论文部分内容阅读
动态电源电流测试对模拟集成电路故障诊断十分有效。采用斜坡电压源代替传统的恒定直流电压,从电源电流波形采样训练神经网络的数据并建立故障字典。利用小波变换具有同时在时-频域分析信号、大量压缩数据的属性,对采样信号进行小波包分解,提取故障特征来训练神经网络,简化了网络结构、提高了训练速度。实验结果表明,该方法能够实现快速故障检测及定位,具有准确率高的特点。
Dynamic power supply current test analog integrated circuit fault diagnosis is very effective. Ramp voltage source instead of the traditional constant DC voltage from the current waveform samples training neural network data and the establishment of fault dictionary. The wavelet transform is used to analyze the signals in the time-frequency domain at the same time, and the properties of the data are greatly compressed. The sampled signals are decomposed by wavelet packet, the fault features are extracted to train the neural network, which simplifies the network structure and improves the training speed. Experimental results show that this method can detect and locate fast faults with high accuracy.