Discrete-time Adaptive Iterative Learning from Different Tracking Tasks with Variable Initial Conditions

Abstract

A new discrete-time adaptive iterative learning control (AILC) approach is developed to deal with systems in presence of time-varying parametric uncertainties. By using the analogy between the discrete time axis and the iterative learning axis, the new adaptive ILC can incorporate a Projection algorithm, hence the learning gain can be tuned iteratively along the learning axis and pointwisely along the time axis. The major advantage of the new AILC is that it can relax the identical conditions on the initial state and reference trajectory, in the sequel achieves an almost perfect tracking performance.