Composite intelligent learning-based tracking control for discrete-time repetitive process

In this work, a new composite iterative learning control (ILC) algorithm for the tracking issue of a class of discrete-time systems that operate repetitively over a finite time duration is developed. Particularly, the proposed intelligent learning process consists of two phases to achieve an enhanced tracking performance: the gain-adaptive iterative learning control (GAILC) phase and the sliding mode iterative learning control (SMILC) phase, respectively. Moreover, the switching of the two phases is determined by the tracking error. For GAILC phase, a prediction of tracking error based adaptive gain sequence is adopted to achieve a fast convergence in tracking error. For SMILC phase, an appropriate sliding surface function in the iteration domain is established, and then a novel SMILC law with a fractional power term is presented to achieve a high tracking precision. Finally, comparative simulations including a DC motor example are provided to validate the effectiveness and advantage of the proposed ILC strategy.