Feature Extraction of the Small Leakage Diagnosis of Oil Pipeline Based on Acoustic Signal

Abstract

This paper introduces the related theory of sound waves , we collect the data through the acoustic wave sensors of the pipeline fault diagnosis system platform, decompose the signals to five layers by Mallat algorithm and wavelet function db4, compare the normal waves and leakage acoustic signal spectrum, and then get the power spectrum estimation for the decomposed signal at each level, we can see the signals energy feature in different frequency band. Feature extraction method based on wavelet transform can make the category of signal characteristics fully displayed in the different resolution band, it has a good application prospect in the field of acoustic signal processing.