Abstract: This paper uses geographic distance and transportation convenience to test its impact on collaborative innovation. The spatial distance is included in the dynamic optimization model to analyze the mechanism of university-enterprise cooperation. Different from using the distance between cities which has commonly been seen in existing related research, this paper examines the impact of distance between a university and an enterprise on their collaboration by mining the geographic information of universities and enterprises and collecting micro-data on patents cooperated by universities and enterprises. And the distance has a significant negative impact on the probability and frequency of cooperation on patents between university and enterprise. The heterogeneity of enterprises and universities also leads to differences in the impact of distance: the influence of distance is more significant in the cooperation between smaller enterprises and universities, in the cooperation between enterprises and "985" universities, and in the cooperation in industries with higher substitution elasities. Further analysis shows that the opening of the high-speed railway and subway has increased the probability and frequency of university-enterprise cooperation, and decreased the negative impact of spatial distance on university-enterprise cooperation. The discussion of robustness testing and endogeneity issues confirms the robustness of the results. The conclusion of the article has certain significance for further improving university-enterprise cooperation.