基于小波概率神经网络的CMOS 电路 IDDT 诊断方法

摘要

近年来利用动态电流（IDDT ）测试研究 CMOS 电路故障的方法得到广泛关注。论文结合小波变换和概率神经网络方法，提出一种基于小波分析和神经网络的 IDDT 诊断方法。小波分析具有时频局部化特征，能有效提取突变信号特征，概率神经网络训练容易，收敛速度快，可以实现任意的非线性逼近，具有良好的分类效果。论文结合二者优点，实现对CMOS 电路故障的诊断，达到90％以上准确率。%In recent years ,IDDT is widely used in studying the fault of CMOS circuit .Combined with wavelet trans-form and probabilistic neural network ,an IDDT diagnosis method based on wavelet analysis and neural network is proposed . With the characteristic of time and frequency localization ,wavelet analysis can effectively extract the characteristic of muta-tion signal ,and makes probabilistic neural network training easier ,possesses fast convergence speed and good classification results ,and can realize any nonlinear approximation .Combine with the advantages of wavelet transform and neural network , the fault diagnosis of CMOS circuit is realized with more than 90% accuracy .