Identification of two-mass mechanical systems using torque excitation: design and experimental evaluation

Abstract

This paper deals with methods for parameter estimation of two-mass mechanical systems in electric drives. Estimates of mechanical parameters are needed in the start-up of a drive for automatic tuning of model-based speed and position controllers. A discrete-time output error (OE) model is applied to parameter estimation. The resulting pulse-transfer function is transformed into a continuous-time transfer function, and parameters of the two-mass system model are analytically solved from the coefficients of this transfer function. An open-loop identification setup and two closed-speed-loop identification setups (direct and indirect) are designed and experimentally compared. The experiments are carried out at nonzero speed to reduce the effect of nonlinear friction phenomena on the parameter estimates. According to results, all three identification setups are applicable for the parameter estimation of two-mass mechanical systems.