Feature Extraction of Acoustic Emission for Offshore Platform Structural Health Monitoring

摘要

Offshore platforms are the basis of offshore oil exploration and development. Living in harsh environment, fatigue damage and processing defects will gradually come out, which may lead to the platform collapse. Acoustic emission (AE) technology is potentially a promising method for crack detection in the offshore structures. The AE device could detect defects' changes timely and effectively with exquisite sensitivity and online real-time monitoring features. It has been successfully used in foreign offshore platform monitoring, but seldom in domestic ones. It is well known that AE is proficient in data acquisition and monitoring data changes, while not in signal processing. Signal processing plays an extremely important part in structural monitoring assessment. Without a good signal processing, the monitoring process is a failure to a certain extent. In this paper, we address an application of the Hilbert-Huang transform (HHT) and wave packet transform (WPT) to characterize the AE signals released from the offshore structure model. Without artificially determining the decomposition level, original signals are adaptively decomposed into several Intrinsic Mode Functions (IMFs) by HHT, while the wavelet (packet) transform could not. Filter out non-acoustic emission frequency band components for signal de-noising and reserve the larger IMFs by Fourier transform and cross-correlation analysis between each IMF and original signal. By HHT analysis, 3D joint time-frequency distribution and wavelet packet transform, the differences between structure normal signal and crack AE are clear and which highlight the AE characteristics. It can be concluded that the joint of HHT and wave packet transform is an effective tool to extract the features in offshore structures, which have a certain sense in the application of acoustic emission technique.