【作者】 李伶；

【导师】 潘双夏；

【作者基本信息】 浙江大学 ， 机械设计及理论， 2004， 硕士

【摘要】 本论文通过对工程机械领域故障诊断技术的系统研究，提出了一种基于专家经验的以神经网络为推理核心的分布式在线故障检测和诊断系统模型，并引入先进的ARM技术进行了可行性试验研究。通过深刻分析维修专家进行设备维修时的思路，采用利于神经网络推理计算的知识表示方法和启发式推理机制实现了故障的定位，充分利用了自然语言描述的专家经验，很大程度地减少了诊断过程的工作量，提高了诊断的准确性和效率。 第一章 介绍了目前故障诊断技术发展的概况和挖掘机结构组成及工作原理。然后介绍本学位论文的研究意义和主要研究内容。 第二章 分析了挖掘机的功能结构和组成，对整个系统进行了分布式划分，建立了液压挖掘机故障分类层次模型。分析比较了各种现代故障诊断方法，状态监测与故障诊断技术在国内外的研究现状、应用及发展趋势，分析了故障诊断领域的主要理论及研究内容，并提出了神经网络与专家系统相结合的故障诊断系统方案。 第三章 详细介绍了本故障诊断系统的推理核心——神经网络模型。包括该模型的神经元构造，神经网络结构和学习算法，以及神经元权值和阈值的定义和意义，给出了构建这种神经网络模型的方法和步骤。 第四章 着重阐述了液压挖掘机故障诊断专家系统的功能结构设计和软件流程。重点介绍启发式诊断推理和神经网络并行推理相结合的故障诊断推理方法。 第五章 介绍了所采用的ARM技术的开发机制、特点和应用，以及发展前景。对Embest IDE for ARM开发平台高度集成化的特点也作了简要介绍。最后给出了便携式故障诊断仪的硬件结构框图并进行了可行性探讨。 第六章 总结了本课题的创新点和已经完成的工作，对课题接下来的发展方向和需要解决的问题进行了展望，并对后续的工作做了分析，提出了几点建议。更多还原

【Abstract】 Based on the systemic analysis of the configuration of the hydraulic excavator and the whole study of its fault diagnosis, this paper bring forward an on line, distributed fault diagnosis framework model which is based on masters experience and with a neural network as its consequence core. Then the fault diagnosis model is discussed about its feasibility with advanced RAM technique. Through deeply analysis of masters thinking when they check or maintain a set of equipment, faults of an fixture can be located by viable knowledge and heuristic consequence method. They are fit for the neural network exactly. The method can use maters experience adequately, so it can reduce the work of the procedure and enhance the precision and the veracity of the result to great extent. This paper can be divided into five chapters as follows:Chapter 1 summarized the development status of the diagnosis detection technology. Then introduced the composing structure and working principle of hydraulic excavators. At the end, introduced the significance and main content of this paper.Chapter 2 is about dividing a hydraulic excavators in distribute way, and founded a fault model with hiberarchy. According with the comparing and analysis of all kinds of modern fault diagnosis method, technique and their applications, brought forward the whole scheme of diagnosis system.Chapter 3 introduced the neural network model, which is the core of the diagnosis system. It is composed with nerve cell, structure of the network and the study arithmetic. And the definition of the threshold and authority is a important part of the neural network. The process and method to made a neural network is provided also.Chapter 4 expatiated the software flow and flow chart of the expert fault diagnosis system for hydraulic excavator. Pivot on the consequence method, thats integrated with heuristic consequence and neural networkconsequence way.Chapter 5 introduced the characters and applications of ARM technique, it can be adopted in a carry-home fault diagnosis apparatus for hydraulic excavator. Then research flat roof - Embest IDE for ARM is introduced in brief. At last, brought the hardware design of this apparatus and made some feasible discuss.Chapter 6 is an ending part. It summarized some innovative features in this papers, and the accomplished parts of this project. Then mentioned some advices and assumptions about the next step of the project.更多还原