Abstract: Abstract: Kalman filter algorithm can effectively cope with the effect of motor parameters in the anti pinch power window system. However, the observation noise, often unknown and changing in the anti pinch power window system, has a serious influence on the performance of Kalman filter. By means of the feature of the wavelet transform separating the signal and the noise, this paper proposes a new anti pinch window lifter system based on the real time estimation of measurement noise. The simulation results in Matlab show that the proposed algorithm can effectively achieve real time estimation of observation noise, and attain better control performance under various noises.