Transfer function and parameters identification of a motor drive system using adaptive filtering

A technique is developed for the identification of mechanical parameters in two-mass-model drive systems, The transfer function of the system model is calculated directly from the input-output signals. This technique involves repeated integration of the data and uses the recursive least squares method. It is shown that when the system is subject to random noises and disturbances, the ordinary least squares method is asymptotically biased, and so an iterative scheme is proposed to remove this bias and to improve the estimation efficiency. Simulation analysis is made using the data of an actual rolling mill plant. The accuracy of the estimated mechanical parameters and the effect of measurement and disturbance noise are analyzed. The developed algorithm exhibits outstanding numerical characteristics and has proven to be very effective for the mechanical parameters identification.