Iterative tuning algorithm for feedforward and feedback control of two-mass motor system with physical parameter identification

Abstract

This paper proposes a novel iterative tuning algorithm for achieving velocity control of a two-mass torsional motor system with uncertain parameters. The command generator tracking (CGT) theory is used to establish a direct relationship between the physical model parameters of the system and the feedforward parameters of a special 2DOF controller. This relationship enables the controller to be tuned by physical model estimates, which are shown to converge to true values its the tracking error decreases. The proposed tuning method employs these estimates to construct an approximate system transfer function that is used to analytically calculate the output error gradient This allows for the application of the iterative feedback tuning (IFT) scheme, without the need to perform additional gradient estimation experiments per iteration. The result is a self-sufficient iterative algorithm that utilizes the established relationship between physical model parameters and feedforward controller parameters to achieve auto-tuning of an unknown system. The effectiveness of the proposed method in realizing trajectory tracking while simultaneously identifying all physical parameters is demonstrated via experimentation of a two-mass motor system