Based On Weld Pool Visual Of The CO₂Welding Defects Characteristics Recognition

Welding technology holds an important place in modern industry, the quality ofproducts and the use of the reliability is very important. But there are inevitable in theprocess of welding technology of interference factors such as strong arc, it is difficult tocollect to the ideal of the molten pool image information and analyze it. Therefore, inview of the above problems that studies on welding molten pool image information ofdynamic characteristics for the real-time control of the welding process and weldingquality monitoring in welding process to provide a guarantee. According to thecharacteristics and advantages of the carbon dioxide gas shielded welding, in themodified passive welding robot vision sensing device, deserve to go up the CCD visualsensor and narrow filter to avoid the interference of strong arc light and noise, so that thecollected as little as possible with stray light of the weld pool image information forresearch. For the welding robot each rod connection of attitude angle transformation inthe space, based on D-H coordinate transformation theory to establish mathematicalmodel for spatial geometric correction, which the weld pool features work of thegeometry provided conditions for further study.Based on Matlab for the image of digital image processing technology to theacquisition of the weld pool are pre processing, advantages of combining spatial andfrequency domain, molten pool image in the frequency domain using the Laplacetransform to use Butterworth low-pass filter When transitioning into a spatial gradientenhanced suppression during the soldering process on light interference in someunfavorable factors, such as the noise generated by impurities. Respectively usingRoberts, Sobel, Prewitt, Log, Zerocross, Canny six kinds of differential operator andmulti-scale wavelet transform algorithm of molten pool image edge detection ofinformation analysis. Found in edge detection threshold [0.1,0.5] Canny operator moltenpool edge extraction full of clear, than Wavelet in dealing with carbon dioxidegas-shielded welding robot gets edge welding molten pool image information is moreappropriate.Through the design of welding robot welding experiment, obtain goodcharacteristics at the optimal conditions of the weld pool image and the synchronizationacquisition in the non normal process of weld pool image case, getting under thecondition of ultra welding current voltage, molten pool appear porosity defect isincreasingly obvious, control current and voltage has a certain effect on improving the welding quality. Explore different angles and flat surfacing welding groove under thesame experimental conditions, carbon dioxide protective gas flow rate is too low in thewelding process also affects the welding molten pool images, the phenomenon of theporosity. In view of the appeared in the process of robot welding wear, lack ofpenetration, weld pool shape defects using image processing after the welding moltenpool image gray value distribution of the weld pool image acquisition of integral closureedge area, according to the variation of the weld pool geometry area can be used todistinguish the shape defect occurrence. Also verified the established mathematicalmodel of space and the image processing method of reliability correctly, and provide newmethod and reference for the welding quality monitoring in the future.