In order to research the topological structure of parking network, a new method of parking network connection based on attractive relationship is proposed. By improving the existing parking demand forecasting model and considering the factors such as parking utilization, parking turnover, and urban center index, a new parking demand algorithm is established. Drawing a conventional Voronoi diagram and using the key properties of the fracture point theory and physical parameters of the conventional Voronoi diagram to divide the influence range of each parking lot, the concept of attractiveness is introduced, and the parameters in the formula of attractiveness calculation are modified and calibrated. The degree of attraction of each node in the parking lot network and the edge weight of the connection between nodes were calculated. Then the connection situation between parking lots was determined. In this paper, the model of parking lot network is constructed under the attraction relationship of some parking lots in Harbin. The complex network parameter analysis method is adopted to analyze the parking lot network. The macro and micro node parameters of parking lot under the scale assignment calculation method and further calculation through the attraction relation are analyzed. Results show that the parking lot network degree distribution conforms to the power law distribution, and the scale-free network features are obvious. Compared with the global coupled network and the unilateral coupled network, the parking lot network under the attraction relationship has a high clustering coefficient and global effectiveness and a small average path length. This method is a practical mode whose performance exhibits equilibrium under the existing parameter evaluation system.