시스템식별과 영상기반 군중하중 추정을 통한 공연장의 동적응답 및 사용성 평가

Abstract

In this study, vision-based monitoring combined with accelerometer-based measurements was conducted in a large-scale concert hall under crowd loads. First, the dynamic characteristics of the structure were identified from ambient vibrations measured in the absence of crowds. Then, an FE model was established that exhibits dynamic attributes similar to those of the structure. Subsequently, camera footage of crowd motion at a K-pop concert was used to estimate crowd loads, which were then applied to the FE model to assess responses. The acceleration response obtained from the analysis was compared with the measured acceleration and the acceptance criteria of design guidelines, such as AISC DG11. The acceleration predicted by the vision-based estimation was approximately 67% of the estimated value based on peak values and 62% of the measured value based on root mean square (RMS) values. Overall, the vision-based estimation showed a peak response around 10 Hz, while the measurement showed a gradual increase in the frequency range above 10 Hz. This is likely because low-frequency components were underrepresented in the vision-based estimation, resulting in relatively underestimated amplitudes and lower responses in the high-frequency range.