摘要：AbstractWe consider Iterative Learning Control for the Furuta Pendulum nonlinear mechanical system, where the goal is to learn the input torque such that the pendulum angle follows a reference. We show that the linearization of the considered system is of variable trajectory-dependent order and thus some existing solutions do not apply. We propose a novel method based on the observability matrix inversion allowing to deal with the variable-order minimum realization. The applicability of the proposed method is illustrated with simulations.
