Economy increasingly depends upon resources, which makes policy simulation a hot research topic because of policy and its execution risks in resources and environmental economics. This paper reviews the worldwide policy simulations, and analyzes the references from CGE model, systematic dynamics and agent-based modeling. These simulations are reparation each other in precondition, modeling, and data dependence and forecast accuracy. An appropriate simulation shall be in accordance with research objects and available data. Due to the complexity of policy and factors during making policy, the integration of these simulations becomes a new trend.

To study and predict China’s petroleum consumption helps improve rationality in petroleum production, import and consumption. This paper adopts BP neural network prediction to establish time-neural network model, which is applied to predict the petroleum consumption quantity in 2010—2012 based on two different petroleum consumption quantity. This model has been proven to be workable. The prediction results acquired from latest data are closer to the growing trend of petroleum consumption, providing references for petroleum demand planning.