Investigations on Training Algorithms for Neural Networks Based Flux Estimator Used in Speed Estimation of Induction Motor

Estimation of exact speed information is must to have speed sensorless drive. This paper presents an artificial neural network assisted speed estimator for induction motor drive. A multilayered neural network (MLNN) is used to estimate the rotor fluxes. These estimated flux values are used in stator current based model reference adaptive system (MRAS) scheme for speed identification. Different learning algorithms are examined for training the MLNN. This estimated speed is used for sensorless vector control of induction motor drive. To investigate the performance, MATLAB software is used.