【作者】 袁红兵；

【导师】 李小宁；

【作者基本信息】 南京理工大学 ， 机械制造及自动化， 2002， 博士

【摘要】 在现今的全球制造时代，更加客户化的产品需求和更短的产品生命周期要求开发更具敏捷性的制造系统，它应当以更快的速度更新其自身，以适应新的技术、产品和组织结构的变化。因此无论是制造工业界还是学术界，人们都更加重视整个制造系统的优化、自适应性和可靠性等问题。来自制造系统内外的各种扰动是实现系统性能优化的主要障碍，而现有的递阶和分布式制造控制系统都不能有效地在维持系统高效运行的前提下处理各种扰动。 合弄制造系统HMS(Holonic Manufacturing System)将制造系统中的机床设备、制造单元、零件、产品、操作人员和工作小组等制造实体都看作为“制造合弄(Manufacturing Holon)”，它同时具有“自治”和“协作”的双重特性。HMS具有分布式的系统结构和决策职责，通过制造合弄间的协调来实现系统重构和优化，是适于敏捷制造环境的制造模式。合弄制造系统HMS的设计涉及制造合弄的定义以及怎样通过适当的系统体系结构和合弄间的协调机制以优化系统性能等基本问题。 本文首先基于“合弄制造”的基本概念，提出了支持制造系统重构的合弄制造系统参考体系结构HMS-RA(Holonic Manufacturing System-Reference Architecture)，它由订单合弄、产品合弄、资源合弄这三个基本合弄和辅助合弄组成。HMS-RA是对现有递阶制造控制和分布式制造控制的综合，它具有系统结构与控制算法相分离、非技术性因素与技术性因素相分离以及系统结构的自相似性等一系列有利于制造系统进化和重构的特点。本文在HMS-RA体系结构研究中进一步论述了HMS-RA中三个基本合弄组合的必要性、功能完整性和通用性。并采用支持面向对象技术的UML建模语言中的“聚集”和“概括”机制来将HMS-RA中的基本合弄组织起来，采用UML中的“协作图”和“顺序图”来描述基本合弄间的典型交互过程，建立了HMS的系统模型。 为了验证HMS-RA体系结构的可应用性，本文进一步基于“分布式系统中的递阶结构”、“分布式决策能力”和“并行的调度与调度执行”这三个有关合弄制造控制的基本概念，深入研究了合弄制造的控制结构和制造合弄间的交互机制，以便将递阶制造控制和分布式制造控制的优点结合起来。建立了合弄制造控制的功能模型和结构模型，并采用一种扩展的Petri网方法对HMS中合弄间的交互机制作了规范化描述，基于“FRFS”规则对多种制造控制结构中决策能力的分布作了形式化描述。 在合弄控制策略研究中，着重研究了制造调度与调度执行的集成优化问题，提出 一ii一摘要博士论文了多个能有效处理系统扰动的调度执行算法，包括两种启发式算法和一种基于扰动分析的自主调度执行算法。结合实际案例分析，对实现合弄制造控制的动态工艺规划与制造调度的集成运行机理、车间混合制造控制系统中单元代理的报价算法、车间代理的任务分派算法和基于作业路径网络模型的虚拟单元重构算法等技术的实现细节作了详细讨论。 本文最后从制造系统体系结构和设计方法的角度，全面总结了新的制造环境对制造控制提出的多层次需求，比较、分析和评价了目前几种主要的制造控制结构满足这些需求的程度。并将代理技术作为合弄制造的实现技术，基于多代理制造系统MAMS(Multi Ageni Manufacturing System)生命周期的概念，对基于代理技术的合弄制造系统体系结构及其设计方法作了初步研究。分析了在制造系统中应用代理技术的前提，研究了从制造实体中构造代理的方法和步骤，总结了多代理制造系统设计所涉及的系统体系结构、交互机制和协议、协调机制、多代理规划策略等基本问题，为合弄制造及其控制系统的规划设计和运行控制提供了设计方法参考。更多还原

【Abstract】 The trends of increased demands for more customized products and decreasing product life cycles in this global manufacturing era point at the need to develop more agile manufacturing systems, that should adapt themselves at an ever-increasing pace to incorporate new technology, new products and new organizational structures. Both manufacturing industry and academia pay much more attention to the optimization, adaptation and reliability of the manufacturing system as a whole. The disturbances from within and outside the manufacturing system constitute a major obstacle to achieve the desired performance optimization, however, the current hierarchical and heterarchical manufacturing control systems are both not sufficient to cope with disturbances while maintaining high system performance.Holonic Manufacturing System (HMS) is a manufacturing system where key elements, such as machines, cells, parts, products, operators and working teams, etc., are modeled as "Manufacturing Holons", which have both "Autonomous" and "Cooperative" properties. The decentralized system structure and distributed decision-making authority, the cooperative relationship among holons to achieve system reconfiguration and optimization, make HMS a new paradigm with great potential to meet today’s agile manufacturing challenges. Critical issues to be investigated include how to define manufacturing holons, the appropriate system architecture and the effective cooperation mechanisms for good system performance.Based on the basic concepts of holonic manufacturing, this dissertation develops firstly a holonic manufacturing system reference architecture called HMS-RA, which consists of four types of holons: The Order Holon, Product Holon and Resource Holon which are referred as there Basic Holons and the Assistant Holon. HMS-RA is a generalization of the Hierarchical and Heterarchical control approaches. The independence of System Structure from Control Algorithms, the independence of logistical and technological concerns, and the self-similarity of the architecture, are the three main innovations essential for the evolution and reconfiguration of manufacturing system. The necessity, functional integrality and generality of the there Basic Holons are discussed in detail. The system models for HMS are established by using the "Aggregation" and "Generalization" mechanism in Object-Oriented UML modeling language as the organizing mechanism of the three Basic Holons in HMS, and using the "Collaboration Diagram" and "Sequence Diagram" to describe the typical interaction scenarios within the Basic Holons.To verify the feasibility of the architecture and investigate its applicability, according to the three basic concepts for holonic manufacturing control: "Hierarchy in Distributed Systems", "Distribution of Decision Power" and "Concurrent Scheduling and Schedule Execution", this dissertation presents a holonic manufacturing control architecture and an interaction mechanism that combine the best of hierarchical and heterarchical control. The holonic manufacturing control architecture is expressed with a functional model and a structural model, while the interaction mechanism is described formally with an extended Petri Net approach. To formally describe the distribution of decision power within several manufacturing control approaches, a modeling framework based on the rule of "Fixed Rules and Flexible Strategies" is proposed afterwards.This dissertation puts emphasis on the integration optimization issue of manufacturing scheduling and scheduling execution. It presents two heuristic algorithms for Schedule Execution and one based on Perturbation Analysis to deal with the system disturbances effectively. As the key techniques to implement holonic manufacturing control, the Dynamic Process Planning technique and its integration operation mechanism with manufacturing scheduling, the Bidding Algorithm of the cell agent, the task Assigning Algorithm of the shop floor agent within a hybrid shop floor control system, and the Reconfiguration Algor更多还原