 2. Modeling problem of scheduling system is studied. And scheduling system modeling is divided up five sub-problem: system modeling, resource modeling, job modeling, activity modeling and environment modeling. System model is built with a high-level Petri Nets which is defined in this thesis. Resource model, job model. and activity model are built with object-oriented technique. And environment model is built with frame-based knowledge discribing method of Al. 3. On the basis of comparing and analyzing some famous production control techniques, some crucial problems that are bottlenecks analyzing, errors analyzing, perturb analyzing, and control strategy are put forward and researched. In the research of bottleneck, bottleneck problem of OPT is extended. Four classes bottlenecks are defined and their relations and influence to system are analysed and studied. In the analysis of errors, stochastic yawp that brings up error of scheduling system and its origin are analysed. In the analysis of system disturbing, some stochastic factor and their influence to system running are analysed. In the study of control strategy, the basic idea and method of public control and local control is discussed. 4. On the base of analysing typical scheduling algorithms, the idea that algorithm research should be developed from microcosmic view to macroscopcal view is discussed and concept of algorithm intergrating are presented. On the base of building algorithm object, Al is used to realize knowlegeble describing of algoritms, intelligent combution of algorithm, and mechanism of dynamic selection of algorithms. Then muti-agent technique is used into study of algorithm intergrating. And muti-agerit based algorithm dynamic intergrated blue print is put forward. Part III. Research of structure of intelligent supporting platfor更多还原