Decoupling control analysis of aluminum alloy pulse MIG welding process based on dynamic fuzzy neural networks

Considering the strong coupling of parameters, unstable and other key issues during aluminum alloy pulse MIG welding process,D-FNN structure and learning algorithm were introduced.The decoupling controllers were designed based on D-FNN.The dynamic decoupling control simulation of aluminum alloy pulse MIG welding multiple-input multiple-output (MIMO) process,setting the duty cycle of pulse current and wire feeding speed as inputs but wire extension and weld width as outputs, was investigated with synchronization,asynchronous and adding interference pulse.The simulation results indicate that D-FNN controller could real-time evolve rules,dynamically adjust the learning factors,completely decouple the MIMO process and meet the real-time control requirements of welding process.In addition,its fast response speed and good robustness provided a new real-time decoupling control method for stabilizing the aluminum alloy pulsed MIG welding process.