Synchronous Control of Flying-type Welder Based on RBF Neuron Network

According to the traveling behaviors of flying-welder and billets being rolling in a EBROS (endless-bar rolling system), an inverse mode of welder driving system is identified off-line based on RBF neural networks, a disturbance observer is established with a filter to predict and at last remove the disturbances existed inside the system or from outside; a single neural composite-controller is used to improve the dynamic and static properties of the system, and strength the tracking performance of welder for the billet being rolled. The simulation results on the system dictates that this control strategy can well remove the mechanical coupling forces between the billets being rolled and the welder, improve the dynamic tracking performance of welder for the billet being rolling.