Application of Measurement Simulation and Least Squares Parameter Estimation in an Undergraduate Servomotor Lab

Laboratory confirmation of theoretical results is important for student confidence in both analytical and experimental techniques. A frequent source of difficulty in obtaining such a confirmation is the improper handling of error-corrupted measurements. This paper discusses measurement simulation and least squares parameter estimation as applied in a dc servomotor laboratory. Simulation illustrates how ideal measurements are obscured by noise, and how usable estimates are recovered from noisy data. Many control system parameters are estimated by the use of linear least squares. Motor time constant is estimated by the method of quasi-linearization combined with least squares. Applications of estimation statistics are described. All measurement and estimation simulations are demonstrated graphically by plots showing both point of measured values and the least squares curve fits.