Abstract: With regard to the nonlinear system having multi-variables and sparse data, a multi-variable discrete grey power model is proposed and its solution method is discussed. Considering the great effect of driving factors on the modelling precision, an optimized model is constructed by introducing a driver control function to recognize the main driving factors in driving stages. Subsequently, approaches to identifying the parameters of the driver control function are discussed and the modeling steps are listed. At last, this new proposed model is employed for forecasting China’s grain production, compared with other competing models. This experimental result shows the proposed model can better describe the nonlinear relationships between the system behaviors and the driving factors, thereby improving the accuracy of grey system modelling.