Learning Position Controls for Hybrid Step Motors: From Current-Fed to Full-Order Models

摘要

The experimental comparison of two different global learning position controls (namely, “adaptive learning” and “repetitive learning” controls) for hybrid step motors performing repetitive tasks has been recently presented in the literature. Related benefits and drawbacks have been successfully analyzed on the same robotic application. However, the design of the two aforementioned learning controls-though relying on a rigorous stability analysis-are based on a simplified current-fed model of the motor. They cannot achieve precise current tracking due to the mere presence of proportional-integral (PI) control actions in the outer current control loops. The aim of this paper is to enrich and update the results of the aforementioned comparison in the light of the latest contributions that generalize the theoretical design to the full-order voltage-fed motor models of hybrid step motors. Learning actions are now included in the outer current control loops: They generalize the corresponding PI actions to the periodic scenario and allow to solve a control problem whose solution was seeming very difficult to be obtained.