Neural network control on direct torque control for brushless DC motor

In order to simplify the control system's structure and reduce the torque ripple restraint, direct torque control (DTC) and neural network control strategy were combined in the control system of the brushless DC motor's control system. The DTC control can leave out complex vector quantity transfer. According to the error between torque and its reference value and the error variance ratio, using the non-linearity control characteristics of the neural network, adjusted the active time of the space voltage and reduce the torque ripple. The experimental result prove that the neural networks DTC method has good dynamic characteristics and simple structure, makes effective control on torque and current, and the fluctuation of the phase current and the torque decrease by about 50%. It enhances the control precision of the BLDCM.