Research On Magnetic Flux Leakage Detection System For Inner Wall Of Natural Gas Pipeline

Pipeline transportation plays a very important role in natural gas transportation.In order to ensure the long-term safe operation of pipelines,they must be regularly inspected and maintained.It is very difficult to detect internal defects of natural gas pipelines while they're in operation.If they're shutdown for detection,major economic losses will be caused.Based on the principle of magnetic flux leakage detection,this paper studies the inner wall detection technology of natural gas pipelines and develops the inner wall detection system for natural gas pipelines.Based on the analysis of the formation mechanism of leakage magnetic field,this paper conducts theoretical studies and analysis of the calculation of leakage magnetic field and the selection of magnetization.The theoretical analysis model of the relevant magnetic flux leakage detection system was built by ANSYS finite element analysis software.By analyzing the influence rules of different defects on the leakage magnetic field of the pipeline,it provides a theoretical basis for the later fault identification.After analyzing the simulation results,the inner wall defect detection system for natural gas pipelines was designed.Under the laboratory conditions,the experimental device of the pipeline inner wall detection system was built,and three types of defects,namely circumferential,axial and oblique,were artificially fabricated.The experimental device is used to collect,process and analyze the experimental data,and the obtained leakage magnetic data of pipeline defects is made into an image data set of defect magnetic flux curves.The deep convolutional neural network model was built by using the deep learning framework Tensorflow,and the training of the deep convolutional neural network model was completed by using the training set of the image data set of defect magnetic flux curves.In this paper,the experimental device of the pipeline inner wall detection system is built,and the leakage magnetic data of the inner wall of the pipeline is collected.The validity and reliability of the magnetic flux leakage detection signal are verified by the experimental data graph,which verifies the rationality of the parameter design of the inner wall detection system for natural gas pipelines.It provides meaningful theoretical support and practical theory for the application of the inner wall detection system for natural gas pipelines.Experiments show that by using the improved deep convolutional neural network model to predict the image data of the new pipeline defect magnetic flux curves,the average recognition rate can reach more than 93.25%.It provides a new method for the rapid and batch identification of pipeline defects.