Geographic knowledge graph-guided twin modeling method for complex city scene

• A three-domain association model for dynamic geospatial knowledge graph. • Designed multi-level fusion mechanism to enable dynamic fusion modeling. • Geographic knowledge graph-guided modeling offers significant advantages over traditional methods in efficiency and updates. The informatization of complex city operations and management is a key trend, and the digital twin (DT) is an effective way of achieving this. However, existing methods for constructing city twin scenes are highly complex and inefficient, making it difficult to meet the efficient construction of digital twin scenes for complex cities. To address this problem, this paper proposes a geographic knowledge graph (GeoKG)-guided twin modeling method for complex city scenes. First, the characteristics and relationships of city scene modeling objects were explored, and a dynamic data-driven algorithm was designed for constructing a GeoKG. Next, the spatial, attribute, and scale relationships within city models were analyzed. From the perspectives of object, attribute, and scene modeling, a multi-level dynamic fusion mechanism was constructed, leading to the proposal of a dynamic fusion modeling method of complex city scene guided by GeoKG. Based on this method, a prototype system was developed for an urban park, and experimental analysis was conducted. The results show that the method proposed in this paper can achieve complex city scene twin modeling, with average fusion modeling times of 1.2, 1.8, 3.0, and 4.1 s for desktop, PC, tablet, and mobile platforms, respectively. The dynamic rendering efficiency remained above 85, 65, 54, and 38 FPS for these platforms, respectively. Compared to traditional methods, the proposed method demonstrates significant advantages in modeling efficiency and dynamic updating, supporting the transition of city operations and management towards refinement and informatization.