AI-informed Rapid Post-earthquake Inspection and Evaluation of Civil Infrastructure

In the aftermath of an earthquake, rapid structural inspection and evaluation are critical to ensure restoration of the normal order of life, work, and production. Traditional manual visual assessments by certified inspectors are slow, risky, and subjective, with limited availability delaying inspections. This lecture presents two approaches for automated rapid post-earthquake safety assessment. The first uses sparse acceleration measurements to define damage-sensitive features, inferred through a convolutional neural network. Validated experimentally at E-Defense in Japan, it proves effective for high-rise buildings. The second employs commercial UAV-collected images and a graphics-based digital twin (GBDT), incorporating finite element (FE) and photo-realistic computer graphics (CG) models. This approach is illustrated for a 45-story building in Guangzhou, China. These strategies enable rapid evaluation and efficient decision-making post-earthquake.