【作者】 高长元；

【导师】 梁静国；

【作者基本信息】 哈尔滨工程大学 ， 管理科学与工程， 2002， 博士

【摘要】 当今世界高技术突飞猛进地发展，高技术产业的规模及水平决定着一个国家的经济发展水平和国际竞争力水平，甚至决定着一个国家的未来。为了尽快地发展我国的高新技术及其产业，国家实施了“863”计划和“火炬计划”。“863计划”和“火炬计划”的实施已使我国的高新技术及其产业有了较大的发展。因此，对高新技术产业评价、规划和产业政策制定等管理工作提出了更高的要求。 高新技术产品的认定与评价是高新技术企业和高新技术产业评价、规划、政策制定和管理的重要基础，目前我国对高新技术产品的划分和认定尚未形成具体的标准和方法，在火炬计划项目中认定高新技术产品的方法是专家会议评价法。这种方法存在着主观随意性大、指标简单、评价结果一次性和一刀切、不能实现实时滚动认定和后评价。在进行国际比较时，参照美国分类的10大高技术领域222类产品从我国的“海关统计商品目录”中分离出高新技术产品，并据此对我国的高新技术产品进出口情况进行统计分析，这种方法未能完全反映出我国的高新技术领域范围，只是在进行国际比较时参考。 根据我国高新技术产品认定与评价工作的需要，必须研究建立一套科学实用的评价方法。由于评价工作的复杂性，完全采用专家评审的方法不仅工作量大而且主观因素影响较大。采用计算机自动评审，由于很多因素无法确定和量化，因此也难以实现。高新技术产品认定与评价工作完全是一种半结构化问题，正适合于建立人机交互的决策支持系统(DSS)。因此，高新技术产品评价决策支持系统的研究对提高我国高新技术产业管理水平具有重要的理论价值和现实意义。 此外，由于决策支持系统的理论和方法还不很成熟，很多技术问题还有待解决，在开发高新技术产品评价DSS过程中，必须首先解决一系列与DSS有关的技术问题，特别是要解决基于Internet环境的DSS和群决策支持系统等方面的理论和技术问题。从这个意义上，本文对决策支持系统理论与开发方法的研究也具有重要的意义。 本文在分析和总结国内外高新技术产品认定与评价理论和方法、群决策支持系统理论和分布式人工智能理论的基础上，提出了采用基于多人多层次多指标综合方法的、建立在Internet环境的分布式智能群决策支持系统(DIGDSS)对高新技术产品进行认定与评价的思想。 在文章中综合运用Internet技术、群决策理论、决策支持系统理论设计了DIGDSS系统总体结构。该系统包括管理者使用的高新技术产品认定与评价管理子系统、企业远程申报子系统和专家使用的分布式智能群决策认定与评价子系统。 为了使模型库系统具有可重用性，研究了应用面向对象和数掘表达相结 哈尔滨厂程人学博十学位论文一合的模型库表达方式构造多人多层次多指标评价问题的通用模型库系统结构。该模型库系统可以实现模型的动态管理。 为了增强系统的智能性，打破了传统 DSS中问题处理系统（PPS）结构模式，研究了基于多Agent技术的问题牛成与问题求解系统运行模式。 应用改进的人工神经网络和遗传算法研究设计了知识发现系统中的模式总结系统。该系统可实现按高新技术产品领域分类训练神经网络，从而对专家评价模式进行机器学习和知识积累。 利用哈尔滨高新技术开发区的部分产品和企业数扼进行了系统模拟运行并给出了模拟结果。 本文的研究对改进我国高新技术产品认定与评价方法，提高高新技术产业管理水平具有实际意义；对Internet技术、分布式人工智能技术和群抉策支持系统的结合研究也具有理论意义。更多还原

【Abstract】 Today with the fast development of high technology, the scale and level of high-tech industries decides the level of economic development and international competitiveness, even the future of the nation. In order to accelerate the development of high technology and its industries of our nation, Chinese government has implemented "863 Project" and "Torch Plan". It brings forward higher requirements to the evaluation, planning and polices making of the high-tech industries.Identification and evaluation of high-tech products is the important foundation of the evaluation, planning, police making and supervising of high-tech enterprises and high-tech industries. Currently there are no specific standards and methods of high-tech products classification and identification. In "torch plan" projects the method of expert conference is used for the identification of high-tech products, which has some disadvantages, such as it is random because of subjective attitudes of experts, the indexes of which are too simple, its evaluation result is only valid once and less flexible, it can not carry out real time and rolling identification, and follow-up evaluation. In time of international compare, the American method that prescribes 10 high-tech fields and 222 kinds of high-tech products is referred, based on which high-tech products are separated from commodities catalogue of Chinese customs. Since this method fails to indicate our scope of high-tech fields, it only acts as a reference in international compare.A series of scientific and practical methods have to be established to meet the requirements of the identification and evaluation of high-tech products in China. Due to the complexity of evaluation work, the method of expert appraisement not only means a high working load but also strongly affected by the subjective factors. Computerized automatic evaluation is hard to realize because there are many factors that are uncertain and difficult to measure. The identification and evaluation of high-tech products is a semi-structure problem, which is suitable for human-machine interactive Decision Support System (DSS) . So the research on the DSS of high-tech products evaluation is of theoretical value and practical significance in improving the managerial level of Chinese high-tech industries.After the analysis and summarization of theories and methods of high-tech products identification and evaluation, the theory of Group Decision Support System (GDSS) and the theory of Distributed Artificial Intelligence from both domestic and overseas research, this thesis gives out an idea to make theidentification and evaluation on high-tech products by a Distributed Intelligence Group Decision Support System (DIGDSS) , which is set up in the internet environment based on combination of multi-person, multi-hierarchy and multi-objective methods.Through the integrated appliance of Internet technology, group decision theory and decision support system theory, it designs a general framework of DIGDSS. The framework includes the managerial system of high-tech products identification and evaluation for managers, long-distance application system for enterprises and the distributed intelligence group decision identification and evaluation system for experts.In order to equip the module base system with the feature of repeated use, it applies the object-oriented method and data expression manner, thus designs a general module base structure of multi-person, multi-hierarchy and multi-objective system. This module base can realize the dynamic management of modules and manage module base with data base management system.In order to strengthen the intelligence of the system, it breaks the framework of problem process system (PPS) of traditional DSS, it applies the agent technique of Distributed Artificial Intelligence, designs the operational structure of problem generating and problem solving systems. This structure realizes dynamic the module control.It designs the machine learning in knowledge discovering system with the appl更多还原