Feature extraction of singular signal based on adaptive wavelet

Aimed at improving the accuracy of wavelet analysis which was used to detect the singularity of the signal, a new method for wavelet basis selection was proposed according to the characteristics of the signal: The regularity of the signal was expressed by the amplitude differential. Then the corresponding wavelet group was determined as candidates. The most suitable basis was obtained by comparing the similarity between the detected signal and different wavelet bases. The simulation experiments show that the proposed method is effective when applied to singularity detections of both slow-change signal and sharp-change signal. Finally, the study result is applied to detect the signal of photoelectrical encoder.