Self-Tuning of Individual Control Components for Systems with Parameter Uncertainties and Disturbances

In this paper, a self-tuning algorithm for the parametrization of a complete feedback control structure or individual components is presented. This approach allows for automatic performance optimization of a motion task w.r.t. a chosen evaluation cost function (ECF). The proposed algorithm is designed especially for systems with only approximate knowledge about the system parameters that are, moreover, affected by external disturbances. Using optimal control techniques and knowledge of the nominal system only, an initial control structure is designed that involves feedback and feedforward control as well as a state and disturbance observer. The proposed algorithm is capable of determining an optimal control parametrization – ensuring closed-loop stability and close-to-optimal tracking behavior. The self-tuning algorithm is applied to an elastic machine tool axis in simulations and the potential of tuning the individual control components w.r.t. the tracking accuracy is investigated in detail. Finally, the obtained simulation results are compared against each other – showing the efficiency and benefits of the overall approach.