摘要
大规模集成电路的发展使得传统的接触法测试在某些场合受到了限制.针对其造成的电量测试信息不足的问题,文中融合电量信息和非电量信息作为故障特征信息,应用自组织特征映射(SOFM)神经网络对模拟电路进行故障诊断.提取电路工作时的电压和温度信息作为故障特征信息,经预处理后作为样本输入给SOFM神经网络进行电路故障诊断.通过输出层各神经元的竞争,得到获胜神经元,从而对样本数据进行故障识别分类.仿真结果表明,应用所提融合诊断方法提高了诊断准确率.
With the development of VLSI, conventional contact test is constrained in some occasions. Aiming at insufficient information of analog circuit electronic test, electronic and non--electronic information is combined as fault feature signature in the paper. Self--Organizing Feature Map （SOFM） Neural network is applied to analog circuit fault diagnosis. Voltage and temperature information is extracted as fault feature information, then processed, which is sent into neural network as input sample to detect circuit fault. Using the competition of output level neurons of SOFM, the winning neuron is attained and the fault recognition for sample data is classified. Simulation results show that fusion diagnosis has more satisfactory accuracy compared with single information diagnosis.
出处
《微电子学与计算机》
CSCD
北大核心
2011年第11期160-164,共5页
Microelectronics & Computer
基金
国家自然科学基金项目(60973032
60673084)
湖南省自然科学基金重点项目(10JJ2045)
