Research On Data Compression Method Of Pipeline Magnetic Flux Leakage Detection Based On Compressed Sensing

With the rapid development of pipeline MFL testing technology,the requirement for the detection accuracy of corrosion and other defect data is getting higher.It is necessary to increase the quantity of sensors in multiple varieties,ranging from single axial pipeline MFL detection to three-axis full-HD pipeline MFL detection,Especially for submarine pipeline detection,an increasingly long detection distance would definitely produce a huge data storage.In each detection process,the storage capacity of the inner detector of the pipeline is limited.Therefore,a reasonable approach to pipeline MFL data compression with a high compression rate,a fast compression processing speed and the retention of characteristic information of MFL data,is crucial to the development of pipeline MFL detection technology.The compression acquisition and data reconstruction process of pipeline MFL data are studied and analyzed in this thesis.It is mainly studied in the following aspects:Firstly,in view of the traditional MFL data compression process,the traditional industrial data compression methods and the performance evaluation index of compression are analyzed and compared from compression performance,and the shortcomings of the traditional methods are obtained.Several main compression methods are compared and analyzed.The main compression performance indexes are summarized and the applicability of these compression methods in the field of pipeline inspection is analyzed and evaluated.Secondly,in the process of sparse acquisition from MFL data,aiming at the influence of the number of observations and the sparsity of MFL data on the accuracy of reconstruction,the sparsity of MFL data is analyzed and the sparse acquisition performance of different observation matrices under different observation values is compared,and the most suitable observation matrix is obtained.The sparsity of MFL data is analyzed,and a suitable sparse transformation basis is selected for compressed sensing.The performance of sparse acquisition by different observation matrices is analyzed and compared,and the quantitative relation among reconstruction accuracy,data sparsity and observed value is analyzed and compared.Thirdly,in view of the contradiction between compression performance and data reconstruction performance,a MFL data compression method based on adaptive compressed sensing is designed from the characteristics of MFL data itself,which improves the compression quantity and guarantees the reconstruction performance.Based on the characteristics of MFL data,the importance of MFL data is categorized.Based on the results of the evaluation,the sparse data acquisition and data reconstruction are implemented to the corresponding importance.Finally,a pipeline MFL data compression method on basis of adaptive compressed sensing is developed,which improves the compression rate and guarantees the effect of data reconstruction.Relevant experimental analysis is carried out to verify the compression effect of the radial MFL data and the MFL data with noise.Fourthly,aiming at the process of MFL data reconstruction,a pipeline MFL data reconstruction method based on compressed sensing is proposed,which achieves better compression performance and higher reconstruction effect.The accuracy and time of these data reconstruction methods based on compressed sensing are analyzed and compared with experiments.These methods are applied to MFL data sparsely collected from observation matrix for data reconstruction,and the reconstructed results are compared and analyzed.In the end,the thesis is summarized and the further research is prospected.