Aiming at the shortcomings of the traditional rotor state estimation method in the sensorless FOC control of PMSM in noise suppression and low speed performance, a high-precision control scheme based on EKF is proposed and a driver circuit system integrated with host computer communication is designed; Firstly, the mathematical model of PMSM in dq coordinate system is constructed, and the back EMF, rotor position and speed are optimally estimated by EKF to solve the estimation error problem of traditional methods; Secondly, the STM32F446RET6 master control, three-phase bridge inverter circuit and high-precision sampling module are used in the hardware to realize the real-time data interaction between the driver and the upper computer combined with UART; Then, the advantages of EKF in steady-state accuracy, dynamic response and low-speed characteristics are verified by Simulink simulation. The physical debugging shows that the driver can achieve stable operation at 0.1 times the rated speed, and the ability to resist high-frequency noise is improved. The upper computer can monitor and configure parameters online in real time; Finally, the experiment shows that the root mean square error of rotor position estimation is reduced by 25% and the torque ripple suppression is improved by 27% compared with the traditional method, which provides an engineering solution for high-performance PMSM control.