发布时间 : 星期三 文章信息融合电路应用论文更新完毕开始阅读5c025a0ceff9aef8941e063a
Abstract
Analog circuit fault diagnosis is a new multi-subject-crossed technique. It has developed a great deal of theory and approaches for fault diagnosis in the past forty years, which keep at a distance with application. With the rapid development of the Electronic Industry, the importance of the analog circuit fault diagnosis is more and more obvious. It has an important signification for working orderly and reliability design of electronic equipment or system, which need some new and effective theories and methods.
Information fusion is a subject formed recently, it has been researched and applied in many fields, however, it is still in starting stage in fault diagnosis of analog circuit. This paper introduces the integration technology of information to the fault diagnosis of analog circuit and set up the general frame of fault diagnosis based on the integration technology of information. Then, a specific diagnosis method was proposed and researched deeply.
To the incertitude of fault diagnosis and based on evidence theory, a decision-making fusion method is researched and provided in this paper, with a simulate demonstration to analize. Fault diagnosis method was researched deeply on the basis of neural network-information fusion, theory and realize steps of neural fusion fault diagnosis were proposed, then instantiation simulation was given to prove that use this method can greatly improve the recognition rate of fault diagnosis. One fault diagnosis model and corresponding algorithm was constructed based on neural network and evidence theory for taking a step forward diagnosis correct rate, which can cut down the imput dimension of neural network、improve classification ability、decrease the error classify rate of diagnosis system. Then, the feasibility and effectiveness of this method was manifested by specific diagnosis example.
To conquer the problem of acquisition high effectiveness in analog circuit fault imformation, wavelet transform was researched in this paper to preprocess the fault signals. Meanwhile, BP neural network diagnosis method based on wavelet transformation was proposed, this method can abstract fault feature effectively and decreases the dimension of imput vector. By this mean, the construct of neural network can be simplified and training time was
3
economized, then the recognize ability of fault type was improved also.
Key Words: Analog circuit; Fault diagnosis; Information fusion; Evidence theory;Neural
network; Wavelet transform
4
绪论
随着电子工业的迅速发展，现代化工程技术系统的规模不断扩大，模拟电路的应用越来越广泛，如在军工、通讯、家用电器、自动控制、测量仪表等各个方面，模拟电路都是重要的组成部分。电子产品日趋大型化、高速化、自动化和智能化，不仅同一电子产品的不同部分之间互相关联，紧密耦合，而且不同电子产品之间也存在着紧密的联系，在运行过程中形成一个整体。因此，一处故障可能引起一系列连锁反应，导致整个电子产品系统甚至整个过程不能正常运行，轻者造成停机、停产，重者会产生严重的甚至灾难性的人员伤亡。在我国，产品的维修费用占总费用的80%，而研发和采购仅占20%。因此，切实保障现代复杂系统的可靠性和安全性，具有十分重要的意义。
故障通俗的说就是“异常”，是指产品或产品的一部分不能或将不能完成预定功能的事件或状态。一般电路的故障诊断可分为模拟电路故障诊断和数字电路故障诊断。数字技术的广泛应用和高速发展，使得数字电路的故障诊断研究取得了空前的发展。对于模拟电路，由于其元件具有容差并存在非线性等原因，使得模拟电路的故障诊断较数字电路的故障诊断复杂得多，其发展比较缓慢，应用也不够广泛。但是，在一个完整的系统中，数字电路并不能完全取代模拟电路。经典的故障诊断主要依靠模拟式仪表，诸如：信号发生器、电子电压表、示波器等，它还要求操作人员应具有一定的理论基础和丰富的实践经验。通常情况下，它的测试速度较慢，测试准确性低。随着电子电路集成化程度越来越高，换代更新迅速，应用也日趋广泛，人们已经逐渐认识到，对故障诊断有必要重新研究，必须把以往的经验提升到理论高度，同时在坚实的理论基础上，系统的发展和完备一整套严密的近代故障诊断方
5
法，并结合先进的计算机数据处理技术，实现故障诊断的自动检测、定位、定值以及预报。自动故障诊断的关键在于诊断程序的产生，而中心问题在于故障诊断理论， 因此，模拟电路的故障诊断研究成为了电路领域世界各国的研究热点。
模拟电路故障诊断理论是以网络理论为基础的，发展至今已成为网络理论中公认的第三大分支。网络理论的发展大致可分为三个阶段：网络分析阶段、网络综合阶段和故障诊断阶段。
在工程技术领域，故障诊断可描述为：根据技术装备若干个可以直接测量的信号，推断装备是否正常，确定故障部位和预测故障的发生。一般来说，故障诊断可以实现故障检测、故障定位、故障辨识、监控等功能。模拟电路传统的研究方法对于模拟电路的故障诊断，比较简单，通用的方法是分类诊断法，其理论基础为模式识别中的统计方法和决策理论。对于模拟电路，其诊断方法的分类一般依据电路仿真在实际测试之前或之后来划分。如果对电路的仿真是在现场测试之前实施，则称为测前模拟诊断；反之，则称为测后模拟诊断。
目前，常见的人工智能技术主要包括专家系统、神经网络、模糊理论、小波变换等。人工智能技术由于其善于模拟人类处理问题的过程，容易顾及人的经验以及具有一定的学习能力等特点在模拟电路故障诊断领域得到了广泛的应用。由此开发出的综合自动故障诊断系统，对于难以建立数学模型的电路的故障，可以实现故障的快速、准确定位，使检修人员对问题的认识更具有全面性、有效性、针对性。
6