A novel method that combines neural net with predictive control is proposed in this paper. Firstly, the neural net is employed to construct the nonlinear systems, the original nonlinear systems are wrapped by the linear timevarying systems described using the polyhedron which is formatted by the maximum or minimum values of the transition function in the neural net. Then,the multiple parametric programming based on combination of the on-line and off-line is utilized to obtain the explicit model predictive control systems. Finally,an example is presented to illustrate the effectiveness of our method. The results show that the developed method can describe the nonlinear system effectively, and improve the control performance.