Abstract
In order to achieve high performance of sensorless direct torque controlled induction motor drive at medium and low speed regions in case of Gaussian-noised stator currents, extended Kalman filter is utilized. At first, sensorless control using rotor-flux-based model reference adaptive system is described. Then, extended Kalman filtering that uses full state-space model of the induction motor is employed to obtain estimated stator currents for the sensorless control. Unmeasured rotor fluxes in extended Kalman filtering are computed based on their relationship to estimated stator fluxes and measured stator currents. The estimated stator currents are utilized to compute input quantities for direct torque control. Simulations are deployed in case of both process and measurement noises of stator currents. Performance comparisons based on two indices: normalized integral of time multiplied by absolute value of speed difference and maximum value of absolute value of relative speed difference between two sensorless control methods with and without extended Kalman filter, are carried out. Through simulations in Simulink environment of Matlab software, theoretical assumptions are confirmed by the fact that the evaluation indices of the proposed method are decreased by at most 75% and 80% compared to the method without extended Kalman filter.