Comparison of different model-free iterative learning control methods concerning a nonlinear mechanical structure

Abstract

Improving the dynamics of systems in the last decades, model-based control design approaches are continuously developed. In the case of nonlinear system control, the task designing accurate models is important, which is time consuming and nevertheless often not precise (enough). For avoiding precise modeling, model-free control methods become attractive. This contribution considers model-free control technique using a steepest-descent method incorporated into the structure of norm-optimal iterative learning control to assure suitable tracking performance as well as robustness against unknown inputs. It is assumed, that an accurate model of the system to be controlled is not used. Only inputs and outputs of the system are used as measurements. Using the proposed control approaches, the control goal can be achieved efficiently without using control design process. Simulation and first experimental results are shown to demonstrate the successful application of the proposed methods.