《基于人工神经网络干衣机故障监测诊断系统论文》由会员分享，可在线阅读，更多相关《基于人工神经网络干衣机故障监测诊断系统论文（65页珍藏版）》请在金锄头文库上搜索。
1、A Dissertation Submitted to Shanghai Jiao Tong University for theDegree of MasterA STUDY ON FAULT DETECTION ANDDIAGNOSIS OF DRYER BASED ON ARTIFICIALNEURAL NETWORKAuthor：Hou XiaoSpecialty:Control EngineeringAdvisor: Prof. Yang YuPuAdvisor: Prof. Hu ZhengYuSchool of Electronics and Electric Engineeri
2、ngShanghai Jiao Tong UniversityShanghai, P.R. ChinaJune , 2007基于人工神经网络的干衣机故障监测诊断系统摘要随着人民群众生活水平的日益提高，家用电器已进入了千家万户，使用频率也非常高，有时甚至感觉一日都离不开它们。因此如果其质量和可靠性出现问题造成停机或更严重的是伤害事故，势必会影响人们的生活。对于家电产品的故障诊断与故障机理分析、性能测试技术及可靠性理论与可靠性试验技术的研究，对家电产品质量与可靠性的提高无疑是具有重要意义。由于家电产品一般是长寿命可修复的复杂机电产品，其内部由控制器类、电机类、加热制冷类以及空气流通类等不同特性的子
3、系统组成，由于它们的故障机理不同，其故障分布也是不同的，另外它们在不同的使用环境条件下，也会表现出不同的故障分布。因此生产企业要研究这些故障分布，常用的研究方法之一是进行可靠性试验来获得第一手的故障数据和资料。获得故障数据和原因的方法有很多，重要的是要寻求实用的、适合自身的，并且也是能持续发展的方法。本文在分析了各种现有的故障监测和诊断方法和人工神经网络原理的基础上，提出了一种可应用于干衣机系统可靠性试验中故障的监测和诊断的方法，即基于故障树分析并结合神经网络多传感器信息融合的故障监测和诊断。在故障监测和诊断中，采集有效的设备运行状态是基础。通过利用神经网络的软测量技术，解决了那些由于硬件上或
4、成本上限制，信号无法直接测量的问题，为实现故障诊断创造条件。在分析了人工神经网络技术在故障监测和诊断系统中固有的优越性后，将神经网络技术引入到该系统中来。同时用故障树技术和神经网络技术相结合的方法，分析了故障原因和特征之间的因果关系。研究了多传感器信息融合技术，通过实验表明多传感器信息融合要优于单传感器检测系统。最后利用虚拟仪器平台构造了基于神经网络的故障监测和诊断系统，应用了结构化的程序设计方法，使程序具备很强的扩展性和生命力，实现了系统的可靠、稳定运行。关键词：故障监测与诊断，人工神经网络，软测量，家用电器A STUDY ON FAULT DETECTION AND DIAGNOSIS O
5、F DRYERBASED ON ARTIFICIAL NEURAL NETWORKABSTRACTHome appliances, one kind of important electromechanical product, arewidely used in various homes. People utilize them almost every day, so ifthere were any faults or even serious incidents occurred, they would impactpeoples life negatively. Many home
6、 appliance makers are studying on thetheories and the technique of the product reliability, i.e. the study on faultmechanism and fault diagnosis, reliability theories and testing method, toestimate and improvement of their quality and reliability continuously.Because that home appliance is one kind
7、of long-life, repairable,complex product, it usually consists of electronic controller, heating/coolingsystem, motor drive system, airflow system, and etc. Meanwhile the applianceused under different environmental condition, so the fault distributions arevarious. One method of the study on the distr
8、ibution is to run the reliabilitytest and acquire data. We can use different method to detect and diagnosefaults, but the most important thing is to find the practical and new diagnosismethod. After studying various conventional fault diagnosis method andartificial neural network, the way of fault d
9、etection and diagnosis based onartificial neural network and fault tree analysis for dryer is provided in thispaper.To acquire data of the running dryer is a basis for the fault detection anddiagnosis, but since the constraint of the hardware and/or cost, some signalcould not be measured directly, t
10、he soft sensing technique based on the neuralnetwork is provided to solve these issues.Presenting the advantage of neural network technology, neural networkis introduced to the system. For decreasing the uncertainty of the diagnosissystem, the information fusion pattern based on neural network is di
11、scussed.The approach of multi-sensor data fusion based on neural network is shown tobe practical and effective through comparing the results of experiment.The concept and structure of virtual instrument is discussed deeply. Thenwe designed a new universal platform of dryer fault detection and diagno
12、sissystem, which is based on the virtual instrument and an integration ofartificial neural network. Meanwhile, modularization design technique isapplied fully on the programming, so the system can work stably and could beupdated easily.Keywords: Fault detection and diagnosis, Artificial neural network, Softsensing, Home appliance目录摘要ABSTRACT第一章绪论 . 11.1课题研究意义和目的. 11.2故障监测和诊断概述. 21.3国内外研究现状和发展趋势. 31.4本文的主要工作. 5第二章故障诊断技术和人工神经网络技术 . 72.1故障诊断技术的定义. 72.2故障监测和诊断技术综述.
