Autonomous Programming Welding And Parameters Adaptive Control Based On 3D Reconstruction Of Seam

In recent years,with the rapid development of sensing technology,robotics,the Internet of Things and artificial intelligence,the intelligence of manufacturing has become an inevitable trend,and welding is an important part of it,so its intelligence is also imperative.However,the currently widely used welding methods of teaching-payback and offline programming are inefficient,poorly adaptable and hard to know the dynamic change of the groove size,thus difficult to obtain excellent welding quality.Therefore,the intelligence of robot welding needs to be improved.An autonomously programmed robotic welding method based on laser scanning is proposed,realizing the welding process for complex curve seam.In order to improve the recognition accuracy of 3D point cloud,the conversion process of multi-polyline scan data is optimized by time-distance interpolation method.Then,based on the high-featured image conversion and targeted image processing algorithms,effectively filter out interference and extract 3D weld feature points.And the 3D spatial model of seam is reconstructed through the cubic smooth spline algorithm,so it is smoother and closer to the real situation.Based on the model,effective welding trajectory planning is performed: the data compression of the welding path through the Douglas-Peucker(DP)algorithm reduced the subsequent computational complexity;the welding torch posture is calculated based on the bevel geometry model,and unusual fluctuations of posture on the welding path are reduced by the smooth interpolation algorithm.In addition,through a series of welding experimental data of controlled variables and Back-Propagation(BP)artificial neural network algorithm,this paper establishes a model of the corresponding relationship between lap gap,welding current and welding voltage,which can be used to predict process parameters based on the gap.And it combined with the polyline fitting method to insert new path points,achieving adaptive control of process parameters in robot welding tasks.Based on the pose and process parameter information on the welding trajectory,the final robot welding task is autonomously generated and executed.Besides,different plane curve workpieces with randomly distributed lap gap were welded by the proposed method,and the trajectory accuracy basically meets the requirement of arc welding.And the uniform and full weld formation results verify the feasibility,effectiveness and compatibility of this new method of autonomous welding.