| At present,the domestic welding in the aerospace fine parts,often uses manual welding.Due to human factors,results in low production efficiency,long cycle,personal factors,difficult stabilized quality,high repairing rate.And some companies have begun to use the robot for welding.However,basically there is no real-time adjustment of welding parameters and trajectory of the "teaching-to reproduce" type of robot.As the fixed welding parameters being set before the start of robot welding,through manual teaching and fixed robot welding path,in the face of complex welding situation,there will be a lot of interference with the robot to achieve automatic welding,Causing the robot unable to continue welding.For example,in the welding process the workpiece heat deformation,assembly error,weld joint thickness difference,teaching accuracy deviation and other reasons,will cause the weld deformation or gap size changes.At this time due to the fixed welding trajectory of the robot can not follow the weld offset in time,fixed welding parameters can not be real-time changes in welding conditions to be modified,that will lead to welding bias,welding or not penetration and so on.In order to solve the above problems of artificial or robot welding,this paper has developed a set of weld quality control system with complex sensing with vision,arc and sound to realize weld tracking and forming quality control of the robot GTAW process,and applied it to the aerospace launch vehicle on the five-way connector flange robot automatic welding.To achieve real-time tracking and forming control of the weld,the first need is to obtain real-time access to the relevant information in the welding process.According to the arc spectrum analysis of aluminum alloy,through the selection of light attenuator and filter,the current and the time of image acquisition experiment,the successful way of the welding process in real time to collect pool image is established.Aiming at the characteristics of aluminum alloy pulsed GTAW pool image,a method of image processing based on image degradation,median filter,canny operator,edge detection and weld inspection was developed.A visual attention theory was used to develop a method of extracting ROI from the Itti visual model of the down-up,which can automatically search the small window of the image processing,that can solve the fault tolerance of manually selecting the small window.So that the image processing can quickly,accurately and effectively lacate the ROI small window,for the follow-up of the weld tracking and forming quality control provides a solid foundation.Based on the arc voltage sensing system designed in this paper,stable arc voltage signal can be collected.And the differential threshold + mean threshold + wavelet Wden function is used to denoise the signal.The interference signal in the arc voltage signal is successfully eliminated.The relationship between the arc and voltage of the aluminum alloy pulse GTAW is tested as: V = 3.06 h + 7.32,and the accuracy of the model is verified at 0.27 mm,and the accuracy can be well satisfied with the requirements of the arc length in welding process.The acoustic signal of the three different penetration conditions is studied by time domain analysis.It is found that the sound signal characteristic in the frequency range of 5.5-9.5kHz has a highest rate of the penetration recognition,that can be chosen to be penetration characteristic.Based on the rough set,the prediction model of the weld backside width of the aluminum alloy pulsed GTAW is established.Based on the information such as current,voltage,wire feed speed,image information and sound information of the molten pool,the rough set multi-information fusion model is established to generate the forecast decision Table,the aluminum alloy pulse GTAW weld back width were accurately predicted for the subsequent weld quality control to provide a stable and effective technical basis.Based on the fuzzy controller based on genetic algorithm,the aluminum alloy pulse GTAW weld is controlled in real time.The weld width of the weld is controlled at about 7mm and the error is less than 0.4mm.The welding quality can meet the requirement of welding quality.Based on the visual sensor,a segmented PID tracking controller is designed to track the plane weld of the aluminum alloy pulse GTAW.The control system successfully tracks the straight line and the curve weld,and the control precision is within 0.3mm.Accuracy of real-time plane tracking for welding.Based on the arc sensing,a segmented PID tracking controller is designed to test the arc length of the GTAW process.It can accurately control the height of the workpiece for the welding torch,that is,the arc length and the control precision within 0.3mm.The accuracy of the welding real-time height tracking.The reliability and stability of the rough set multi-information fusion prediction model and the fuzzy control system based on genetic algorithm is testified through welding experiments of the flat welding flange.Based on the research of this paper,the reliability and stability of the rough set multi-information fusion prediction model and the fuzzy control system based on genetic algorithm are proved,which has a strong applicability,and provides a solid technical support for aerospace automation welding development. |
PDF Full Text Request