基于CLIPS的光电跟踪设备故障诊断专家系统研究

发布时间：2018-03-07 14:46

本文选题：故障诊断　切入点：专家系统　出处：《中国科学院研究生院（光电技术研究所）》2015年硕士论文　论文类型：学位论文

【摘要】：光电跟踪设备是集光机电于一体的大型光电设备,其层次结构复杂,单元部件多,采用传统的建模方法和信号处理方法均不能从系统层面上实现故障诊断的要求,而人工智能领域中的专家系统可依据领域经验知识与理论知识实现故障诊断。近年来专家系统在航空航天领域大型设备故障诊断方面取得了突破进展,为光电跟踪设备故障诊断系统的研究奠定了技术基础。本文分析了光电跟踪设备的组成和工作流程,对设备故障进行了分类并总结了各类故障的底层故障原因,然后分别以妨碍设备运行的主要故障事件为顶事件建立了光电跟踪设备故障树。故障树是一颗由故障现象到故障原因的倒置树,具有逆因果性,可以显示故障知识相关信息以及故障间的联系。根据设备故障特点,采用了产生式规则和框架表示法相结合的知识表示方式将故障树表示的设备故障知识转化为电脑中可存储的知识,建立专家系统知识库。采用正向推理方式与高置信度优先的控制策略设计推理机,其中置信度是由故障发生概率与排查难易程度共同决定的,并设计了专家规则来解决知识间的相斥情况。本文采用擅长编写谓词逻辑、语义分析类程序的专家系统编程语言CLIPS设计专家系统核心部分:知识库和推理机,实现故障诊断;采用VC的应用程序框架MFC设计专家系统界面,实现友好的人机交互;在数据库SQLSEVER中设计了故障知识表、维修建议表和诊断案例表存储设备故障相关知识,以便更好地维护和管理知识。本文采用直接嵌入法和ADO对象技术将CLIPS与SQLSEVER嵌入MFC,同时利用代码自动生成技术实现数据库与CLIPS知识库的同步更新,协调这三个部分共同运行,完成光电跟踪设备故障诊断专家系统的设计。本文最后对专家系统主要功能做了介绍并通过具体故障实例对系统进行了测试,测试结果表明该故障诊断专家系统能准确诊断设备故障并对故障知识进行有效管理。
[Abstract]:Photoelectric tracking equipment is a large optoelectronic device, which has complex hierarchical structure and many unit parts. The traditional modeling method and signal processing method can not realize fault diagnosis from the system level. The expert system in the field of artificial intelligence can realize fault diagnosis according to the field experience and theory knowledge. In recent years, the expert system has made a breakthrough progress in the field of large-scale equipment fault diagnosis in the field of aeronautics and astronautics. It lays a technical foundation for the research of the fault diagnosis system of photoelectric tracking equipment. This paper analyzes the composition and work flow of the photoelectric tracking equipment, classifies the faults of the equipment and summarizes the underlying fault causes of all kinds of faults. Then the fault tree of photoelectric tracking equipment is established based on the main failure events which hinder the operation of the equipment. The fault tree is an inverted tree from the fault phenomenon to the cause of the fault, which has the inverse causality. Can display fault knowledge related information and the relationship between the fault. According to the characteristics of the equipment failure, In this paper, the fault knowledge of equipment represented by the fault tree is transformed into the knowledge that can be stored in the computer by the way of knowledge representation which combines the production rule and the frame representation. The knowledge base of expert system is established. The inference engine is designed by using forward reasoning and high confidence priority control strategy, in which the confidence is determined by the probability of fault occurrence and the degree of difficulty in checking. In this paper, the expert system programming language CLIPS, which is good at writing predicate logic and semantic analysis program, is used to design the core parts of expert system: knowledge base and inference engine, to realize fault diagnosis. The expert system interface is designed by using VC application program framework MFC to realize friendly man-machine interaction, and fault knowledge table, maintenance suggestion table and diagnosis case table are designed in database SQLSEVER. In order to better maintain and manage knowledge, this paper adopts direct embedding method and ADO object technology to embed CLIPS and SQLSEVER in MFC. at the same time, we use code automatic generation technology to realize synchronous updating of database and CLIPS knowledge base, and coordinate these three parts to run together. Finally, the main functions of the expert system are introduced and the system is tested by a specific fault example. The test results show that the fault diagnosis expert system can accurately diagnose equipment faults and manage fault knowledge effectively.
【学位授予单位】：中国科学院研究生院（光电技术研究所）
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN29

【参考文献】