Feedforward parameter identification by output oversampling iterative tuning

This paper proposes an algorithm for iteratively identifying the physical parameters of an elastically coupled two-mass torsional system directly from input-output data. The identification process is conducted under close-loop conditions with a 2-degree-of-freedom controller that incorporates a suitable reference model. It is shown that with convergence of the output error to zero, the controller parameters give a precise estimate of the true physical parameters. The tuning of controller parameters is accomplished in an iterative manner similar to the Iterative Feedback Tuning (IFT), utilizing a linearized plant transfer function obtained through direct closed-loop identification based on the output oversampling approach. The effectiveness of the proposed method in identifying the physical parameters of a two-mass motor system is verified via experimentation.