Real time model of artificial neural networks-based speed estimator for an induction machine drive

Thanks to the theory of Vector Control (VC), high performance speed and torque response are achieved from the induction machines nowadays. Driven by a VC controller, an ac machine behaves similar to separately excited do machine in which the torque and flux are controlled independently.; In this work the advantages and drawbacks of the different schemes for sensorless vector control of the induction machines are studied and finally a back emf based Model Reference Adaptive Systems (MRAS) technique using a recurrent Neural Network as the speed estimator is chosen which is believed to be a new and robust method. Neural networks are analytical systems that address problems whose solution have not been explicitly formulated. A neural network is an information processing system that is non-algorithmic, non-digital, and intensely parallel which makes it a suitable choose for using as the mathematical technique in simulating our estimator. The system is trained online to achieve a real-time application. Such a kind of training has no need to pre-computations and makes the system flexible and robust in a wide range of operation. It is believed that the work is an important contribution in this area of research. (Abstract shortened by UMI.)...