The paper recapitulates recently conducted investigations of non-proportional Luenberger observers, applied to reconstruction of state variables of induction motors. Three structures of non-proportional observers are analyzed, a proportional-integral observer, modified integral observer and observer with integrators. Criteria for gain selection of the observer are described, classical ones based on poles, as well as additional, increasing observer’s robustness. Fulfilment of the presented criteria can be ensured with the three proposed methods for gain selection, two analytical, based on dyadic transformation and one based on optimization.
In this paper, we propose sensorless backstepping control of a double-star induction machine (DSIM). First, the backstepping approach is designed to steer the flux and speed variables to their references and to compensate uncertainties. Lyapunov”s theory is used and it demonstrates that the dynamic tracking of trajectories tracking is asymptotically stable. Second, unfortunately, this law control called sophisticated is a major problem which leads to the necessity of using a mechanical sensor (speed, load torque). This imposes an additional cost and increases the complexity of the montage. In practice, this variable is unknown and its measurement is expensive. To restrain this problem we estimate speed and load torque by using a Luenberger observer (LO). Simulation results are provided to illustrate the performance of the proposed approach in high and low variable speeds and load torque disturbance.