| Spatial interaction is the connection of cities and regions over space, it can be broadly defined as movements of people, capital, commodities or/and information over geographical space resulted from a decision process. If there is movement of some element, it's sure that some kind of spatial interaction exists. Spatial interaction has developed for nearly a century, through which spatial interaction has formed its own frame of theory system. The paper analyzed the basic theory of spatial interaction, including its meaning, classification and conditions. The paper placed much emphasis on the spatial interaction models, discussed the principles, methods and differences between models. From the classic gravity model to neural spatial interaction model, spatial interaction modeling as well as its application has been greatly improved.Based on the understanding of spatial interaction theory and its models, the paper then focused on the empirical research. According to the relevant statistic data of Jiading district, the author proposed some assumption and then obtained the data of trans-town public transportation flow. Using the theory of artificial neural network, the author established the neural spatial interaction model of Jiading district, and then calculated the potential value of each town of Jiading district. Combined with GIS, the paper also discussed the attraction of each town and explained the factors that influence the attraction, aimed at helping to make better and more effective decisions for Jiading district.In the finality, the author gave the conclusion of the paper, and the problems requiring further studies are discussed. |