【作者】 资武成；

【导师】 罗新星；

【作者基本信息】 中南大学 ， 管理科学与工程， 2010， 博士

【摘要】 在供应链管理中,各合作伙伴是具有自身利益的独立企业,它们在合作的过程中会存在大量的冲突。协商是解决冲突的基本手段,也是实现各合作伙伴协作的重要环节,对提高供应链整体效益和竞争力具有重要的意义。论文把人工智能中的Agent技术应用到供应链伙伴协商当中,针对供应链伙伴协商Agent的特点,提出通过构建供应链协商关系网来组织Agent进行协商。为了提高协商效率和协商成功率,论文把基于案例推理技术和劝说式协商方法应用到供应链伙伴的协商当中,对其中的若干关键技术进行了研究,并以此为基础,提出了一种基于Multi-Agent的供应链协商模型。最后,基于JADE平台开发实现了供应链协商仿真原型系统,以验证供应链协商模型的有效性。论文研究内容主要表现在以下几个方面：(1)供应链协商模型中的相关理论分析。对供应链和供应链管理中的内涵、特征、供应链管理的原则与目标进行了详细的阐述；界定了供应链伙伴协商的内涵,对协商类型、协商策略、协商协议等相关理论进行了研究；研究了Agent技术、BDI模型理论和Multi-Agent协商机制等相关理论。这些理论为论文的研究提供了理论基础和技术支持。(2)基于供应链协商关系网的Multi-Agent协商组织形式研究。供应链实际上是介于自由开放的市场和单个企业组织之间的一种中间性组织,其协商环境和协商Agent具有自身的特点,如何构建合理的协商Agent组织形式对提高供应链伙伴的协商效率和协商成功率具有重要的意义。论文提出了一种基于信任的Multi-Agent供应链协商关系网,利用供应链协商关系网组织Agent进行协商,深入研究了供应链协商关系网的形成和演化机制,给出了供应链协商关系网的形成和更新算法,对供应链协商关系网中节点Agent的关系表示和Agent组织框架进行详细的论述,对供应链协商关系网的性能进行了深入的研究。(3)供应链伙伴协商中基于案例推理的关键技术研究。针对供应链伙伴协商具有相对稳定、重复性的特点,提出把基于案例推理机制应用到供应链伙伴的协商当中,通过案例推理初始化协商Agent的背景信息、预测协商伙伴的议题保留值。论文对供应链环境下基于案例推理机制的关键技术进行了深入研究,主要研究了运用XML和本体技术表示供应链协商案例的语法和语义技术、应用ID3算法和灰色关联理论检索最相似案例的方法,对供应链伙伴协商案例库的维护和重用技术也进行了理论探讨；最后提出了一种融合Agent和基于案例推理的供应链协商案例逻辑推理模型。(4)基于劝说的供应链协商提议交互机制研究。本文把劝说的方法应用到供应链伙伴Agent的协商当中,研究了劝说式协商中的提议产生、表示和效用评估机制,深入分析了协商Agent在劝说中的让步方式、交互原则和提议接受机制,并用算例说明了协商Agent劝说的过程。为了提高供应链伙伴的协商成功率,解决协商中可能出现的僵局,论文针对供应链伙伴协商的特点提出了一种协商僵局判断方法和僵局解决机制。(5)提出了一种基于Multi-Agent的供应链协商模型。在前面几章理论研究的基础上,论文提出了一种基于Multi-Agent的供应链协商模型。首先对基于Agent的供应链协商模型框架、协商Agent的状态进行了形式化的定义,这是规范模型的基础；第二,协商策略是协商模型的重要组成部分,论文研究了效率型供应链和反应型供应链中协商伙伴的基本协商策略,并把Bayesian学习机制应用到协商策略的学习当中,阐述了应用Bayesian学习机制在线更新协商策略的方法和步骤。第三,提出了一个供应链伙伴协商的协商协议,该协商协议具有通用性、规范性、较好的协商效率,能提高协商成功率等特点。第四,对协商模型中的协商过程进行了阐述,并把本文的供应链协商模型和相关的协商模型进行了比较。(6)在JADE平台的基础上,设计、开发实现了供应链协商仿真原型系统。论文介绍了JADE仿真平台的特点、Agent运行机制,对仿真原型系统中协商Agent的静态结构、行为交互以及通信机制都进行了详细的设计。仿真实验结果表明,基于Multi-Agent的供应链协商模型能提高供应链伙伴的协商效率和协商成功率。更多还原

【Abstract】 Every partner is a single enterprise and a lot of conflicts maybe come out during the process of cooperation in supply chain. Negotiation is most important to improve the total profits and Core Competencies of supply chain for it is the basic method to solve conflicts and the crucial step for cooperation. We apply the technology of intelligence agent to the negotiation of supply chain and a negotiation relation network is put forward for negotiation among agent. In order to improve the efficiency and successful cooperation ratio, some key technologies of Case-Based Reasoning and Persuasive method used in the negotiation process are studied in detail. Based on the research of these theories, a negotiation model for supply chain base on multi-agent is presented. At last, A simulation prototype system for supply chain negotiation is developed based on the platform of JADE in order to verify the model validity. The main content of this dissertation are as follows:(1) Study on the relative theories of supply chain negotiation model. Firstly, we analyze the connotation characteristics of supply chain and the principle objectives of supply chain management. Secondly, we study on the relative theories of negotiation, which including the types of negotiation negotiation strategy、negotiation protocol. At last we also study on the theory of multi-Agent, which including the model of BDI and the negotiation mechanism of multi-Agent. These theories provide important theoretical basis for our study.(2)Study on the negotiation organization form among multi-agent based on the negotiation relation network of supply chain. In fact, we think that supply chain is an Inter-form which between the open market and single enterprise, so the negotiation environment and negotiation agents of supply chain have their own characteristics. It is important to improve the efficiency and successful cooperation ratio for negotiation between agents by building an appropriate negotiation organization form among multi-agent. A Multi-agent negotiation relation network and its building algorithm based on trust are presented. Also we study on the data structure and the organization framework of negotiation agents which are consider as the nodes of negotiation relation network. At last, the performance and characters of the negotiation relation network based on multi-agent also are researched.(3)Study on the technologies of Case-Based Reasoning (CBR) for Supply Chain Negotiation. We think that the negotiation among supply chain partners has its own characteristic for the negotiation partners are relatively stable and the negotiation cases have certain repeatability. We present that we can use the technology of Case-based Reasoning in the negotiation of supply chain partners. The negotiation agents can initialize their background information and predict the negotiation strategies and the reserved issue values of their opponents. According to the characters of supply chain negotiation, some key technologies of CBR are studied in detail, which including the technologies of XML and ontology in order to describe the semantic and syntax of negotiation case and the retrieve technology of similar case by using the ID3 algorithm and gray relevancy theory. Also the technologies of revise and reuse for negotiation case database are studied. At last, we put forward a Web logical framework model based on multi-agent and CBR for negotiation among partners.(4)Study on the mechanism of alternating negotiation issues based on persuasive. In this dissertation, the method of persuasive is used in the process of negotiation. The technologies of creation, alternation and evaluating of negotiation offer in the persua-sive process are studies. We also discuss the concession method、alternating principle and acceptable rule in the process of persuasive negotiation. In order to improve the successful negotiation ratio and get rid of the deadlock which maybe come out in the negotiation process, a method to define negotiation deadlock and a mechanism to get rid of the deadlock are present.(5)Study on a negotiation model for supply chain negotiation base on multi-agent. On the basis of these theories which are studied in this dissertation we present a negotiation model for supply chain negotiation base on multi-agent. At first, the negotiation model framework and the status of negotiation agent are defined, which provide the basis to formalize the negotiation model. Secondly, we study on the basic negotiation strategies for efficiency supply chain and responsive supply chain. When we studying the negotiation strategy, we applied the Bayesian learning to learn the opponent negotiation strategy and update his relief. The online learning method and learning process of Bayesian are expatiated in detail. Thirdly, we present a negotiation protocol which is universal and canonical for negotiation among supply chain partners. At last, we research on the negotiation process which is one aspect of the negotiation model and compare the negotiation model of supply chain with the relative negotiation models.(6)Based on the platform of JADE, a simulation prototype system for supply chain negotiation is developed. Through the analysis of traits of JADE and the running mechanism of multi-agent, the static structure, the interacting behavior and the communication method of multi-agent are designed in detail. The simulation results show that the negotiation efficiency and successful cooperation ratio are improved based on the negotiation model of supply chain.更多还原