The dynamic Gray Radial Basis Function (DGRBF) prediction model is the improvement of traditional dynamic grey prediction model GM (1,1). It gives the dynamic algorithm of acquiring optimized initial conditions and identifying parameters like metabolism, and then the model combines the characters of RBF Neural Networks, and therefore has the ability of dynamic prediction on small volume of samples. The DGRBF prediction model has been applied in the practical engineering successfully, and the experiment results demonstrate that model and its intelligent optimizing algorithm are capable of predicting in a long term and the desired data could be acquired accurately, easily and conveniently.