NEURAL NETWORKS SOLUTIONS FOR AC MOTOR CONTROL

摘要

The paper presents alternative neural network (NN) solutions for the control of induction motor (IM) and permanent magnet synchronous motor (PMSM). Both NN methods, based on vector control, were compared with the classical respective method control for IM and PMSM. The proposed NN for IM is related with sensorless drives, based on a model reference adaptive system (MRAS) and the NN is designed to estimate the electrical rotor speed. As an alternative solution to classical vector control of PMSM, an inverse model neural controller with a simplified mechanism to reduce the speed offset during the control process is proposed. All the algorithms were experimentally tested in order to prove capabilities of NN to control the AC machines. The experimental tests have been developed around Motorola DSP56F805 fix-point processor to control the IM and respectively with DS1104 for PMSM control.