Design Of Visual Positioning System For Welding Of Structural Stiffeners

In the field of welding,the automatic welding of structural stiffeners is a common type of engineering application.Structural stiffeners are such a type of component,which is designed to strengthen the bearing capacity of the structural surface in structural design.This type of component tends to have a little deformation and a large number of splash interference points when it is fixed in the early stage,In the past,this type of components was welded manually,which had high repeatability,high labor intensity,and a certain degree of danger.In the automatic welding stage,most of the weld positioning methods are based on two-dimensional images,which used geometric methods to extract the welds.The welds extracted by this method lack applicability to structural stiffeners,and the overall positioning accuracy still has some room for improvement.Therefore,we design and construct a structural stiffeners weld positioning system,where 3D vision sensor.is used,the specific research contents are as follows.Firstly,we design the visual positioning system for structural stiffeners welds.Then analysis the basic components of the visual positioning system and different types of structural stiffeners.after that,the positioning process of the entire system is determined.According to the actual welding requirements of structural stiffeners,three indicators of the measurement system are designed.The three indicators are Stability,accuracy and speed,among them,the accuracy indicator requires that the hand-eye calibration error is plus or minus 0.5 mm,the final positioning error of the weld is plus or minus 2 mm,and the rapidity requires that the execution time of the algorithms are less than two seconds.Secondly,we design the pivotal hand-eye calibration algorithm in the positioning system.In view of the actual welding requirements,we determine the installation method of the vision sensor,after analysis the two coordinate transformations involved in the positioning system,we derive the hand-eye calibration equation,where we use the calibration coordinate axis registration method to achieve the equation parameters.The hand-eye conversion matrix is obtained by experiment combined with the mark points on the three-dimensional calibration board,which is designed for hand-eye calibration error analysis.Finally,the experiments result shows that the average error is 0.52 mm,which meets the requirement of accuracy indicator.Thirdly,we study the principle of thin-plate spline interpolation deformation registration,and analyze its mathematical model,then we introduce soft registration technology and deterministic annealing optimization to achieve the registration results,which means we can transfer the target structure stiffener to the template structure stiffener.By marking the control points in sequence on the contour of the template stiffener made offline,we determine the welding sequence of various structural.Stiffeners.Then we use registration to obtain the control points corresponding to the contour of the target structure stiffener.For the control points on the target contour,cubic spline interpolation is used to fit the continuous weld under the coordinates of the vision sensor.The experimental results show that the target structure stiffener can be better registered to the template stiffener,stable control points can be obtained,and the fitted welds are continuous and smooth.Fourthly,we design the software module of the system.In the light of the operating demands in the actual welding process,the functional modules are designed,which mainly contain communication module,camera module,weld positioning processing module,calibration module,display module and monitoring module.In terms of touch screen interface design,the touch screen editing software is used to build interactive functions that mainly include optional operation of welding types,real-time display of welding trajectories,robot reset control,and start-stop functional buttons.The designs mentioned above increase the interactivity of the system and the traceability of data during the welding process.The welding project is built on the Visual Studio 2015 with C++,where relevant algorithms are complied.Combined with the actual welding experiment,the three indexes of the system are analyzed and evaluated.The experimental results show that the execution of the system algorithm takes about 1.7s,and the final positioning accuracy is within 2mm,which meets the designed indexes of the system.