Research On Welding Seam Deviation Correction Technology Of Welding Robot Based On Vision Guidance

With the general application of welding robot in manufacturing field,automation and intelligent welding has become the development direction.At present,welding robots mainly conduct welding in the way of teaching and playback.Due to the influence of workpiece size,installation error,clearance change and thermal deformation,there may be a certain deviation between welding teaching track and actual track,thus affecting the welding quality of workpiece.In order to improve the intelligence and automation of welding robots,this paper studies the key technologies involved in the realization scheme of robot welding seam rectification guided by linear structured light vision.Aiming at the size and installation error of the workpiece and thermal deformation and other problems,the off-line deviation correction and online deviation correction are used to correct the welding teaching track,so as to improve the welding quality of the workpiece.Firstly,in view of the problems of interference and noise in the collected weld image caused by arc light,soot and spatter during the measurement process of the linear structured light vision sensor,the method of weld image processing and feature point extraction is studied.First identified the interested area of the weld image,a clear laser fringe image was obtained by means of median filtering method,maximum interclass variance method and morphology closed open operation processing.Then,the center line of the laser stripe in the weld image is extracted by the gray-scale centroid method,combined with the weld characteristics of four typical grooves,designed the slope analysis method and the straight line fitting intersection method to extract the weld feature point information of the corresponding groove.Secondly,aiming at the problems of low efficiency and poor accuracy of the traditional ICP welding seam track off-line rectification registration,an improved ICP off-line rectification registration method was proposed.The unit quaternion method was used for rough registration of weld detection track and teaching track,and then the improved ICP method was used for fine registration,so as to solve the problems of slow convergence rate and low search efficiency caused by a long search time for corresponding point pairs.The simulation results of straight weld and arc weld show that the improved method not only improves the registration accuracy,but also improves the registration efficiency.Then,aiming at the problem that the weld track is prone to thermal deformation in the process of online correction,an online correction method based on GA-ADRC is proposed.According to the principle of online deviation correction,an appropriate forward-looking data storage strategy and calculation method of deviation correction were designed.GA was used to optimize the relevant parameters of ADRC to realize the online correction of welding torch trajectory,so as to ensure that the welding torch can approach the welding seam quickly without overshooting.The effectiveness and stability of GA-ADRC online correction method are verified by simulation experiments on straight and arc welds.Finally,a robot welding seam correction system based on visual guidance is constructed,which realizes the functions of image acquisition,data processing and analysis,welding seam track off-line correction and online correction.The feasibility test of the visual-guided correction system was completed through the straight butt weld.The absolute error of correction was less than 0.35 mm,and the weld was formed evenly.The experimental results verified the effectiveness of the relevant methods in this paper.