Neural network based terminal iterative learning control for tracking run-varying reference point with initial state variance

In this paper, a neural network based terminal iterative learning control (NNTILC) method is proposed for a class of discrete time uncertain linear systems to track run-varying reference point. The zero error initial condition in most of the previous work on terminal iterative learning control (TILC) is removed by the use of neural network. A radial basis function (RBF) neural network is developed to approximate the effect of initial state and reference on terminal output iteratively. By involving these information as well as the reference signal in the control scheme, the proposed NNTILC can drive the system to track run-varying reference point fast and precisely beyond the initial state variance and reference change. Stability and convergence of this approach are proved and computer simulation results are provided to confirm its effectiveness further.