The traditional permanent magnet synchronous motor vector control needs to obtain the position information of the rotor in real time. The common rotor position sensor has the problems of increasing system volume, high cost, and being susceptible to interference, and it is difficult to operate normally in a harsh external environment. Based on the mathematical model of permanent magnet synchronous motor and the vector control method of rotor field orientation, this paper studies the current speed sensorless technology and proposes a model reference adaptive system (MRAS) in synchronous rotating coordinate system. A method of estimating the rotor position. The stator voltage equation of permanent magnet synchronous motor with rotor position parameters is used as the adjustable model, and the permanent magnet synchronous motor itself is used as the reference model. The adaptive rate is designed based on the theory of super stability and positive dynamics. At the same time, the steady-state stator voltage equation is used to feed forward the dq axis voltage, which improves the system convergence due to the cross-feedback relationship between the two components of the dq axis current. Finally, the new MARS sensorless control is simulated and analyzed, and the core algorithm code of DSP is generated directly in Simulink by using MBD method for experimental verification. The results show that this method can estimate the rotor position well and can replace the position sensor.