Comparison of stochastic and deterministic parameter identification algorithms for indirect vector control

The indirect vector control method provides accurate, rapid control of an induction machine's developed torque. To maintain the control performance, the position of the rotor flux vector has to be known accurately. In this paper, a stochastic and a deterministic state-space estimator are compared using experimental results, as regards their estimate of the rotor resistance. Simulation results are used to compare stationary and synchronous reference frame versions of the extended Kalman filter (EKF). Induction machine core losses are usually neglected due to the required increase in model complexity. However, a standard method of core loss compensation which does not require higher order estimators is shown to improve the EKF and extended Luenberger observer accuracy, while requiring only a minimal increase in computation. The experimental system used to implement vector control and online rotor resistance estimation is also described.