节点文献

电控发动机故障诊断专家系统的研究

Study on Diagnosis Expert System of Electric Controlled Engine

【作者】 王丽荣

【导师】 韩宗奇;

【作者基本信息】 燕山大学 , 车辆工程, 2004, 硕士

【摘要】 电子控制技术在汽车上的广泛应用,使汽车的结构、性能发生了根本性的变化,新的结构原理和装置相继涌现,对汽车维修人员提出了更高的要求。目前,我国汽车诊断技术跟国外相比还有很大差距,电控发动机故障诊断专家系统的研究对尽快改善和提高我国汽车检测诊断技术尤为重要。  本文阐述了电子控制技术在发动机上的使用情况,总结出电控发动机的故障特点,深入分析了汽车故障诊断理论和技术、传统专家系统的原理和局限性、基于神经网络技术的专家系统的原理和优点,进一步采用将基于符号处理的传统专家系统和基于神经网络技术的专家系统相结合的集成故障诊断专家系统。本文的核心部分是建立确定性故障现象的知识表示和推理策略、故障码诊断和数据流分析模块设计、基于神经网络技术的故障波形诊断的软件设计;根据发动机的结构、功能原理及维修专家的经验,建立故障树,由故障树提炼出训练样本供神经网络学习,学习后生成的权值和阈值组成了知识库的一部分。文中对BP网络及其在sigmoid函数激发下的算法步骤、程序设计作了详细的说明。该软件在Windows2000操作系统中,用Visual Basic6.0和Access技术开发设计的。系统收集了部分车的故障码、数据流分析、点火系统检测方法等,以方便用户添加、删除、修改和查询。如何更好地实现在线检测与故障诊断,使故障诊断简单化,智能化,建立适合我国国情的发动机故障诊断系统,并能在实践中得以推广使用,是今后故障诊断系统研究的主要方向。因此对电控发动机故障诊断专家系统的研究具有重要的经济价值和广阔的应用前景。

【Abstract】 The wide use of electronic technology in automobile totally changes the automotive structure and performance, new structure theory and devices are springing up and putting forward of higher demands to the maintainers. Now compared to the foreign countries, there were a wide gap of domestic fault diagnosis technology, the study of diagnosis expert system of electronic controlled engine has significance to improved and enhance domestic fault diagnosis technology.This paper elaborates the utilization conditions of electronic technology in engine and sums up the fault feature of electronic controlled engine, deeply analyze the theory and technology of fault diagnosis, theory and limitation of traditional expert system, theory and advantage of the expert system based on neural network. Integrated expert system base on fault diagnosis combined by base on symbol processing and expert system based on neural network is further put forward.The core of this paper is to establish knowledge representation of deterministic fault appearance and reasoning strategy, module design of fault code diagnosis and data flow analyze, software design of fault shape diagnosis based on neural network. According to the knowledge of the structure, principle of the engine and the experiences from experts, the fault trees are established. Extracting the training samples from fault trees is given to neural network for study and for the formation of part acknowledge database of weight value and threshold value after study. In this paper there was a detail account of BP network and its algorithm steps stimulated by "sigmoid" function.The software is designed by VB6.0 and Access, its operation system is windows2000. This system has collected fault code, data flow analysis, inspection method of ignition system and so on, the user can add, delete, alter and inquire conveniently.The main direction of studying to diagnosis expert system is how to better realize inspection and diagnosis on line, establish a sort of diagnosis system that is simple intelligent, adapted to the conditions of our country and can be popularized in practice. So, the studying to fault diagnosis expert system has important economic value and wild utilization prospects.

  • 【网络出版投稿人】 燕山大学
  • 【网络出版年期】2004年 04期
  • 【分类号】U472
  • 【被引频次】11
  • 【下载频次】962
节点文献中: 

本文链接的文献网络图示:

本文的引文网络