A fault diagnosis method for analog circuits based on wavelet packets is developed in this paper. The sampled signals from the test nodes are decomposed by wavelet packets and the feature vectors extracted are applied to neural networks for identifying the faults. The proposed method is characterized by minimizing the ambiguity groups as well as simplifying neural network architecture, reducing training cost and maximizing the diagnosability. Simulation results of diagnosing a four-op-amp filter circuit have confirmed the validity of the propose technique.
|Title of host publication||Procs of the 2004 IEEE Region 10 Conference, TENCON|
|Publisher||Institute of Electrical and Electronics Engineers (IEEE)|
|Publication status||Published - 2004|