【作者】 张磊；

【导师】 乐晓波；

【作者基本信息】 长沙理工大学 ， 计算机应用技术， 2008， 硕士

【摘要】 信息技术的高速发展给人们带来了极大的方便,但与此同时,发展中的信息网络也给人们提出了挑战,主要体现在网络设备复杂化使网络管理变得复杂,网络的经济效益越来越依赖网络的健壮运行。网络故障诊断的重要性越来越突出,网络故障诊断技术的研究,特别是网络的智能诊断技术研究已成为目前一个重要的研究领域。目前,基于多Agent的故障诊断系统能够有效的提高故障诊断的灵活性,加快诊断速度。在传统的多Agent诊断系统中,故障诊断过程往往采用案例与专家系统相结合的推理方式。基于规则的专家系统具有规则表现形式易读、知识变更容易,解释方便、推理过程简单等优点,但是存在知识获取困难及逻辑推理的组合爆炸问题。模糊Petri网是基于模糊产生式规则的良好建模工具,可以很好的描述网络故障的动态产生和传播过程。因此,本文提出一种基于FPN推理的多Agent网络故障诊断模型,即在模块化的多Agent诊断模型中,使用模糊Petri网推理算法进行网络故障诊断专家系统的推理,结合实例给出推理结果的可信度,并对系统性能进行了分析。本文的主要研究工作如下:(1)分析了模糊Petri网推理机制,探讨了模糊Petri网构建过程中的约简方法,研究了模糊petri网推理算法。(2)提出基于FPN推理算法的网络故障诊断方法,并将FPN推理算法嵌入多Agent诊断模块中,设计出一个基于多Agent的网络故障诊断系统模型,详细描述了系统模型中各组成部分的作用和各个模块的工作原理。(3)从JAVA语言角度,对系统各模块功能进行了设计。针对网络故障实例,进行故障诊断推理,最后进行了性能分析。更多还原

【Abstract】 The high speed developed information technique bring people tremendous convenience, but at the same time, the information network in the development also put forward a challenge to people, which the network equipments’ complication makes the network management become more complicated, the economic performance of network depends on the stabilize movement of network more and more. The research of the network fault diagnosis, especially intelligent diagnosis, has become an important field currently.Currently, fault diagnosis system based on multi-agent can raise the vivid of fault diagnosis effectively and release the diagnosis time. In the traditional multi-agent diagnosis system, fault diagnosis often uses the reasoning method that combining the case and the expert system. The expert system based on the rule has the advantage that the rule is easy to read, the changing, explaining and the reasoning is easy, but it has the problem, for example: the difficult to get the knowledge, the explosion during the logic reasoning. The fuzzy petri net (FPN) is a model tool based on the fuzzy rule, it can distribute the fault state creation and speed of the network. Therefore, this article puts forward a kind of network fault diagnosis model, which using fuzzy petri net reasoning diagnose the network fault in the model based on the multi-agent, it get the credibility during an example and analyze the system function.The researches of this thesis include:(1) To analyze the backward reasoning algorithm and discuss the method of simplifying a FPN model to reduce the calculation complexity.(2) The FPN reasoning is embedded into the diagnosis module of the mobile agent. The network fault diagnosis method based on fuzzy petri net reasoning is proposed. The function of the components and the working principle of the module are minutely illuminated.(3) The model is descripted using Java language. Taking a network fault for example, the diagnosis reasoning is circulated. At last the performance is analyzed.更多还原