This invention discloses a quantitative discrimination method for post-earthquake RC frame
structure collapse modes, belonging to the field of
earthquake resistance and disaster reduction technology in
civil engineering. First,
raw data such as color images, depth data, and point clouds of post-earthquake ruins are acquired. Instance-level identification and
visibility determination are performed on load-bearing components to distinguish between
occlusion boundaries and physical fracture boundaries. Then, geometric features such as the three-dimensional position, tilt angle, and residual length of the components are extracted to construct topological relationships such as beam-column nodes, support contacts, and facade meshes. Subsequently, geometric and topological parameters such as overall tilt angle, residual height ratio, cavity scale, and mesh continuity are calculated. Finally, a hierarchical discrimination conclusion including the
impact range, damage degree, cavity development, and
structural integrity is output. This invention can achieve accurate mapping from component-level damage to the overall
collapse mode, effectively
handle occlusion and missing scenarios, and provide intelligent and quantifiable
technical support for post-earthquake
emergency rescue, survival space analysis, and secondary collapse
risk assessment.