Artificial neural network-based estimation for rotor-flux model reference adaptive system

Abstract

At the start, this paper focuses on the function of a rotor-flux model reference adaptive system (RF-MRAS) and in the following part on the realization and application of artificial neural networks (ANN) in a sensorless induction motor drive. Afterwards, a data collection and usage process for the training of ANN is described. In the final part, experimental results of ANN's ability to estimate rotor flux are presented. According to simulations, ANN estimations are accurate and its application as a part of a control scheme looks promising.