【作者】 王辉；

【导师】 彭良玉；

【作者基本信息】 湖南师范大学 ， 电路与系统， 2015， 硕士

【摘要】 信息处理技术在当今得到了快速发展,电子设备中的电路变得日益复杂,由模拟电路引起的设备故障问题,要得到有效处理却日益棘手。集成电路集成度的变高,元器件本身固有的不稳定性等原因给快速定位故障及处理故障带来更大挑战。面对众多出现的问题,传统故障诊断方法已经不能满足社会发展需求,新的诊断技术迫在眉睫。各国研究者开始尝试新的理论研究,其中神经网络作为智能技术运用于模拟电路故诊断研究得到快速发展,在新的诊断技术方面开辟了新路径,并在一段时间内取得了丰硕的成果。现如今,广大学者开始重视将小波分析,遗传算法等多种理论及其融合理论结合神经网络进行故障诊断的新技术,这为智能化故障诊断技术提供了新的思路。LabVIEW软件作为一款功能强大的图形编程软件,可以提供良好的人工交互界面,已经开始运用于故障诊断技术中,为实现故障诊断的简易化提供了便捷之路。本文以新的诊断技术为背景,将小波分析,遗传算法理论融合到神经网络,结合虚拟仪器(Lab VIEW平台),实现电路故障的可视化诊断。介绍了模拟电路故障诊断的研究背景意义、国内外发展现状、存在问题及分类方法。概述人工神经网络理论,包括其特点、应用以及学习方式。以BP神经网络理论为基础,对小波神经网络结构进行构造及其改进算法进行详细讲解,通过仿真实例进行验证所提算法的正确性,其中包括使用软件ORCAD10.5对待诊断电路进行原始数据提取;利用MATLAB软件平台编程对数据进行多分辨分析,提取故障特征值,构造样本集;基于小波神经网络故障诊断方法的实现:针对神经网络权值问题,利用遗传算法进行优化,改善网络性能,最后通过Lab VIEW软件平台实现编写程序的图形化,搭建神经网络模拟电路故障诊断系统界面,实现诊断过程的可视化,操作简易化。更多还原

【Abstract】 With the quick development of information processing technology today, the circuit in electronic equipment become increasingly complex, thus to deal with equipment failure problems caused by analog circuits effective become increasingly difficult. As the integrated circuit getting higher and the inherent instability of component itself and other reasons,it is more challenges to locate faults and process faults quickly.Faced with numerous problems, the traditional fault diagnosis methods have been unable to meet the needs of social development and new diagnostic techniques demand is imminent.Researchers from different countries began to try new theory, which the neural network as an intelligent technology applied in analog circuit diagnosis research get fast development, and has opened up a new path in term of new diagnostic technologies, and obtained a number of achievements over a period of time.Nowadays, more and more scholars have begun to pay attention to the new technology that many theory based on wavelet analysis, genetic algorithm and fusion theory combined with neural network processing fault diagnosis,which provides a new way forintelligent fault diagnosis technology.Lab VIEW software as a powerful graphical programming software, can provide good human interaction interface, has been applied to fault diagnosis technology and provides a convenient way to achieve the facilitation of the fault diagnosis.Based on the new diagnostic technique,this paper integrate the wavelet analysis and the genetic algorithm theory into neural network, combined with virtual instruments(Lab VIEW platform), to achieve the visualized diagnosis of electrical faults. This paper introduces research background and significance of fault diagnosis of analog circuits;domestic and international development present situation;the existing problems and classification methods. Then overview of artificial neural networks theory, including its features, applications and learning styles. Based on BP- neural network theory,this paper construct the structure of wavelet neural network and introduces its improved algorithm in detail,and validate the correctness of the proposed algorihtm through the simulation example, including use the ORCAD10.5 software to extract the raw data in analog circuit diagnosis;using the MATLAB software platform programming for multi-resolution analysis of the data, extracting the fault characteristic valueand constructing the sample set; the achievement of the fault diagnosis based on wavelet neural network method. For the problem of neural network weights, by using the genetic algorithms to optimize and to improve network performance.Finally it realize graphical programming and builds the interface of analog circuit fault diagnosis system through Lab VIEW software platform, then diagnosis is visualization and simplification.更多还原