Estimation of Nonlinear Dc-motor Models Using a Sensitivity Approach

A nonlinear model structure for a permanent magnet DC-motor, appropriate for simulation and controller design, is developed. The essential nonlinearities are due to coulomb friction and to voltage and velocity dependent brush resistance. The physical parameters of the nonlinear model are estimated directly, using a sensitivity approach for input design and evaluation of relative accuracy of the estimates. Experimental results demonstrate that linear models, as well as nonlinear models with the parameters determined from traditional static measurements, fit dynamic measurements poorly. A nonlinear model with the parameters estimated from dynamic measurements, however, fits the measurements very well, and the sensitivity measures combined with cross-validation results indicate that this model is robust and accurate.