摘要
针对模拟参数型故障的测试难题,提出了一种基于小波滤波器组的模拟VLSI测试方法.将测试响应经小波变换实现的数字滤波器组完成子带滤波;然后对各子带滤波后的序列计算相关系数、相关函数、协方差函数、相干函数以及功率谱;从以上5个方面完成模拟响应的多个特征提取,解决了采用单一特征量可能出现的正常与故障下特征差异不明显或混叠的问题.采用正常与故障电路差异明显的特征完成故障诊断.实验结果表明:基于文中方法,采用金字塔结构的Haar小波滤波器组对模拟元件参数型故障的测试效率较高.
Aiming at the difficulties of testing parametric faults, an approach based on wavelet filter bank is presented. Subband filtering to test response signal is achieved by virtue of wavelet filter bank. The computation of correlation coefficient, correlation function, covariance function, coherence function and power spectrum is followed on the subband filtered. The signature extraction to analog response signal is carried out based on the above five operations. As a result, the difficulty to distinguish fault-free and faulty circuit by a single signature can be solved effectively. Signature with clearly apparent difference between normal and faulty circuit is chosen to diagnose faults. Experimental results show that our approach based on pyramid type Haar wavelet filter bank is of high effectiveness in testing parametric variation of analog component.
出处
《计算机辅助设计与图形学学报》
EI
CSCD
北大核心
2007年第11期1459-1464,共6页
Journal of Computer-Aided Design & Computer Graphics
基金
国家自然科学基金(90407007)
关键词
模拟电路测试
故障诊断
小波滤波器组
特征提取
testing of analog circuits
fault diagnosis
wavelet filter bank
signature extraction
