Eddy Current Detecting of Leak Hole in Pipeline by Wavelet Packet Signal Analysis Method

Abstract

In the process of detecting leak hole in pipeline by intelligent pig with eddy current measuring method, the cut off effect of eddy current field in the closed cross section of leak hole in steel plate is weak due to geometric deformation of eddy current on the edge of the hole produced by destroyed drilling. As a result, the measured signal is very unobvious. Meanwhile, owing to the rough surface of steel plate, the periodic interference generated by movement of the detecting probe cannot be eliminated or be inhibited easily by conventional signal processing methods for its large amplitude and approximate frequency band with the leak hole signals, the signal to noise ratio (SNR) of measured signal is very low and the accurate identification of leak hole cannot be guaranteed. The wavelet transform, with characteristics of time-frequency localization and multiple scales, is a useful and effective method for identifying singularity of the signals and adapts to detect the transient signal or extract non-stationary information in the signals with strong periodic interference and noise. The reconstructing signal SNR will be increased greatly in eddy current detecting of leak hole in the pipeline with wavelet packet analysis of the signal by constructing the self-defined cost function based on maximum SNR to obtain optimal wavelet packet basis function. This ensures good detection and location of leak hole in the pipeline.