【作者】 江志华；

【导师】 朱国宝；

【作者基本信息】 武汉理工大学 ， 载运工具运用工程， 2005， 硕士

【摘要】 当代社会已经由工业社会转向信息社会。信息社会里知识得到充分的应用，企业中人的地位日益重要，科技进步步伐明显加快，产品生命周期越来越短，交通变得越来越便捷，各种先进通讯工具得到充分的利用，信息交流越来越容易，企业经营可以突破地理位置的束缚，进行横向和纵向的扩展，相关企业之间的联系变得更加复杂。对企业来说，在信息时代，在激烈的市场竞争中，只有摆脱以往陈旧的经营观念和管理手段，将最前沿的科学管理方法引入管理机制，才能获得持续发展的生机和活力，在这种条件下，无论是企业的组织形式还是管理方式都发生了巨大的改变。虚拟企业理论正是在这种知识经济、网络经济和知识管理的大背景下产生的，构建虚拟企业，实施经营，正逐步成为许多企业实现迅速发展的有效途径。 组建虚拟企业成为未来制造企业提高市场竞争力的有效手段。如何快速、低成本、低风险的集中最优秀的合作伙伴组建虚拟企业，便成为制造企业、专家学者所必须解决的首要问题。伙伴选择是建立虚拟企业的关键环节之一，直接关系到虚拟企业最终的成败。如何通过对各潜在伙伴企业的特征信息进行分析、整理、评估，确定一个最优的伙伴企业构成方案，继而确定各伙伴企业的参与方式，显得十分重要。 论文首先回顾了虚拟企业的产生背景及其发展过程，概述了虚拟企业特点，并较详细的介绍了虚拟企业以及合作伙伴选择的研究现状；接着介绍了目前研究得比较广泛深入的各种伙伴选择方法(AHP法、遗传算法、Agent法和模糊推理法等)，分析了它们的优点和不足，结合前人的研究，应用FNN法选择合作伙伴；之后重点介绍了FNN法的相关知识(模糊逻辑和神经网络)，并简要讨论了虚拟企业伙伴选择评价指标确定问题；最后在上述工作的基础上借用Matlab和ASP．NET编写了一个基于FNN法的伙伴选择框架，该框架充分地利用了Matlab强大的计算功能和．NET的丰富表现能力，实现了网上选择合作伙伴。 好的神经网络需要大量数据的训练，在FNN框架中可以灵活自山的动态调整训练数据集，以达到更好的应用效果；网络时代，作为一款强大科学计算工具的Matlab同网络无逢结合一直是研究者们的梦想之一，FNN框架也对这方面进行了有益的尝试，提出用共享文件的方式实现Matlab和Web交换数据。更多还原

【Abstract】 The contemporary society has already turned from industrial society to the information society. In the information society, knowledge gets abundant application, the status of the enterprise middleman is increasingly important, scientific and technological progress paces are obviously accelerated, products life cycle is shorter and shorter, the traffic becomes more and more convenient, various kinds of advanced communication tools get abundant use , information interchange is easier and easier, enterprises management can break through the geographical position , carry on horizontal and vertical expansion, the connection between relevant enterprises becomes more complicated. To enterprise, at information age, in the fierce market competition, only get rid of the outmoded management idea and management, lead the scientific office procedure of the most forward position into the mechanism of management, could obtain life and vigor of sustainable development. Under this kind of condition, the enormous changes have all taken place in both the organizational forms of enterprises and management style. The virtual enterprise’s (VE) theory was just produced under the great backgrounds of this kind of kownledge economy, cybereconomy and information management. Structuring VE is the effective way that a lot of enterprises realize being developed rapidly progressively.Setting up VE becomes the effective means that the manufacturing company raises the market competitiveness in the future. How the fast, most outstanding cooperative partner of concentration of low cost, low risk sets up VE, become primary problem that manufacturing company, experts and scholars must solve. The partner chooses to be one of the key links of setting up VE, concern VE ’s final success or failure directly. How to analyse , put in order , assess through the characteristic information to every potential partner’s enterprise, confirm a optimum partner’s enterprise form the scheme, then confirm the participation method of every partner’s enterprise, seem very important.The paper has reviewed the production background and evolution of VE at first,sum up the VE ’s characteristic , and the research current situation which VE of more detailed introduction and cooperative partner choose; Then has recommended studying the partner that must be more and choosing various kinds of methods (AHP method, hereditary algorithm , Agent methord and fuzzy reasoning method ,etc.) at present, analyse their advantage and deficiency, propose choosing the cooperative partner with the FNN law; Later introduced the relevant knowledge of FNN law especially (fuzzy logic and neural network), and has discussed briefly the VE ’s partner chooses the evaluation index to confirm the question; last using Matlab and ASP.NET on above-mentioned foundations of working writed a framework for selecting of VE partner, which has realized that chooses the cooperative partner on the net.The good neural network needs training for a large number of data, the dynamic adjustment that can be flexible and free in FNN framework trains the data collection, in order to reach better application result.In cybertimes, it is one of researchers’ dreams to combine Matlab, as a strong science computational tool, and network. The FNN framework carry on beneficial try to respect this too. is it realize between Matlab and Web exchange the data by means of sharing the file to propose.更多还原