Abstract: The hysteresis nonlinear characteristic of the nanometer positioning system based on piezoceramic actuator decreases the accuracy of the nanometer positioning stage badly. In order to compensate for the hysteresis nonlinearity and improve the precision of system with hysteresis, this paper studies the inverse hysteresis modeling of the piezoceramic actuator. A dynamic Preisach inverse model is built based on the asymmetric exponential function hysteresis operator, and identified by the neural networks. Several groups of experiment data are used to verify the accuracy of the promoted inverse model. The experiment result shows that voltage deviation between the neural networks which is added in historical input displacement and the given voltage are reduced. The max deviation from the original 16V is reduced to no more than 6V. Performance has improved obviously.