Research And Design Of Automatic Laser Welding System Based On Machine Vision

Welding technology has a wide range of applications in both the metal manufacturing and processing industries and other industrial processing fields.In recent years,with the rapid development of laser technology and automation technology,a new type of automatic welding method using laser beams as a heat source can meet the needs of high-standard requirements in cutting-edge manufacturing,the aerospace and military manufacturing fields,while automated and intelligent welding technology also shows great development potential.Combining with machine vision technology is an effective way to improve the automation and intelligence of automatic welding systems,which has great industrialization potential and broad application prospects in the field of automatic production and processing.This paper mainly studies the platform design of automatic laser welding system,weld identification and smooth planning method of center line path of weld trajectory.In terms of the design of automatic laser welding system platform,after the preliminary market research on the automatic laser processing equipment developed and produced by leading domestic and foreign laser processing equipment manufacturers,thispaperdesigned an automatic laser welding system experimental platform based on machine visionindependently.In this paper,three parts are designed and assembled to build an experimental platform for automatic laser welding system based on machine vision.Frist of all,this paper designed industrial cameras,industrial lenses and lighting systems,as the visual sensor for guiding the motion execution system.The second,designed the motion control by using the three-axis linear module with ball screw structure as the motion actuator.The motion control card and drive are designed as the motion control drive part.The last,selected and matched the laser and the laser welding headas the welding heat source and the laser beam shaping mechanism in order to design thewelding execution system.Weld identification includes two parts: weld track identification and center line extraction of weld track.In view of the narrow weld gap and the similar characteristics of welds' surface scratches and welds,which can easily lead to misidentification,this paper first completed the research of camera calibration model and calibration method,and applied Zhang's calibration method to carry out camera calibration experiments to eliminate lens distortion.Then,according to the characteristics of the plane curve weld,the edge detection of the welding object is carried out by the method of directional wavelet transform,and the interference factors such as scratches and rust on the surface of the welding object are filtered out,and the edge of the welding object and the weld image are extracted.Finally,the extracted edges of the welding object and the weld image are further processed,the morphological processing method is used to extract the binary image of the weld track,and the morphological refinement method is used to extract the center line coordinates of the weld track.The reliability of the weld trajectory centerline extraction method was verified by experiments.In the aspect of the smooth planning method of the center line of weld path,the weld path curve fitting method is studied.Firstly,in view of the large amount of data in the fitting operation of complex weld trajectory curve,the Harris-Laplace corner detection algorithm was adopted to segment the complex curve.The idea of piecewise fitting was adopted to reduce the complexity of the curve without losing the curve data.Secondly,in view of the phenomenon of stutter and jitter when the motion actuator moves directly according to the coordinate of the center line of the weld trajectory as the coordinate of the type value point,the NURBS curve fitting method is applied to plan the welding trajectory curve.Based on Matlab development environment,this paper completes the algorithm verification of weld identification and smooth path planning of the center line of weld trajectory.Based on Visual Studio 2010 software development platform,C# language is used to develop welding control software.Through the laser welding experiment on metal workpiece,the design of automatic laser welding system is completed.