A neural network adaptive observer for field oriented control

A neural based adaptive observer is proposed for field oriented asynchronous machine control systems. It comprises two artificial neural networks (ANN) which are trained to learn the rotor flux and stator voltage dynamics respectively. After training, the ANN adjusts the slip signal command, thus providing decoupling of the flux terms. The ANN are subject to online training performed on the basis of voltage errors. Simulation results show that the proposed observer behaves very satisfactorily under rotor time constant changing conditions.<<ETX>>