Computer vision-based reliability analysis in digital twinning for structural dynamic safety assessment

The emergence of digital twinning (DT) technology provides a superior platform for representing buildings and their assets by linking physical form with their digital counterparts. While the development history of DT in the construction industry is relatively short, its application potential is promising. This study proposes a computer vision (CV)-based digital twinning assessment method for structural reliability analysis, with a special focus on data processing and transmission for vision-based measurements. Firstly, traditional sensor-based structural motion estimation in the DT process is replaced by video measurements, which are used to determine the interlayer displacement of structures under excitation. Then, based on the structural dynamic equation, the vision-based measurements are input into a finite element (FE) model for structural analysis. Finally, the reliability assessment is performed based on the FE analysis results, and a unified visualization platform is developed using the BIM engine. To demonstrate the developed approach, a structural dynamic analysis for a three-storey steel frame is conducted. The results are compared with test results in terms of interlayer displacement and section stress. The findings indicate that the computer vision-based digital twinning method is accurate and feasible for structural safety analysis.