Research On Real-time Tracking And Control Technology Of Three-dimension Welding Seam During Welding Robot GTAW Process Based On Vision Sensor And Arc Sensor

Nowadays, robot welding is a trend of welding automation technicaldevelopment; more and more welding robots have been applied in theautomatic manufacturing process at home and abroad. However, most ofthem are primary teaching-playback robots, the welding path and processparameters of which are set in advance, and the requirement of conformanceto the welding operation conditions is very strict. That means if thework-pieces of weld are changed, they must be taught and programmed again.Especially, the seam position is often disturbed by distortion, ways ofspreading heat, variability of gap, stagger edge, etc. which will affect thequality of welding forming. In the practical welding application, this type ofwelding robot cannot meet the enterprises’ requirements on superior qualityand top efficiency because of the lackness of adaptability in teaching trajectory offset and no function of real-time tracking control, which limits itsapplication in many fields. In order to overcome the uncertainty which affectsthe quality of welding, there is an urgent need to track the seam in real timefor “teaching-playback” welding robot, which will make it more flexible andintelligent.In the practical welding process, most of the welding seams arethree-dimensional, which are difficult for the only vision technology to trackand control in real-time. However, among what have been studied in weldtracking, the passive vision technology can accurately get the2Dseam-tracking information, but the height information should be acquiredonly through3D reconstruction method. It is a pity that most of thealgorithms are too complex to be used in real-time control, which causes thatthe vision technology can hardly meet the requirements of real-time heighttrack and control during actual robotic welding process. Thus, anothertechnology must be used in height tracking while the visual sensortechnology is used in real-time3D seam tracking. The composite sensorsbased on visual sensor and arc sensor can be used in the weld of3D seam,which provides a possibility to realize the function of real-time tracking of3D welding seam during robotic GTAW process, and will develop theteaching-playback robot into an intelligent welding robot. No doubt that theseresearches supply important theoretical significance and practical value forexpanding the applied range of robot welding automation technology in manydifferent fields, such as national defense, aviation, shipbuilding, machinery,automobile, electronics and electrics.There are three problems which should be resolved in real-time trackingof3D welding seam during robotic GTAW process based on visual sensor and arc sensor. The first one is accurately retrieving the3D welding seam.The second one is that the right signal processing method must be found,which should be accurate enough to meet the requirements of real-timetracking of3D welding seam during robotic GTAW, and the characteristics of3D welding seam should be extracted. The third one is that the design ofreasonable controller for the seam tracking must meet the requirement ofreal-time welding seam tracking for welding robot GTAW process.Firstly，this paper designed a set of tracking and control system forthree-dimension welding seam during welding robot GTAW process based onvision sensor and Arc sensor, including the setup of experimental platform,the design and processing of sensors, the calibration of welding equipment,the design of system control software and welding fixture, etc. Theexperiment platform can make better effect of real-time control onthree-dimension welding seam during welding robot GTAW process.To achieve accurate real-time seam tracking during welding, theaccurate information which is used to describe welding process should beobtained first. Thus, this paper designed a set of system of passive visionsensor for the robot to track the plane welding seam in real-time. The qualityof image acquisition is associated not only with dimmer-filter system, butalso with the value of base current and the time of capturing image. Througha large number of experiments, the parameters of Al alloy pulsed GTAW areascertained, such as the composite dimmer-filter system, the base current,the time of capturing images, etc. By using the passive vision sensor system,we can get some clear images during the robot weld automation process.Aiming at the deficiencies of existing edge detection algorithm, a newimproved Canny algorithm has been proposed to detect the edges of seam and pool, which effectively enhances the precision, accuracy and selfadaptability. By analyzing the features of welding images, a set of completealgorithm of image processing is proposed to extract the characteristicparameters of welding images. Finally, the algorithm of the image processingwas validated through the experiment. The precision can be controlled aboutwithin±0.169mm, which can completely meet the requirement of real-timewelding seam tracking for welding robot GTAW process.By analyzing the characteristics of pulsed GTAW for Al alloy, the arcvoltage signals are successfully gathered by the arc sensor, and A newde-noising method of anti-impulse interference moving average based onwavelet packet is studied in this paper. The characteristics of arc voltage areextracted, which can be accurate to0.273mm. The relational model betweenarc voltage and arc length, which is U3.108h21.472, has been worked outand then confirmed by tests in the paper. The experiments have suggestedthat the characteristics of arc voltage, the arc-length calculated by therelational model and the real arc-length can be well matched, for the minimalerror is about0.218mm. Therefore, in accordance with the above points, wecan conclude that the accuracy of the relational model between arc voltageand arc length can completely meet the requirements of real-time trackingand control in height during robot GTAW processBy analyzing the deviation rectification of welding robot and thecharacteristics of3D welding seam, a compound Fuzzy-PID double modecontroller based on fuzzy rules switching has been designed for the system of3D seam tracking control during robotic GTAW process. By using fuzzylogic rules, it can shift from Fuzzy controller to PID controller according todifferent seam deviations, which ensures the maximum real time, anti-interference performance, and good stability of seam tracking in weldingprocess. It has been proved by the simulation-system test that the compoundFuzzy-PID double mode controller enjoys the advantages of excellentrobustness, strong anti-interference and tracking abilities, which cancompletely meet the requirements of real-time tracking and control duringrobot GTAW processFinally, aiming at the complex three-dimension curve seam, someexperiments have been done to verify the precision of3D seam trackingbased on visual sensor and arc sensor technology. The results demonstratethat the precision range of planar seam tracking and arc length tracking canbe controlled within±0.3mm and0.4mm, respectively, which is accurateenough to meet the practical welding requirements of the real-time trackingand controlling during the welding robot GTAW process.