电路板故障诊断中神经网络信息融合技术的应用

将信息融合技术应用到导弹电子设备的电路板故障诊断之中．提出了一种基于红外热成像实验的温度故障隶属度函数构造形式以及改进的BP神经网络算法。进行了BP神经网络信息融合故障诊断的实验研究．从结果看信息融合能较好地解决电路板元件故障诊断的不确定性问题。     

基于单片机控制的实验线路故障诊断系统

介绍了基于单片机控制技术的强电实验线路故障自动诊断系统的结构及功能,阐述了系统硬件设计、软件设计的基本原理．     

Analysis of Spectral Reflectance Using Normalization Method from Endoscopic Spectroscopy System

Objective assessment of gastrointestinal mucosal color is extremely important in the endoscopic diagnosis of digestive tract disease. In this paper, we propose a method to clarify the spectral characteristics of gastric and colon cancer. A large number of spectral reflectance data of mucous membrane are measured by the endoscopic spectroscopy system (ESS) in the National Cancer Center Hospital East, Kashiwa, Japan and the Department of Internal Medicine, Self-Defense Force Center Hospital, Tokyo, Japan. We assume that early cancer appears primarily in the spectral data of short wavelength, because it is usually present in a superficial cell where short wavelength light is scattered more strongly than long wavelength light. To identify the features in the short wavelength components, the spectral reflectance was divided by the reflectance of a long wavelength. We investigated the possibility of distinguishing early cancer from normal spectral data through statistical analysis, employing the projection axis as the mean difference between them. Early cancer and normal spectral data were projected on the projection axis, and the Student’s T-test was applied to evaluate the mean of the distribution between these data.     

模拟电路的模块级故障诊断

本文以多端口元件或子网络为基本单元，提出了一种模块级故障诊断方案，这种方案的故障模型更符合电路的实际情况，因而具有更广泛的实际意义，文中首先推导了模块故障诊断议程，并对模块故障可测性问题进行了一般性讨论。然后，给出了模块故障可测的几个拓扑条件，并通过定义广义独立通路使这些拓扑条件适用于一般的非互易网络。     

Measurement selection for parametric IC fault diagnosis

This article presents experimental results which show feedforward neural networks are well-suited for analog IC fault diagnosis. Boundary band data (BBD) measurement selection is used to reduce the computational overhead of the FFN training phase. We compare the diagnostic accuracy between traditional statistical classifiers and feedforward neural networks trained with various measurement selection criteria. The feedforward networks consistently perform as well as or better than the other classifiers in term of accuracy. Training using BBD consistently reduces the FFN training efforts without degrading the performance. Experimental results suggest that feedforward networks provide a cost efficient method for IC fault diagnosis in a large scale production testing environment.This work is supported by NSF-IUC CDADIC, Project 90-1.     

一种用模拟信道进行数字通信的方法

本文介绍了一种利用现有机载模拟通信设备进行数字通信的高性能机载报警系统。文中，着重论述了该系统在编码和解码技术方面，提高报警信号的抗干扰性、保密性、防止误码以及提高传输距离等关键技术；较详细地介绍了系统的工作原理及调制解调电路的特点。     

一个光纤综合业务环网的故障自动恢复

光纤综合业务区域网是近年来迅速发展的一个领域，环形网是这种网最常用的一种拓扑结构。由于环形网结点串联的特点，网络的可靠性是必须解决的问题。本文针对一个具有３４Ｍｂ／ｓ速率，可以传输计算机数据、话音、传真和会议电视的光纤综合业务环网（ＩＳＲ），研究了其故障的自动隔离。文中分析了ＩＳＲ的故障，讨论了故障的检测、定位和隔离的方法。为了适应综合业务的混合电路交换和分组交换技术，本文提出了采用二分搜索法来定     

基于主元分析的多变量统计过程的故障辨识技术

为了更好的进行故障检测与诊断，介绍了主元分析理论，给出了基于主元分析的过程故障辨识机理及策略。仿真实例表明，利用此方法建立的故障诊断模型，能够在不依赖过程机理的前提下高效抽取原始数据空间的主要变化信息，对过程的非正常变化做出反应，同时还能较正确地找出发生故障的原因以及相应环节。     

Railway wheel fault diagnosis using a fuzzy-logic method

A simple method for diagnosis of railway wheel defects using fuzzy-logic is proposed. The method is based on vibration measurements at different train speeds on healthy wheels and wheels with defects known a priori. The measured vibrations are statistically analyzed and confidence intervals for healthy and defective wheels depending on train speed and frequency of analysis are established. To facilitate the implementation of the method a fuzzy-logic is adopted. The fuzzy-logic model stores the obtained experience in a database and performs the decision making on damage extent and consequently the need of preventive maintenance. In view of the obtained results, the advantages and the limitations of the proposed approach as well as suggestions for further improvements are presented and discussed.     

基于神经网络的模拟电路故障诊断优化近似方法

本文提出了线性电阻电路故障诊断的一种神经网络优化方法。其基本思想是应用Ｈｏｐｆｉｅｌｄ网络原理来处理Ｌ１范数问题，以此来进行诊断。对实例的计算证明此法是可行的。