Research On Acoustic Emission Signal Detection Technology For Boiler Pipeline

Nowadays, industrial equipment such as boiler pipe has been widely applied to all walks of life, and plays an important role in different fields. However, with the rapid development of boiler pipe, the leakage and explosion accident is also growing. It brings the serious threat to our country and people’s life and property. Boiler pipe aging and corrosion are two main reasons causing leakage explosion accidents. Before the boiler pipe leakage, firstly the crack appears within the material, fracture part of the material will generate transient elastic waveforms, namely acoustic emission. Acoustic emission is a technology that can be used for nondestructive detection method for dynamic testing whether the boiler pipe cracks. Analysis and processing the signal of acoustic emission we detected, a large amount of early crack information can be obtained. But the scene the detected acoustic emission signals are often accompanied by industrial complex background noises, it is difficult to identify and extract the acoustic emission signal, therefore, research on a reliable acoustic emission technology has important significance.Firstly, according to the characteristics of acoustic emission signals, In this paper, CS theory and EMD method is applied to the acoustic emission signal processing, from two aspects of data compression and decomposition to improve detection technique of simulated acoustic emission signal. Then, this paper attempts to apply approximate entropy; recurrence plot and recurrence quantification analysis method to acoustic emission signals processing, analysis of acoustic emission signal feature extraction, to provide more useful information to accurately identify acoustic emission signals. Finally, using the support vector machine for the extraction of the acoustic emission signals classification.In this paper, the experimental results show that the compressed sensing signal processing method is applied to detection of acoustic emission signals, it can save data transfer and storage costs; by contrast signal EMD, EEMD and mask signal decomposition method, verify the mask signal method can effectively improve the acoustic emission signal decomposition process exists modal aliasing; the EMD and approximate entropy algorithm is introduced to detect acoustic emission signals can effectively extract the characteristic parameters of acoustic emission signals using SVM to classify; with RQA value as the characteristic parameter input SVM classifier can get higher classification accuracy.