Research On The Acoustic Emission Leak Detection Technology Of Two Phase Flow Pipeline

As we all know,two-phase fluid pipeline system is common and important oil and gas transmission system in Petrochemical industry.In practical applications,due to the large number of pipelines and complicated and changeable structures,most of the environment have serious corrosion and harsh other characteristics,such as high pressure,serious corrosion,pipeline vibration.The change of of pattern of two-phase flow and the existence of pipeline leaks and other effects,will result in changes in process parameters and pipeline leakage accidents.It is well-known that pipeline leaks can lead to the destruction of the integrity of the entire transportation system,resulting in the leakage of transmission media,fire,explosion and even catastrophic accidents.Acoustic emission detection technology is capable of detectiong and monitoring leaks,and quickly identifing the state of leakage to take emergency response,and we can minimize the consequences of the accidents.This article provides a tool for the normal oil and gas pipeline technology and technology,Intelligent technical support.Based on the gas-liquid two-phase flow leakage experiment system,the leakage experiment was carried out with the advantages of adjustable system and readable flow rate.Using the basic principles of acoustic emission detection technology,the whole process of converting acoustic emission sources into acoustic emission signals is studied,and the leakage of pipelines with two-phase flow is used as the acoustic emission source to study characteristics of acoustic emission signals.Two-phase flow type and pipeline leakage parameters were used to analyze the factors of acoustic emissions from pipeline leakage in two-phase flow.Analyze and investigate the relationship between the acoustic emission signal parameters and the parameters of leak hole aperture,leak hole orientation,flow medium and other parameters.Wavelet packet decomposition,empirical mode decomposition and local average decomposition are respectively carried out on the original acoustic waveform signal of twophase flow pipe leakage collected in the experiment.Then the EMD signal is reconstructed,and the LMD signal is reconstructed,Then these signals is decomposed by wavelet packet decomposition to obtain the energy proportion of each frequency band,then,the difference between each leakage signal and the background noise signal is calculated respectively,so that the characteristic extraction method of the input data of the optimal machine learning is the local mean decomposition method.Based on the deep learning theory,BP neural network,Stacked Auto Encoders and Deep Belief Networks,these method are used to study the application of deep learning theory in the application of leakage recognition of two phase flow.Based on the feature parameters extracted by the method of local mean decomposition,the data is input as network,and the applicability and accuracy of several intelligent learning networks to leakage identification are studied respectively.