Vulnerability Assessment of Road Networks in Valley Cities Considering Spatial Heterogeneity

The road network of valley cities is constrained by mountains and rivers, developing along main river channels with a cluster structure. While internal roads within clusters are well connected, inter-cluster connectivity is weak, creating significant spatial heterogeneity. Existing vulnerability assessments overlook these topological traits, limiting their ability to identify critical nodes. This study proposes a methodology that quantifies spatial heterogeneity using Relational Graph Convolutional Network (R-GCN) models. The road network is decomposed into subnetworks with similar structures, forming a weighted heterogeneous network. Two attack strategies assess vulnerability and identify key nodes, offering new insights into road network resilience. This study provides a new insight into vulnerability assessment of road networks from the perspective of topological signatures of intersection connections in…