MIMO Model Identification And Control System Simulation For Aluminum Alloy MIG Welding Process

In resent years, with the lightweight development of transportation means such as automobile, high speed train and so on, the demand of aluminum alloy is more and more great. It requests that the production of aluminum alloy welding structures be more automatic and intellectualized. But the particularity of aluminum alloy welding and the poor adaptability of teaching-playback robot limit the application of arc welding robot in aluminum alloy welding production seriously. In order to overcome the influences on welding quality from kinds of uncertainty factors in robot welding process and enhance the intellectualized level and the work reliability of arc welding robot, the welding robot system is requested to realize the online adjustment of welding parameters and real-time quality control of welded joint. Multivariable control system based on visual sensor will be one of the popular methods to solve above welding problems.Visual sensing system of aluminum alloy welding area image was used in this thesis and clear images of welding area were captured. The disturbance of arc light was wiped off by narrow band filter effectively. After analyzing the characteristics of welding area images, the extraction algorithm of wire extension was purposed, succeeding the morphology weld pool width extraction algorithm. The experiments show that the above extraction algorithms can obtain the stable and reliable pixel values of weld pool width and wire extension.The relationships between welding pool width, wire extension and welding parameters including welding current duty ratio, base current, peak current, welding speed and wire-feeding speed were studied. Through large numbers of welding experiments, aluminum alloy pulsed MIG welding dynamic process were identified by the curve fitting method, and the mathematical models between welding current duty ratio, base current, welding speed, wire-feeding speed and wire extension were obtained, succeeding the model between the above parameters and weld pool width. After model parameters being revised, the multi-input multi-output (MIMO) model of aluminum alloy robot pulsed MIG welding process was established. Through the MIMO model, the weld pool width control system of aluminum alloy MIG welding process was realized and the welding wire extension control system of aluminum alloy MIG welding process was simulated. The aluminum alloy MIG welding experiments of the weld pool width control system indicate that the welded joint under only the weld pool width real-time control still not satisfy the high grade welding requirements of aluminum alloyand needs many parameters to control its quality. The simulation results show that the welding wire extension PID control system can satisfy the control requirements of the actual aluminum alloy pulsed MIG welding process.Based on aluminum alloy pulsed MIG welding MIMO model, the different double-variable control systems were used to control welding process in the thesis. The simulation results of different control systems show that the double-variable PID control system without decoupling unit can not obtain the suitable system performance and the neural network inverse plant decoupling control system with the neural network PID controller can obtain the satisfied dynamic and steady-state system performance.