A neural network-based commutation optimization strategy and drive system design for brushless DC motor

An optimized commutation method based on backpropagation(BP) neural network is proposed to resolve the low stability and high-power consumption caused by inaccurate commutation point prediction in conventional commutation strategy during acceleration and deceleration.This article also builds a complete brushless DC motor drive system based on the GD32 F103 micro control unit(MCU),with an Artix-7 XC7 A35 T field programmable gate array(FPGA) to meet the performance requirements of neural network calculation for real-time motor commutation control.Experimental results show that the proposed optimization strategy can effectively improve the system stability during system acceleration and deceleration,and reduce the current spikes generated during speed changes.The system power consumption is reduced by about 11.7% on average.