The neural network speed controller based on Fuzzy PI for direct torque control

To increase traditional direct torque control (DTC) of induction motor control precision and decrease large torque ripple, a Fuzzy PI speed controller was proposed. On the basis of conventional PI regulator, a Fuzzy PI speed controller was designed according to speed error and its rate of change, which could adjust the proportional coefficient k_p and integral coefficient k_i dynamically to adapt the speed variations. A new strategy of feed forward neural network (NN), which replaces the Fuzzy PI speed controller is also proposed and applied to DTC system. Meanwhile, a backward propagation (BP) algorithm was used to train the network. The comparison with conventional PI speed controller shows that the proposed method reduce the flux, speed and torque ripples. The validity of the proposed method is verified by the simulation results.