Techniques de commande avancées de la machine asynchrone

Abstract

The aim of this work was carried on the experimental comparison of the performance of field oriented vector control and DTC control, and to develop control and observation algorithms with conjoint on-line estimation of the induction motor (IM) parameters, taking into consideration the constraints imposed by the practical realization. We took a special interest to the experimental test-bed as it is an essential step for validating each new control law design. A theoretical study is devoted to the basic concepts of variable structure systems with sliding mode, and after describing the conventional method, we proceeded to the synthesis of control law that is based on exponential reaching phase (ERL) to achieve more robustness of the system even at transients and to improve its dynamic response. At the end, a new sensorless control by MRAS with an on-line conjoint estimation of the IM parameters using artificial neural networks is developed. To do so, we proceed with a theoretical study of the MRAS technique evoking its different methods. Then we continue with the process of estimating the IM parameters espacially the stator resistance and the inverse of the rotor time constant using the concept of Popov’s hyperstability. After describing the principles of neural networks and their application to our case in the estimation of the parameters we go further to the practice implementation and interpretation of results.