Research On Welding Quality Monitoring For Pulse Multi-control GMAW Of Aluminum Alloy

The fierce market competition in modern manufacturing industries demands improvement in weld productivity.Increase the welding speed is an effective and economic approach to achieve enhanced productivity.Aluminium alloy which is characterized by low melting point,high heat conductivity and high coefficient of linear expansion.However,welding speed higher than a critical value often causes weld defects such as undercut,lack of fusion and humping.These defects can cause reduction in static,fatigue and fracture strengths of a welded assembly,which becomes the chief factor that restricts the improvement of production efficiency.Pulse Multi-Control Gas Metal Arc Welding is a newly developed high-efficient GMAW method.Its Optimized pulse characteristic and improved ignition can significantly reduce the overall heat input,which makes it suitable for the welding of aluminum alloyIn this paper,an on-line welding monitoring system based on welding current and welding quasi-stable state temperature field was developed in order to realize the real-time dynamic monitoring of the aluminum alloy PMC GMAW process.The time-frequency analysis method,feature extraction and classification model refinement have been further studied.The welding disturbance and weld defects multi-class recognition has been selected as the research goalBased on Lab VIEW graphical programming language,the welding monitoring system control software was designed and developed.The software is not only simple to operate but also realizes the function of real-time signal displaying and historical data query.The welding current signal and quasi-stable state temperature field was real-time captured and stored during the welding process,which provides reliable data sources for the subsequent data analyzing processThe emissivities of weld and base metal with different surface conditions were calibrated,which proved theoretical support for analyzing the abnormal temperature data which was collected by the infrared sensing system.5000 aluminium alloy welding wire would generate dark-grey oxides on the surface of the weld and heat-affected zone(HAZ)during the welding process.The calibration results indicate that the dark-grey oxides can seriously change the emissivity.An effective method based on analyzing the characteristic temperature distribution curve of the quasi-stable state temperature field was proposed for identifying the high-speed weld defects.Typical high-speed welding defects such as undercut,lack of fusion and humping were created by applying improper travel speed.The temperature data of the weld and HAZ was measured by an infrared sensing system during the welding process.These data formed the quasi-stable state temperature field of a sound weld and welds with above three high-speed weld defects.Based on the analysis of the quasi-stable state temperature field with different weld defects,characteristic temperature distribution curves were extracted.The relationship between the characteristic temperature distribution curves and the above three high-speed welding defects was studied.The result shows that skewness and kurtosis of the characteristic temperature distribution curve are sensitive to the defects,which can effectively represent the morphological characteristics of the characteristic temperature distribution curve.When taking these two parameters into account,the sound weld would be identified form welds with defects Kalman Filtering method was applied to generate the IR image of the quasi-stable state temperature field.The temperature distribution of weld and HAZ can be displayed in a more intuitive way.The type and location of weld defects in the IR image matched the real weld appearanceAn efficient approach based on the frequency spectrum of the characteristic intrinsic mode function(IMF)of the welding current was presented for identifying the interference during the welding process.First of all,the ensemble empirical mode decomposition(EEMD)was used to decompose the welding current signal.The time-frequency analysis method was applied to the decomposition result.Based on the analysis result,the characteristic IMF which is closely related to the short-circuit frequency was identified The relationship between the short-circuiting frequency and the process stability is established.Experiment results have shown that the frequency components would centralize in a narrow band when the welding process is stable.Experiments were conducted by applying improper travel speed and under the inappropriate welding condition(insufficient protection gas and oxide film on the surface).The changing law of the frequency spectrum of the IMF was studied.The ratio of dominant frequency energy of the IMF frequency spectrum to the overall energy of the IMF frequency spectrum was selected as the characteristic parameter.Combine with the dominant frequency of IMF frequency spectrum,welding current,travel speed,different types of interference can be identified and classifiedA particle swarm optimized multi-class Support Vector Machine(SVM)model was built in order to identify and classify the weld defects and disturbance during the welding process.Welding current,travel speed,skewness and kurtosis of the characteristic temperature distribution curve,The ratio of dominant frequency energy of the IMF frequency spectrum to the overall energy of the IMF frequency spectrum,the dominant frequency of IMF frequency spectrum were used as the model inputs.The sound weld,undercut,lack of fusion,humping,subsidence,spatter were selected as the outputs for the weld defects multi-class recognition model.The stable welding process,excessive welding speed,insufficient protection gas and oxide film on the surface were chosen as the outputs for the welding interference multi-class recognition model.The results have shown that the correct recognition rate for weld defects reaches 96.85%and the correct recognition rate for welding interferences reaches 95.14%.