A Novel Implementation of Neural Network and Multi-Fuzzy Controllers for Permanent Magnet Synchronous Motor Direct Torque Controlled Drive

To reduce torque ripples and improve dynamic performance, a novel implementation of neural network and multi-fuzzy controller for permanent magnet synchronous motor (PMSM) direct torque controlled (DTC) drive is presented, which replaces the conventional hystersis controller with fuzzy controller (M) and includes a neural network speed controller (NC). To tune the weights and biases of the neural networks online, another fuzzy controller (FC2) is adopted. It combines the capability of fuzzy reasoning in handling uncertain information and the capability of neural network in learning from processes. Results of simulation are provided to demonstrate that the proposed drive has low flux linkage and torque ripples, and has quick response performance even under the occurrence of parameter variations and external disturbance.