Research On The Localization And Identification Of Defects Of Deep Sea Oil And Gas Pipeline

Due to the accidental impact or corrosive medium in the tube,the metal loss defects appear on the wall of the pipe during the process of pipeline transportation,installation and operation,which is a serious threat to the safety of oil and gas pipelines.How to eliminate this potential hazard and ensure the safety of pipeline transportation is the most important and urgent problem to be solved.In particular,deep sea oil and gas pipelines are more sensitive to defects,because they are under high presure and their inside and outside temperature are quite different,requiring higher requirements for the detection of pipeline defects.In view of this,a series of research is carried out on magnetic flux leakage detection of pipeline defects and localization and quantitative identification of defects in this paper.The reconstruction of pipeline defects requires a lot of magnetic flux leakage testing data.Due to the limitation of the measured data,previous researchers use finite element method to simulate the inside magnetic flux leakage detection of pipeline.On the basis of understanding the principle of electromagnetic field finite element analysis,a three-dimensional finite element of the pipeline and the testing device is established and magnetic flux leakage detection of pipeline is simulated,whose prototype is Liwan large thick wall offshore pipeline in the South China Sea.By extracting the magnetic flux density,the spatial distribution of the magnetic field in the pipeline and the relationship between the pipeline defects and the magnetic field is obtained.Through the finite element simulation,the inside magnetic flux leakage detection data is obtained.By analyzing the data,it is found that the magnetic flux leakage detection signal curve changes suddenly on the position of defect,which is helpful to identify the position of defects on pipelines.In addition,it is also found that the magnetic flux leakage detection signal curve changes corresponding to the change of the defect size,indicating that there is certain relationship between the two above.In order to reconstruct the size of pipeline defects,a large number of finite element models are established in this paper.By extracting the characteristic value of the magnetic flux leakage signal,the training samples and verification samples for network reconfiguration are established.By analyzing the influence of the characteristic quantity,sample number and SPREAD value on the reconstruction accuracy of RBF neural network,the optimal value or optimal range of the three above are determined,and the RBF neural network is trained.Finally,the trained network is tested by the verification samples and the results show that the precision of the reconstruction of pipeline defects is more than 98%,indicating that the goal of quantitative identification of pipeline defects is achieved.