Application of frequency band energy decomposition using wavelet packet in extracting sucker rod defect characters

Sucker rod is the hey component for the rod pumping system. In the working process, there are mainly four kind of defects: cracks, corrosion, partial wear and artificial damage, which will cause downhole accidents, so feature extraction is a premise to distinguish different defects. In the paper, by using wavelet energy spectrum, sucker rod defects signals that came from magnetic flux leakage (MFL) detection method are decomposed to the different frequency bands by arbitrary time-frequency resolution, the frequency signal energy is an eigenvectors to judgment sucker rod defects. In practice, the detection method is effective and feasible, and there is a new way for the sucker rod defects pattern recognition.