Underground transportation infrastructure in cities with dense river networks is at risk of waterlogging during extreme rainfall. To evaluate the flood vulnerability and failure mechanisms of urban tunnels in disaster scenarios, a case study is conducted on the tunnel network in Suzhou, China, and a dual-weighted model of the road tunnel network, incorporating spatial geographical attributes and traffic flow characteristics, is constructed using complex network theory and geographic information system technology. Critical inundation criteria and a node failure simulation method are established to quantitatively analyze the dynamic attenuation law of the operational efficiency of urban tunnel networks under extreme rainfall conditions. The results show that: (1) the road tunnel network in Suzhou displays small-world characteristics; (2) during the initial phase of extreme rainfall, operational efficiency drops by more than 20% before the water level reaches 3 m. The simultaneous failure of key low-lying tunnel nodes, such as the Chengbei Expressway West Tunnel and Xinggang Street Tunnel, primarily contributes to this decline. As the water level rises further, the degradation of network performance stabilizes due to elevation differences at tunnel entrances and exits. To effectively interrupt the failure chain, targeted road network resilience enhancement strategies are proposed based on an analysis of topological vulnerability and spatial heterogeneity of network nodes. These strategies include the differentiated and tiered deployment of flood control facilities, multidimensional optimization of underground drainage systems, and the construction of an integrated intelligent water level monitoring and early warning platform.