| Polyethylene pipes have good corrosion resistance and toughness,and have been widely used in nuclear power and oil and gas transportation fields.Butt fusion(BF)welding is a commonly used welding method for polyethylene pipes.The welding quality is high and the cost is low.In the BF welding process,welding parameters,personnel operation and environment will affect the welding condition of polyethylene pipes,which may cause defects in BF joints and affect the safe operation of pipeline.Safety inspections of BF joints need to be carried out with the help of phased array ultrasonic testing(PAUT).PAUT of BF joints on polyethylene pipes adopts fan-shaped scanning for detection,and the sound pressure of the fan-shaped scanning beam deflection decreases with the increase of the scanning angle,resulting in shape distortion of defect detected image,thus it is difficult to identify the defect.In this paper,the characterization and reconstruction methods of defects in PAUT for BF joints of polyethylene pipes are studied through nondestructive testing of polyethylene pipe BF joint,numerical simulation of defect response and statistic analysis of defect simulation data.The main works are as follows,(1)Considering that the crimping structure at the fusion surface of BF joint restricts the detection position of the phased array probe,PAUT process of BF joint is formulated,the types of defects in BF joint are classified and the causes of the defects are analyzed.Use this inspection process to conduct on-site inspection of gas pipelines in a certain city,obtain joint types with frequent defects and common defect types,then determine the research object of defect simulation.(2)Based on the CIVA simulation software,the sound field simulation and hole defect simulation of DN315 pipeline BF joints by PAUT are realized.The corresponding relationship between the defect simulation results and the actual defect characteristic parameters was established,and the characterization method of the defect quantitative and location in the BF joint inspection results was proposed.We compared the characterization effects of different characterization methods based on the defect database.In the defect quantification,the law of F/B value and defect size is analyzed,the influence of buried depth is explored.In defect location,the law of the calculated value of axial distance,the calculated value of buried depth and the actual value of buried depth is analyzed,the influence of defect size is explored.(3)Based on the F/B values,axial distance values and burial depth values in the defect characterization results,the relationship between the actual defect characteristic parameters and the defect characterization results is obtained through the matrix scatter diagram.Two regression models are used for defect reconstruction,and the influence of regression model on polynomial regression,and the influence of kernel function,kernel function coefficient and penalty coefficient on support vector regression are discussed.Considering the error and the degree of fitting comprehensively,the polynomial regression equation and the support vector regression model of the defect size and position difference are determined.On this basis,the polynomial regression equation and the support vector regression model are simply weighted into a combined regression model with a weight of 50%,thereby forming a defect reconstruction method. |