A neural networks based field oriented control scheme for induction motors

This paper attempts to introduce the artificial neural network (ANN) principles in the development of an ANN based field oriented control (FOC) of an induction motor. A two-layer ANN has been trained offline to estimate the slip, the direct and the quadrature current commands that are to be used in implementing an indirect FOC scheme. Simulation work indicates the feasibility of this approach as an alternative to the conventional approach, and the practical considerations for implementing this network on a DSP development system are discussed. This paper also proposes a method for estimation and adaptation of the rotor time constant of indirect FOCs. Simulation results show excellent potentials of the proposed scheme.