According to specificity of the complex system process with stochastic uncertainty and correlation, a kind of predictive method based on gray system theories is presented. A nonlinear and multi-variable predictive model approach is established by the Taylor series expansion extended the single deformation prediction model in order to estimate more. The effectiveness of the model is confirmed through a computational experience, and the feasibility of the solutions is stipulated at the same time. Thus this predictive control strategy introduces a new way for generic nonlinear control.