An optimizing model for B-spline networks prediction has been proposed for the effective realization of the real-time prediction based on nonlinear time series. A global optimization algorithm has been adopted for the optimization of the design of the network structure, with the weight parameters and node location of spline functions synchronous independent variables to be optimized. The industrial simulation results show that the proposed optimized model for networks prediction has a relatively higher prediction accuracy. With its structure much simpler and less parameters to be optimized, this is an effective prediction method based on nonlinear time series.