基于案例推理的机车故障诊断专家系统研究

【作者】 赵明；

【导师】 黄汉永；

【作者基本信息】 中南大学 ， 计算机应用技术， 2004， 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故障诊断专家系统原型，以实现对机车故障的智能诊断。 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新兴起的基于案例推理方法的基础上，给出了基于案例推理的机车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，探讨了基于案例推理方法构造专家系统的关键技术，主要包括基于案例推理的知识工程以及案例检索技术。 在运用知识工程技术进行案例的表示、组织以及案例库的构建中，论文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，以基于特征的故障案例表示为基础，依据故障的特征属性和内容将机车故障案例按照故障种类分类，从而将整个案例库划分成了多个代表故障案例种类的案例集。在案例检索中，论文基于按故障分类构造机车故障案例库的思想，从故障案例种类中抽取出能与其它种类不同的关键特征，根据故障案例种类的层次结构，进而将整个案例库组织成多级分层索引结构。这种按故障分类，分层索引机制的策略思想不仅降低了案例表示、组织以及案例库的构建难度，而且极大地提高了案例检索的效率。 最后基于上述思想，设计和建立了系统的概念模型、物理模型，并构建了软件原型系统。通过对机车故障样本案例进行仿真测试，表明了设计思想是合理的，实现方法是有效的。更多还原

【Abstract】 Expert system is a computer system which emulates the decision ability of experts in certain field. Aiming at developing a locomotive fault diagnosis system, the paper proposed a series of approaches to solve the problem encountered in intelligent locomotive fault diagnosis on the basis of artificial intelligence and fault diagnosis.Based on the research on fault diagnosis technology in common use and burgeoning Case-based reasoning method in artificial intelligence, a scheme and a prototype of Case-based reasoning locomotive fault diagnosis is accomplished. Substantive log of maintenance and the characteristics of the knowledge in locomotive fault diagnosis is analyzed and the key technology of Case-based reasoning fault diagnosis system, such as case-based knowledge engineer and case index technology, is discussed in the paper. The strategy of layered index mechanism after the classification of fault is put forward originally.In expression and management of case and the construction of knowledge base with case-based technology, the locomotive fault case is classified as case sets by characteristic and detail according to the demand of fault diagnosis expert system and the peculiarity of locomotive fault diagnosis on the foundation of characteristic-based case expression. The multi-layered structure of the knowledge base of the system is formed in accordance with the key characteristic abstracted from fault cases, which reduce the complexity of case expression and management and construction of knowledge base and enhance the index speed and efficiency dramatically.Based on research above, the conceptual model and physical model is founded and the prototype of the system is finished. The emulation test to the software with locomotive fault cases in field proved the reasonability of the design and the validity of the implementation.更多还原