Study Of Intelligent Control And Processing Of Multi-sense Information For Pulsed MIG Welding Of Aluminum Alloy

In order to fit the requirement of lightweight of transportation, more and more aluminum alloy welding structures have been widely applied. Pulsed MIG welding is one of the important techniques of aluminum alloy welding, so research on a system of real-time sensing and control of robot pulsed MIG welding for aluminum alloy is of vital importance. Because dynamic process of the aluminum alloy pulsed MIG welding is a multivariable complex system with strong non-linearity , strong coupling and time-varying, taking on strong arc and many kinds of metal transfer ,accompanied by smoke dust , electromagnetic interference and other disadvantage factors, traditional control theory cannot achieve good results. Therefore, intelligent control method is used for this problem.PID controller and intelligent controller for weld pool width of aluminum alloy pulsed MIG welding has been used in this paper based on an identified function model, simulations also been carried through. A Fuzzy controller combining with an expert system was designed after comparing the two controller's control results. This control system can solve the problem of multivariable coupling in the aluminum alloy pulsed MIG welding. Surfacing experiment on the flat using the PID controller and fuzzy-expert system controller were done, the results show that the control effect of the fuzzy-expert system is better than the PID controller.The experiment show that wire extension and arc length can influence the stabilization of welding process and the welding-seam modeling when using fuzzy-expert system .In order to control the wire extension and the length of the arc, the vision sensor and the image processing arithmetic of the wire extension were researched and the identified function model of the wire extension via the wire feed speed was set up by least square method when using the step as the input. The simulation results with this model can fit well the actual experiment results. The control port of the MIG/MAG welding machine was analyzed. This can use as the direction for controlling the wire extension and aluminum alloy multivariable decoupling control.The sensor methods often being singleness in traditional welding control, this cannot reflect the all-sided of the welding process. Themulti-sense information together model was set up. This model uses the vision system to detect the welding pool and use the arc sound to check the stabilization of the welding process. Different signal process methods were discussed in disposing the arc sound. The wavelet transformation and power-spectrum analysis method were finally used to analyze the feature extraction of the sink point of the pool and different metal transfer in the aluminum alloy MIG welding. This can be used as valuable references and direction for detecting pool sink and metal transfer in welding process.