A Sensorless Model Predictive Control for Induction Motor Based on Ultra-Local Model

In this paper, a sensorless model predictive control (MPC) for induction motor using an ultra-local model is studied. Firstly, the speed, stator flux, and stator resistance are estimated online using a sliding mode observer (SMO). Secondly, in sensorless model predictive control, due to the sensitivity of traditional predictive models to changes in motor parameters and the existence of errors between estimated and actual speeds, the accuracy of the predictive model will be reduced. In response to the above issues, this paper optimizes the prediction model. For the flux linkage prediction model, a correction term is introduced and the stator resistance in the flux linkage prediction model is updated in real-time through a sliding mode observer. For the current prediction model, an Ultra-local model is introduced instead of the traditional current prediction model, which does not consider any motor parameters. Therefore, this method has better robustness performance. Finally, the method is validated through simulation.