Research on vibration fatigue damage identification of oil and gas pipeline under the condition of measured noise injection

Abstract

The acoustic emission signals collected by oil and gas pipelines on offshore platforms are disturbed by noise, which makes it difficult to locate the pipeline damage. To solve this problem, in this study, the measured noise signal of the offshore platform pipeline was first obtained through vibration noise acoustic emission tests. The measured field noise of the offshore platform was injected into a lead break signal collected in the laboratory, and then the injected noise signal was identified by the damage identification method based on empirical mode decomposition (EMD) combined with a probabilistic neural network (PNN). Finally, the method of time difference localization was used for preliminary localization, and then a backpropagation (BP) neural network was used to correct the localization results. The results showed that the recognition method based on EMD combined with the PNN network achieved a 97% recognition accuracy in cases with large amounts of measured noise interference. The fatigue crack localization method based on the BP neural network time difference correction method was used to locate the acoustic emission fatigue crack in the test data, and the average localization error after correction was less than 2%.