Improved performance of motor drive using RBFNN-based hybrid reactive power MRAS speed estimator

Model Reference Adaptive System (MRAS) represents one of the most attractive and popular solutions for sensorless control of induction motor drives. However, the performance of this scheme deteriorates at low speed. A new method is described which considerably improves the performance of MRAS-based sensorless drives in low speed regions of operation. It is applied to a vector-controlled induction motor drive. This new technique uses Radius Basis Function Neural Network (RBFNN) to entirely replace the conventional Proptional- intergral (PI) adaptation mechanism of classical hybrid reactive power MRAS speed estimator. The simulation results show great improvement in the speed estimation performance at low speed.