Sensorless in duction Motor Drive Based on Reactive Power MRAS Estimator Using Artificial Neural Network

This paper describes the problematics of modeling advanced methods of control of an induction machine such as sensorless control with the use of type of reactive power model reference adaptive system (Q-MRAS) observer with a neural network. In the beginning, problematics of the sensorless control of an induction machine is proposed, mainly in the way of MRAS possibilities with a specific description of Q-MRAS. Then, variants of Q-MRAS implementations are mentioned, also with the modern use of artificial neural networks within the structure of this type of observer. Further, there is described the realization of the simulation model of Q-MRAS observer containing the artificial neural network in Matlab Simulink interface. Simulations are made for the different values of the desired speed and its step changes. In conclusion, there is discussed the suitability of this type of observer due to the simulation results.