【作者】 王英；

【导师】 房至一；

【作者基本信息】 吉林大学 ， 计算机系统结构， 2007， 硕士

【摘要】 传统的Workflow技术在跨企业的供需链管理系统中,普遍存在耦合程度高、灵活性低和缺乏智能性等缺点,已不能满足现代应用的需要。本文在深入分析分布式Agent和Workflow技术的基础上,提出了集成分布式Agent和Workflow技术的协同控制框架,来构建支持动态联盟的供需链管理系统。本文采用基于ACPN的建模方法建立了以订单为驱动的协同控制模型;在分析协同控制机制的过程中,利用MAS间的协商模型,实现了协商发起Agent和协商参与Agent的协同过程;通过基于RAWF的分布式事务处理模型,阐述了任务之间的依赖关系,设计了协同控制引擎,在ECA规则的控制下,协同控制引擎根据Agent间的协商结果进行调度;最后应用JADE平台和J2EE平台的开发框架,实现了供需链管理系统。用分布式Agent和Workflow技术实现供需链管理的协同控制具有更好的柔性和敏捷性,能够有效地完成企业过程再造,动态控制供需链企业活动,从而使联盟企业在合作与竞争中更好地发展。更多还原

【Abstract】 The research on collaborative control in supply chain management is that the core business integrates the distributed resources of the supply chain to implement dynamic management between synchronized coordination and integrated seamless of various nodes of the supply chain by information technology and management methods, in order to make the supply chain in an optimal operational status. Workflow technology plays a decisive role in promoting technological progress of the enterprise and the development of the information industry. However, traditional workflow technology has not been able to satisfy the heavy demand of enterprise application modernized integration for its shortcomings such as tight coupling, worse flexibility and passivity. Therefore, it is necessary to import new technology to develop and improve business process of the supply chain management system. Agent technology is an important branch in artificial intelligence domain. It is paid much attention by its characteristic in many research fields and the application prospect is extremely broad. Therefore, based on analyzing the distributed Agent and the workflow technical profoundly, this paper puts forward a management frame based on Agent and workflow technology, constructing the supply chain management system which supports dynamic alliance agile supply chain. The frame based on supply chain management system adds a layer of Agent, which is defined formally to support dynamic interaction for the existing supply chain management system by the ability of Agent autonomy, cooperation, negotiation and reasoning. The specific business process is also completed by the bottom workflow system.Each node in supply chain has its own resources, capabilities and goals, and makes the materiel flow, information flow and financing flow smooth for the products and services by mutual cooperation under the supply chain management. This paper makes the independent or semi-independent entities as an abstract model for Agent. Supply chain management system is decomposed into five business Agents and three service Agents. Business Agents contain PurchaseAgent, SaleAgent, WarehouseAgent, FinanceAgent, and TransportationAgent. Service Agents contain CustomerAgent, SupplierAgent and transporterAgent. Each Agent has an internal workflow management system to implement and monitor the business process.The business process is driven to collaborate by the orders, which are the core status in supply chain management. Based on the basic procedure of supply chain for collaborative control, this paper puts forward a method of Agent-based Colored Petri Nets (ACPN) to establish the collaborative control model which is driven by the orders. The model is run by the state changes of activities, which analyses the implementation process of the orders and the Agent dynamic behaviors effectively by the reciprocity and interaction of material flow and information flow.Collaborative control mechanism in supply chain management can access information timely and feedback on supply and demand at any time to help enterprise make right decisions timely. Implementation of collaborative control in supply chain management by using distributed Agent and workflow technology in this paper is analyzed mainly from the following four aspects:①Construct a negotiation model, analyze and implement the collaboration process between the sponsors and participants. Negotiation is an important strategy of collaboration, which means each Agent involved in collaboration considers overall interests of the MAS, looking for a collaborative method widely accepted by discussing, understanding and complementing. The scheduling result is obtained through the negotiation among the different types of Agents. ②A distributed workflow model based on Role-driven Multi-Agent Workflow(RAWF) is presented. The distributed transaction processing model based on RAWF refers to implementing the environment of the distributed business process management by a group of different Agents. In this model, Agent is considered as a person performing some tasks, and each Agent has a role-base which provides an agent one or more roles. Multi-Agent collaboration in describing the workflow process has been linked up with each other in the implementation phase. Such different Agents can collaborate together for the workflow tasks. Through Agent collaboration, when a problem occurs, this system is able to predict its possible consequences and tries to reorganize workflow processes to reduce the possible harms.③Analyze the coherence relations of tasks. The collaboration between tasks is based on the method of the interdependent information flow to describe. It may prevent multi-Agent disorder and chaos, and break the constraints of the resources, the environment and the time.④Design the collaborative control engine. The function of monitoring collaborative control engine may monitor the movement of the workflow in the whole organization and provide a range of management functions. Further more, through Agent collaboration, a workflow process can receive relevant information from other processes or outside. By ECA rules, Agent can make necessary workflow process changes. After the remote Agent receives orders, it analyzes the data about the process and schedules its own business process. According to the order from the data, it will start or end in one or more activities.When it comes to implementation, this paper presents a development framework based on Java Agent Development Environment (JADE) platform and J2EE platform, in order to implement the supply chain management system based on distributed Agent and workflow technology. JADE can facilitate the rapid development of collaborative engineering applications through the provision of reusable packages of agent level components and programming tools. In order to enable the user to access Agent by Internet and visit the resources in application server by Agent, this paper designs and implements the JADE control interface to communicate with the application server. Under the JADE development environment, this paper adds JSP/Servlet access Agent and Agent access business logic to implement process scheduling. Moreover, this paper analyzes the key codes in the collaborative process of the Agent. At last, this paper takes the PurchaseAgent as an example to illuminate the collaborative control process.By the above analysis, the implementation of collaborative control in supply chain management by using distributed Agent and workflow technology has the following advantages:①Flexible allocating mechanism. Agent can determine which one is responsible for the implementation of a task by negotiation. If it finds some Agent can not finish the task, it will resolve the conflict by collaborative mechanism.②Distributed system framework. Multi-Agent is suitable for the distributed environment, and Agent in the system encapsulates task execution module of the workflow management system and executes the local business process. It meets the needs of the cross-organizational workflow in supply chain management system and is available to integrate business progress in distributed environment.③Flexible communicating mechanism. Messages based on FIPA ACL reduce the burden of communication.As time is limited, the research on collaborative control in supply chain management is not enough, and the supply chain management system implemented by this paper still needs further improvement.更多还原