BLDCM (brushless direct current motor) is a complex system with multi-variable, nonlinear, time-varying parameters and strong coupling. The traditional double closed loop PID (proportional integral differential) algorithm has some problems to drive BLDCM such as bad parameter tuning, poor adaptability, low control accuracy and weak anti-interference ability. In order to achieve high precision control for BLDCM, a single neuron neural network PID algorithm was proposed for motor speed loop control. The mathematical model of BLDCM was studied by the single neuron neural network PID algorithm and the operation characteristics were analyzed based on this system. Finally, the speed step function response, the operation state of three-phase current, back EMF (electromotive force) and electromagnetic torque were compared and analyzed, respectively. The results show that the speed step function response controlled by the single neuron neural network PID algorithm has a faster rise time, a smaller overshoot and a more stable operation state.