Abstract: The hysteresis modeling of nonlinear system is studied in this paper. A new sorting & taxis model of hysteresis is realized using neural network to describe the hysteresis of the piezoceramic actuator. Taking both the wiping-out property and the precision into consideration, a BP neural networks is proposed to solve function F. In this way the error resulting from interpolation is avoided. The modeling experiment of the piezoelectric ceramic is realized by the network. The results of experiments prove that the sorting & taxis scheme using neural network effectively improves the precision and reduces the error of the model.