Neural-network-based high-order iterative learning control

The purpose of this work is to develop an effective approach to design the high-order iterative learning control scheme based on neural networks. The idea adopted here is to propose the high-order learning controller which is able to combine the control information from the previous trials and in this way to improve its convergence speed. Sufficient conditions for convergence of the developed ILC scheme are derived and constructively incorporated into training process of neural controller. The efficiency of the proposed approach is illustrated on the example of a pneumatic servomechanism.