Research On Sparse-Representation Based Ultrasonic Composite Array Imaging For Pipeline Defect

Pipeline is one of the main energy transportation for oil and natural gas.Due to the long-time suffering of different environmental stresses,such as corrosion,pressure and temperature,various defects can build up in the pipeline which can lead to the leakage accident.This phenomenon will cause environmental pollution,economic losses and fire or explosion accidents.Therefore,the research of pipeline defect detection and imaging is of great significance to the quantitative evaluation of pipeline defects.Ultrasound imaging technique can provide intuitive and reliable visual judgment basis for quantification and evaluation of pipeline defects by constructing the acoustic image of pipeline internal structure based on acoustic characteristic parameters in ultrasonic echo signal.Due to the advantages of showing more abundant and intuitive quantitative information,3D ultrasound imaging technique has been gradually introduced into the field of pipeline defect inspection.However,the application of this technique in the visual inspection of pipeline defects is limited by the high computational complexity and low time efficiency of composing different 2D slice images into a 3D image.This thesis employs an ultrasonic transducer composite array to obtain the ultrasonic echo signals of pipeline defects.The sparse representation algorithm is used to directly extract 3D acoustic characteristic parameters from ultrasonic echo signal,which can improve the anti-interference performance of characteristic parameter extraction,and avoid the complex operation of composing 3D information based on different 2D images.Meanwhile,a self-adaption Gabor atomic dictionary construction method and a CSO(Chicken Swarm Optimization)based atomic search algorithm are proposed to improve the operational efficiency of characteristic parameter extraction.A 3D ultrasound imaging model of pipeline defect is established by projecting the 3D acoustic characteristic parameters into the pipeline space model,which is divided by sector voxels.It can achieve the 3D imaging and provide reference basis for quantifying and evaluating pipeline defects.The main research works of this thesis are as follows.(1)The pipeline defect inspection principle of composite array is analyzed.Based on the geometry of composite array and acoustic propagation property,the algorithm of solving the defect space location is given.A testing system of pipeline defect inspection based on ultrasonic transducer composite array is built.(2)Based on analyzing the basic theory and key techniques of signal sparse representation,the traditional Gabor atomic dictionary construction method and the specific process of ultrasonic echo signal sparse representation are researched based on ultrasonic echo signal characteristics.(3)A self-adaption Gabor atomic dictionary construction method is proposed aiming at the problems of large scale and low efficiency of traditional Gabor atomic dictionary.Simulation and real experiments show that this dictionary has a good anti-noise performance and can significantly improve the operational efficiency of sparse representation.The operational time is only about 19% of traditional Gabor atomic dictionary.(4)In order to overcome the disadvantages of the high complexity and time-consuming of the ergodic atom search algorithm,an atom search algorithm based on CSO is proposed.Experiments indicate that the operational time of this algorithm is only about 20% of the ergodic atom search algorithm under the condition of keeping the same precision as the latter algorithm.(5)The 3D acoustic characteristic parameter extraction method is researched using the sparse representation algorithm.Meanwhile,a 3D pipeline defect imaging model is established based on the 3D acoustic characteristic parameters.In order to verify the 3D imaging method,3D imaging experiments of crack and hole defects located at internal or external surfaces of the pipeline with different directions and sizes are carried out.Experimental results demonstrate this method can display the quantitative information of defects,such as location and size,which can provide intuitive judgment basis for the quantification and evaluation of pipeline defects.