Assessing Urban Waterlogging Risk for Built Environment Resilience: An Interpretable HEVR–ICWGT–Cloud Assessment of Drainage and Flood Control Infrastructure Deficits in Eastern Henan

Urban waterlogging is increasingly a test of built environment resilience; cities with similar rainfall can suffer very different losses depending on their drainage and flood control infrastructure. This study assesses the waterlogging risk for Kaifeng, Shangqiu, and Zhoukou (Eastern Henan, China) over 2015–2024 using a 15-indicator Hazard–Environment–Vulnerability–Resistance (HEVR) system, in which resistance indicators are reverse-coded so that a higher resistance expectation (R − Ex) denotes a larger capacity deficit. Subjective AHP weights and objective entropy weights are reconciled by an improved game-theoretic combination method (ICWGT) with reliability-proportional coefficients (0.516:0.484), and cloud model aggregation reports both risk position and classification dispersion. The mean risk gradient is Shangqiu (Ex = 5.252) > Kaifeng (5.023) > Zhoukou (3.797); high-risk years occur only in Kaifeng (2015 and 2018) and Shangqiu (2018). Contribution decomposition identifies Kaifeng as the largest resistance-deficit case (R − Ex = 7.174; 37.6% of its total contribution), and first-order screening indicates that closing this deficit to Shangqiu’s level would lower its mean Ex by 0.388. Machine learning classifiers serve strictly as a consistency check (SVM-RBF reproduces 83.3% of labels), supporting internal learnability rather than independent validation. The framework provides an interpretable way to prioritize drainage and flood control investment in the built environment.