摘要
介绍了基于BP神经网络(BPNN)进行模拟电路故障诊断定位的理论,设计并实现以数字信号处理器(DSP)为核心的故障诊断系统.该系统采用模块化设计,具有扩充方便、高速采样的特点;将采样数据构成故障特征向量,利用BP网络训练这些特征向量并进行故障模式分类,实现模拟电路及PCB故障诊断.并给出了PCB板的故障诊断硬件结构及对单软和双软故障的诊断设计方法.
The theory for identification and diagnosis of faults in analog circuits based on Back-Propagation Neural Network (BPNN) was introduced. The hardware system for fault-diagnosis with digital signal processor (DSP) was designed and implemented. This system was designed with modularized construction, convenient enlargement and high-speed sampling. The eigenvector was constituted from sampled data and trained by BP network. Then the output of BPNN will indie.ate the classification of faults in analog circuits and PCB. At last, the illustrative example of hardware structure of the fault-diagnosis of PCB with single soft and double soft faults was given.
出处
《微电子学与计算机》
CSCD
北大核心
2008年第6期53-56,共4页
Microelectronics & Computer
基金
国家自然科学基金项目(50677014)
国家"八六三"计划项目(20060104A 1127)
教育部新世纪优秀人才支持计划(NCET-04-0767)
教育部高等学校博士学科点专项基金(20060532002)
湖南省自然科学基金项目(06JJ2024)