Neural-network-based nonlinear iterative learning control: Magnetic brake study

The objective of the paper is to provide an approach to design an effective nonlinear iterative learning control scheme for the magnetic brake system. Also, the discussion on the conditions to guarantee the convergence of the tracking error is carried out. The proposed control scheme takes into account the initial system state error and initial feedback controller state, as well as disturbances and noise acting on the system. The control scheme uses two neural networks. One for the modeling the plant dynamics and the second for implementing a learning controller with a trial-varying structure. Thus, the control system is able to adapt to changing working conditions of the plant. Developed control strategy was applied to a magnetic brake system in order to follow the desired reference with the acceptable level of the tracking error.