Fatigue crack identification method for oil and gas pipelines on offshore platforms

Abstract

In studying the practical application of acoustic emission technology to the monitoring of pipeline cracks in oil and gas platforms, the problems of pipeline vibration interference and effective feature extraction of acoustic emission signals of fatigue crack need to be solved. The use of fatigue crack identification based on wavelet packet as a feature extraction method is proposed based on a probabilistic neural network to address the above problems. First, wavelet time-frequency analysis is used to determine the feature frequency range of acoustic emission signals of steel structure fatigue cracks. Second, only the features of the reconstructed acoustic emission signals that contain the feature frequency are extracted. Finally, fatigue crack recognition is performed through the probabilistic neural network. Test results show that the proposed method can identify cracks in oil and gas platform pipelines, and has a certain anti-interference ability, thus providing a test basis for the subsequent testing and practical application of offshore oil and gas platform pipelines.