| The quality of welding on the tube plate in large high-pressure vessels in nuclear power plants directly affects the heat exchange efficiency of the equipment,which in turn affects the overall system operating efficiency and significantly shortens the lifespan of the equipment.Although commonly used semi-automated and specialized welding equipment for tube plate welding has been adopted,when welding tens of thousands of tubes onto the tube plate,these equipment face many problems such as shallow weld pool,long cycle time,high labor intensity,and low yield.Therefore,to establish a high-quality and efficient fully automated tube plate welding system,research and innovation are needed to improve the quality and efficiency of tube plate welding.This project proposes a machine vision-based fully automated tube plate welding method,and provides a design scheme and key technologies.The main research work is as follows:First,the design requirements of the fully automated tube plate welding platform and the robot body are introduced.The laser sensor and visual hardware selection for the welding platform are detailed,using the Big Dipper Mercury series MER-2000-5G M/C-P industrial camera with Gig E interface,the M0814-MP2 lens produced by Comp utar,and a linear infrared laser generator to complete the construction of the fully aut omated welding robot platform.Analyze the quality requirements of obtaining the image of annular weld of tube plate,and analyze the calibration process of laser vision platform built by this topic.Steger algorithm was used to extract the center line of laser fringe based on Hessian matrix.According to the calibration principles of welding robot tool coordinate system,industrial camera calibration,distortion correction,laser plane calibration and hand-eye calibration,the corresponding calibration experiment flow is designed respectively.Based on the characteristics and quality of the collected images of the torus weld,the image preprocessing method and laser center line extraction process are analyzed in detail,and a scheme most suitable for this topic is selected.Finally,the pixel coor dinates of the detected torus weld image feature points are converted to the base coor dinate system of industrial welding robot through hand-eye calibration matrix.Experiments are conducted on industrial camera calibration,laser plane fitting,han d-eye calibration,and circular welding seam image recognition,and precision error ana lysis is performed.The experimental results show that the maximum error of industrial camera calibration does not exceed 0.0003 mm,the maximum error of laser plane cali bration is 0.052 mm,the average error of hand-eye calibration between the industrial c amera and welding gun is 0.1mm,and the average error of tube plate image recogniti on is 0.0204 mm.Finally,the accuracy of the fully automated TIG tube plate welding designed in this project is 0.17 mm,which meets the engineering design requirements. |
PDF Full Text Request