In order to improve the modeling accuracy of piezoelectric ultra precision positioning stage, the hysteresis model of BacklashLike is used to describe the hysteresis characteristics of piezoelectric ultra precision positioning stage. The recursive least squares method based on genetic algorithm is utilized to identify the parameters of BacklashLike model. Combined with the dynamic characteristics of the piezoelectric ultra precision positioning stage, the secondorder dynamic hysteresis model is established for the piezoelectric ultra precision positioning stage. The experimental results show that, compared with the BacklashLike model under frequencies of 30 and 40 Hz, the maximum output displacement error of the dynamic hysteresis model decreases from 1.21 and 1.39 μm to 0.32 and 0.44 μm, and the maximum relative error are 3.5% and 4.4%. The average displacement error decreases from 0.53 and 0.76 μm to 0.17 and 0.21 μm, the average relative error decreases from 1.93% and 3.38% to 1.11% and 1.37%. The experimental results verify the dynamic hysteresis model and prove that the proposed model can not only reduce the system error caused by the dynamic characteristics of the piezoelectric ultra precision positioning stage, but also simulate the hysteresis characteristic and dynamic characteristic, and avoid the repeated model parameters identification under different frequencies. The accuracy of piezoelectric ultra precision positioning stage is improved in high frequency, fast, and large stroke location. This method is simple and has strong adaptability, easy to implement in engineering.