In view of the deficiency of the extended Kalman filter (EKF) in determining the appropriate system noise matrix Q and measurement noise matrix R in parameter identification of permanent magnet synchronous motor (PMSM), a PMSM parameter identification method has thus been proposed based on improved self adaptive differential evolution (SADE) -EKF. Firstly, based on an analysis of the working principle of the extended Kalman filter, the double thread identification model has thus been established. Then, the mutation strategy of the improved differential evolution algorithm (DE) is adopted to jump out of the local optimum, followed by a design of a suitable fitness function. Finally, the Q and R of EKF are optimized by using Sade algorithm. The experimental results show that the improved SADE-EKF is characterized with a better convergence speed and identification accuracy than the traditional EKF in identifying motor parameters.