基于边界扫描的模拟电路故障诊断

发布时间：2018-05-07 05:07

本文选题：边界扫描 + Wiener级数　；参考：《哈尔滨理工大学》2015年硕士论文

【摘要】：模拟电路故障诊断在电路测试领域占据着重要地位，一直以来都是研究人员关注的对象。随着科技的发展，电路板已经进入大型化与高度集成化的时代，电路可及节点少使得电路故障诊断变得越来越困难，能否高效准确的解决可及节点少的问题、获得大型集成模拟电路的故障信息显得尤为重要。本文针对这一问题提出了基于边界扫描的模拟电路故障诊断方法，用“虚拟探针”代替传统的物理探针，大大提高了电路的可测试性，利用Wiener级数描述非线性模拟电路，并通过训练神经网络实现电路状态的在线识别，完成电路故障诊断。本文分析了边界扫描技术中混合信号测试的IEEE1149.4标准，介绍了在模拟电路诊断中边界扫描所使用的指令。在此基础上，为了将边界扫描更好的应用于模拟电路，提出了三种基于边界扫描IEEE1149.4标准的电路可测性设计方法，，为利用边界扫描进行各种模拟电路的故障诊断奠定了基础。针对模拟电路的非线性问题，用Wiener泛函级数描述模拟电路并研究了Wiener核的获取方法。用MATLAB编程提取了基于边界扫描的电路各状态Wiener核，又将前三阶Wiener核及相应状态的编码作为训练样本训练神经网络，通过训练好的神经网络在线识别出电路状态，达到智能故障诊断的目的。在进行理论研究和仿真的同时，本文设计了基于ATmega128单片机的故障诊断系统，其中硬件包括JTAG控制器、信号发生模块、数据采样模块、数据存储模块等。软件设计包括故障诊断系统、上位机软件、神经网络库的生成。可实现获取Wiener核、提取核特征以及训练学习神经网络等功能。最后以包含了具有边界扫描特性的STA400芯片在内的实际电路为例，对电路的各故障状态进行诊断，在上位机界面显示故障类型和故障元件编号，实验结果验证了基于边界扫描与Wiener核的电路故障诊断方法的可行性。
[Abstract]:Analog circuit fault diagnosis plays an important role in the field of circuit testing. With the development of science and technology, the circuit board has entered the era of large-scale and highly integrated. The lack of accessible nodes makes the circuit fault diagnosis more and more difficult, and can solve the problem of less nodes efficiently and accurately. It is very important to obtain fault information of large integrated analog circuits. In this paper, a fault diagnosis method for analog circuits based on boundary scan is proposed. The "virtual probe" is used to replace the traditional physical probe, which greatly improves the testability of the circuit. The nonlinear analog circuit is described by Wiener series. The on-line recognition of circuit state is realized by training neural network, and circuit fault diagnosis is completed. In this paper, the IEEE1149.4 standard of mixed signal testing in boundary scan technology is analyzed, and the instructions used in boundary scan in analog circuit diagnosis are introduced. On this basis, in order to better apply boundary scan to analog circuits, three circuit testability design methods based on boundary scan IEEE1149.4 standard are proposed, which lays a foundation for fault diagnosis of various analog circuits using boundary scanning. In order to solve the nonlinear problem of analog circuits, the Wiener functional series is used to describe the analog circuits and the method of obtaining Wiener cores is studied. The Wiener kernel of each state of the circuit based on boundary scan is extracted by MATLAB programming, and the coding of the first three Wiener cores and corresponding states is used as the training sample to train the neural network, and the circuit state is recognized online by the trained neural network. To achieve the purpose of intelligent fault diagnosis. At the same time of theoretical research and simulation, a fault diagnosis system based on ATmega128 microcontroller is designed. The hardware includes JTAG controller, signal generation module, data sampling module, data storage module and so on. Software design includes fault diagnosis system, upper computer software, neural network library generation. It can obtain Wiener kernel, extract kernel feature and train learning neural network. Finally, taking the actual circuit including the STA400 chip with the characteristic of boundary scan as an example, each fault state of the circuit is diagnosed, and the fault type and fault component number are displayed in the upper computer interface. The experimental results verify the feasibility of the circuit fault diagnosis method based on boundary scan and Wiener core.
【学位授予单位】：哈尔滨理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN710

【参考文献】