An iterative learning control algorithm based on predictive model

Abstract
An algorithm of iterative learning control(ILC) based on predictive model is proposed for a kind of repetitive tracking process of the discrete time system with CARMA model. The repetitive tracking process is operated along with the reference trajectory with performance of predictive control based on predictive model of the one step minimum variance. The convergence of this algorithm is analyzed and convergence conditions are derived. The algorithm for linear stable process can be achieved one iteration unbiased tracking for any changing trajectory when the estimation of model parameters is unbiased. In the car suspension system as an example, the simulation results demonstrate this algorithm can achieve fast unbiased tracking for the changing trajectory. It can still achieve unbiased tracking by 4~5 times of iterative learning control while errors of model parameters estimation are changing in ±30%.