Abstract：Permanent magnet synchronous motor（PMSM）has the advantages of fast dynamic response，high

power density，and high torque at low speed，but the temperature variation and complex working conditions will

cause the variation of PMSM parameters，thus affect motor performance and reduce the output efficiency. To

address the controller parameter mismatch problem caused by the change of motor parameters in the model

predictive current control，firstly an adaptive linear（Adaline）neural network was used for the online identification

of the parameters of the PMSM such as inductance，flux and resistance，and then the normalized least mean square （NLMS）algorithm was introduced to improve the Adaline neural network algorithm in order to improve the

convergence speed and computational accuracy of the algorithm. In addition，the high-frequency current component of the model predictive control was utilized to calculate the PMSM rotor position and the parameters of rotor angle and speed were adopt to achieve sensorless control. The experimental results show that the improved NLMSAdaline neural network is of practical value in terms of speed and accuracy compared with recursive RLS and

traditional Adaline online identification，along with a nice adaptation to parameters mismatching.

      Key words：permanent magnet synchronous motor（PMSM）；parameter online identification；adaptive linear

（Adaline）neural network；normalization；model predictive current contro（l MPCC）