Vulnerability Analysis of House Construction on Cut Slopes Based on Coupled Data-Driven and Physics-Based Simulations

To assess the risk of landslide geohazards induced by House Construction on Cut Slopes (HCCS) in mountainous areas, this study used HCCSs in the granitic slope region of Yanling County, Hunan Province, China, as case samples. An intelligent vulnerability assessment framework was developed by coupling multiple machine learning algorithms with SPH–FEM numerical simulations. Field geological investigations indicate that the rear wall of houses built against cut slopes is the primary element at risk. Data-driven analysis further identifies slope-cutting height (SCH) and toe distance of slope (TDS) as the key factors controlling house vulnerability. For houses with SCH exceeding 5 m, the vulnerability level should be upgraded. In addition, the existing houses generally have insufficient TDS, suggesting that a mandatory safety setback should be reserved during construction. SPH–FEM simulations were subsequently performed to reveal the motion–deposition characteristics and energy-transfer mechanisms between the landslide mass and the house rear wall under different TDSs. When TDS ≤ 0.5 m, a large proportion of the landslide kinetic energy is transferred to the wall, resulting in severe structural damage. As TDS increases, the energy absorbed by the wall decreases progressively, and the disaster intensity is significantly reduced. When TDS > 5 m, the wall exhibits negligible deformation, indicating that this distance can be regarded as a safe TDS setback for HCCS.