Intelligent Learning Controllers for Nonlinear Systems using Radial Basis Neural Networks

Iterative learning controllers are a good choice for repetitive trajectory tracking tasks because they do not need identification of a nonlinear system. Starting from zero knowledge of the system, these types of learning controllers take a certain number of iterations before converging to the desired trajectory. In the case of many desired trajectories, learning takes almost same amount of iterations for every desired trajectory. In this article intelligence is incorporated in the iterative learning controllers using neural network for a class of nonlinear systems. The experience of iterative learning controller with different desired trajectories is stored in the neural network. For a new desired trajectory, this neural network generates the initial control input and feeds it to the iterative learning controller. This approach is proved to be very effective in improving the convergence of the tracking error. Our proposed method is very general and applicable to most of the iterative learning controllers without modifying their simple learning structures.