Aiming at problem that while traditional extended Kalman filter（ EKF） is used to estimate velocity and position of permanent magnet synchronous motor（ PMSM）

the model is not accurate

the estimation precision is not high and the real-time performance is poor when the noise is uncertain

and it may lead to filtering divergence

a Sage-Husa adaptive fading extended Kalman filtering（ AFEKF） algorithm is adopted. In order to improve the robustness of model errors and reduce overshoot

a new type of sliding-mode speed controller（ SMC） based on exponential reaching law is adopted. The experimental result shows that

the proposed control strategy can estimate the rotating speed and position accurately compared to the traditional proportional integral（ PI） and EKF algorithms. At the start time

it can reach the preset speed quickly

and has no overshoot phenomenon

and quickly reaches the stable state; the maximum deviation after loading is reduced by 1. 77 % compared with traditional algorithm

and the rotating speed error is decreased by 0. 371 % in steady state

and the steady-state position error is decreased by 0. 45 %. It is proved that the algorithm is stable and practical in PMSM sensorless control system.