Traditional position sensorless control systems generally adopt PI regulator for position information processing. A new adaptive Luenberger observer is proposed to solve the drawbacks of PI regulator such as parameter setting, poor tracking performance and weak interference suppression ability. This method uses pulsating high frequency current injection method to obtain the high frequency pulse vibration position signal, and then establishes Luenberger observer for speed and load disturbances according to the dynamic equation of the motor. Neural network is adopted to build the autotuning parameters controller for replacing the PID controller in Luenberger observer. The control of permanent magnet linear synchronous motor (PMLSM) without position sensor is realized. The simulation results show that the maximum error of velocity estimation is 2×10^-3m/s，and the maximum error of position estimation is -3×10^-5rad. PMLSM control system based on the adaptive Luenberger observer has good tracking performance and antiinterference performance.