Nonlinear Dimension Reduction on Analog Circuit Fault Diagnosis Using L-Isomap

Yue-hai WANG, Yu-ying MA, Shi-ming CUI, Yong-zheng YAN, Xiong LI


Due to the nonlinear characteristics of analog output signal, the extracted fault feature vectors tend to be high-dimensional and general need to reduce dimension. Focus on this problem, this paper puts forward a feature reduction method for analog circuit fault diagnosis based on wavelet packet energy spectrum and L-Isometric feature mapping (L-Isomap). In our approach, L-Isomap is used to reduce the high-dimensional fault vector of analog circuit and extract the optimized low-dimensional features as the characteristic vector. Compared with Isomap, L-Isomap has better dimension reduction effect and lower time complexity. Several experiments running on a two-stage four-op-amp biquad low-pass filter circuit, and the experiment results validate the effectiveness of our proposed method.


Analog circuit fault diagnosis, L-Isomap, Wavelet packet energy spectrum


Full Text:



  • There are currently no refbacks.