Research On Welding Quality Inspection And Evaluation Technology Of HDPE Pipe Joint

High Density Polyethylene(HDPE)pipes are widely used in pipeline transportation systems such as natural gas transmission and urban water supply because of their light weight,corrosion resistance,wear resistance and long service life.In the fusion process of HDPE pipeline,it is susceptible to various defects at the joints due to the adverse effects of process,equipment,environment and other factors,which may lead to leakage or explosion accidents.With the rapid popularity of HDPE pipes,it is especially important to ensure the quality and safety of HDPE pipeline joints.Therefore,this paper thoroughly researches the HDPE pipeline fusion joint quality inspection system to realize the automatic detection and identification of defects in HDPE pipeline hot fusion joints,and the main research contents and conclusions are as follows:(1)The overall design of the HDPE pipe fusion quality inspection system was completed by integrating the defect information of HDPE pipe hot fusion joints and the development of fusion welding equipment,which was divided into visual inspection system for external defects and ultrasonic phased array inspection system for internal defects.The inspector uses the visual inspection technology to get the characteristic parameters of the external flap profile of the pipe to realize the detection of external defects of the pipe joint;the ultrasonic phased array inspection technology is used to get the D-sweep map of the internal defects of the pipe to realize the detection of internal defects of the pipe joint.(2)In ultrasonic phased array internal defect detection,an improved Faster-RCNN target detection algorithm model based on attention mechanism is proposed for ultrasonic phased array D-sweep HDPE pipe joint internal defect detection.For the problem that there are many small defects inside HDPE pipe hot melt joints and feature information is easily lost,the deep residual network(Res Net50)is integrated with the feature pyramid network(FPN)to form a feature extraction network,and the convolutional block attention module(CBAM)is introduced to adaptively refine the features.SSD model and Faster-RCNN model are used for training and testing,and VGG16,Res Net50,Res Net50+FPN,CBAM+Res Net50+FPN are used as feature extraction networks in turn for training and comparison of ultrasonic phased array HDPE pipe hot melt butt joint internal defect samples.The experimental results show that the Faster-RCNN algorithm model works better than the SSD algorithm model for the detection of internal ultrasonic phased-array defects in HDPE pipeline hot-fusion joints,and the accuracy(m AP)of the improved Faster-RCNN algorithm model for the detection of internal defects in HDPE pipeline joints is 90.5%,which verifies the effectiveness of the improved scheme.(3)In visual external defect detection,the Single-scale Retinex(SSR)algorithm is improved to solve the problem of uneven illumination in images.The gray-scale morphological filtering and the removal of image marker lines algorithm are used to avoid the interference of the external wall of the pipe and ensure that the external flap contour of HDPE pipe joints is not blunted,and the maximum inter-class variance adaptive threshold segmentation method and Canny edge detection algorithm are combined to obtain the external flap contour of the pipe,which ensures the integrity of the flap contour of HDPE pipes.(4)Based on the flap contour to obtain the feature parameters required for defect quality determination,four defect parameters,namely flap feature point size deviation,flap left and right feature point relative distance,pipe angle offset,and flap feature point size deviation,are calculated for quality determination of external defects of pipes.The experiments show that the visual inspection algorithm can achieve the quality determination of HDPE pipe external defects for HDPE pipe hot melt joints,which verifies the feasibility and reliability of the visual inspection algorithm.(5)Using QT to design and develop HDPE pipe fusion quality inspection system,the human-computer interface mainly includes three parts: image acquisition,internal defect detection,and external defect detection.