In order to accurately predict the ownership of CNG automobile, Grey System Theory GM Model and Elastic Coefficient Method were investigated to and compared in terms of their prediction errors. . A new Combined Model was established based on the Weight Factor, which is calculated by using Suboptimal Combination Method and the Linear Combination Model. Compared to the Single Model, the prediction accuracy of the new Combined Model is greatly improved when it was applied to predict the CNG automobile ownership in S province and the errors achieved are no more than 10% than before Thus the study demonstrated that the new Combined Model was effective and practical in its application on the rational layout of the gas stations.