【作者】 程礼；

【导师】 莫易敏；

【作者基本信息】 武汉理工大学 ， 机械电子工程， 2007， 硕士

【摘要】 专家系统是一种模拟专家决策能力的计算机系统。论文以机车途中故障诊断为应用背景，利用人工智能、故障诊断的理论和方法，研究并开发了一个机车途中故障诊断专家系统原型，以实现对机车故障的智能诊断。本论文在研究目前常用的故障诊断智能技术以及人工智能领域中的基于规则推理方法的基础上，给出了基于规则推理和故障树技术的机车途中故障诊断专家系统的设计方案和原型系统。分析和研究了机车故障诊断领域知识特点和大量的故障维修日志，探讨了基于规则推理方法和故障树分析法构造专家系统的关键技术，主要包括基于规则推理的知识工程以及故障树检索技术。本论文以机车途中故障诊断领域为研究和应用背景，根据机车故障诊断领域的特点以及故障诊断专家系统的应用需求，对产生式规则表示法和故障树分析法等理论、方法和应用技术等一系列问题进行了深入的研究和探讨，主要研究工作有：1)经过现场的实地调查研究以及在对国内外大量资料的收集、整理、分析、研究的基础上，从应用实践的需求出发，在较高的抽象层次上探讨了机车故障智能诊断的基本内容和特点；并针对机车故障诊断领域的特点，利用混合推理策略和局部最佳优先搜索算法对机车途中故障进行诊断。2)对产生式规则和故障树技术作了深入的研究和探讨。采用产生式规则表示法的基本原理，借助故障树技术，设计基于规则的故障树表示方法；构造了一个按层次结构组织的多级索引机制的机车故障规则库。并在此基础上，设计了知识库，然后运用混合推理和局部最佳优先搜索算法进行匹配率计算的故障检索机制。3)基于以上的研究，设计和建立了基于规则推理的机车途中故障诊断专家系统的概念模型和物理模型。在此基础上，用Borland C++Builder5开发完成了软件原型系统。并在江西向塘机务段投入使用，反应良好，验证了设计的合理性和实现的有效性。最后，通过近一年来该系统在江西向塘机务段的应用实践表明，该机车途中故障诊断系统是切实可行的。更多还原

【Abstract】 Expert system is a computer simulating systems of the expert decision-making ability. This Paper taking locomotive fault on the railway diagnosis as application background, using artificial intelligence, fault diagnosis theory and methods, studied and developed a diagnostic system of locomotive fault on the railway to achieve to the locomotive fault intelligent diagnosis.The paper produced design proposal and the prototype system of diagnose expert system of the locomotive fault on the railway based on the RBR and the Fault-tree technology, on the base of studying the fault intelligent diagnosis technology at present in commonly used and the RBR method in artificial intelligence domain. The paper analyzed and studied the knowledge characteristic of the locomotive fault diagnosis domain and massive fault service diary. And the paper discussed essential technology of structuring expert system based on the RBR method and the Fault-tree analytic method, which mainly includes the knowledge engineering of the RBR and the Fault-tree retrieval technology.The paper taking locomotive fault diagnosis domain as research and application background, according to the locomotive fault diagnosis domain characteristic and application demand of fault diagnosis expert system, studied thoroughly a series of questions such as the RBA method and the Fault-tree analytic method, and which main research work includes following several.After scene on-the-spot investigation research and on the basis of collecting, collating and anglicizing a large number of domestic and foreign information, according to application demand of the practice, discussed basic content and the characteristic of the locomotive fault intelligence diagnosis in a higher abstract level, and using mix reason strategy and partial best-first search algorithm to diagnose locomotive fault.The paper researched and discussed thoroughly the production pattern rule and the Fault-tree technology. Using the basic principle of the production pattern regular method of portrayal, with the aid of Fault-tree technology, designed the RBR Fault-tree expression method.Designed and established the conceptual model and the physical model of the locomotive fault diagnosed expert system. And put it to use in the Xiangtang locomotive depot in Jiangxi province, get a good appraisal, and which confirmed the design rationality and the realization validity.更多还原