ANFIS BASED MRAS SPEED ESTIMATOR FOR SENSORLESS CONTROL OF PMSM

Sensorless Control of PMSM has gained a great attention recently. Many algorithms were proposed in this topic among which is the MRAS based speed and position estimators due to their great immunity against motor parameter variations and nonlinearities through different operating conditions. In this Paper an online tuned Adaptive Neuro-Fuzzy Inference System (ANFIS) based MRAS speed estimator is proposed, where the conventional adaptation algorithm based on Popov Hyperstability theory is replaced by ANFIS algorithm adaptation. The reference model for the MRAS is the motor itself while the current model in the rotor reference is the adjustable model. The error between both the reference and adjustable model is utilized to tune the ANFIS parameters through a hybrid training algorithm. The estimator is able to track the speed efficiently at different operating speeds and varying motor parameters.