Adaptive Predictive Iterative Learning Control for Constrained Nonlinear Systems Under Varying Operating Environments

In this work, a new adaptive predictive iterative learning control (APILC) scheme is designed for a class of multiple-input-multiple-output (MIMO) discrete-time nonlinear systems, which simultaneously addresses the problems of randomly varying iteration lengths, iteration-time-varying system uncertainties on parameters and disturbances, iteration-time-varying reference trajectories, and system constraints. First, in order to compensate for missing output/state data caused by randomly varying iteration lengths, a new search decision compensation mechanism (SDCM) is constructed to select optimal data from historical data, estimated data, and predicted data, thereby mitigating the impact of randomly varying iteration lengths on the tracking performance. Next, a new adaptive learning algorithm is developed based on the compensated data, which not only estimates and predicts iteration-time-varying system uncertainties on parameters and disturbances, but also constructs a more accurate adaptive prediction model to effectively capture future dynamic characteristics of the system. Furthermore, the constructed adaptive prediction model is employed to design an APILC scheme that simultaneously handles both iteration-time-varying reference trajectories and system constraints. Theoretically, the convergence of both the adaptive prediction model and the tracking error is rigorously guaranteed, even under system constraints and various iteration-time-varying operating environments. Ultimately, the simulation results verify the effectiveness of the proposed SDCM based APILC (SDCM-APILC) scheme.