Design of Permanent Magnet Synchronous Motor Controller Based on Recurrent Neural Network

A novel vector control method of a permanent magnet synchronous motor (PMSM) using recurrent neural network(RNN) is presented.The RNN controller is used as a speed controller to mimic an optimized speed output under the condition of motor parameters variations and load perturbation.The RNN is trained online by using the extended Kalman filter(EKF) algorithm,and the learning rates are obtained in the sense of Lyapunov stability theory.The proposed RNN vector controller has shown good performance in the transient and steady states,and also at either variable-speed operation or load variation.The validity and the usefulness of the proposed algorithm are thoroughly verified with experiments on fully digitalized 1.2 kW PMSM drive systems.