Akademik Veri Yönetim Sistemi
Structures, cilt.69, 2024 (SCI-Expanded)
This study introduces a novel nondestructive approach to Structural Health Monitoring (SHM) using computer vision and optical flow methods to analyze structural vibrations. It combines advanced image processing techniques, like the Lucas-Kanade Optical Flow method, with spectral analysis tools including the Autoregressive Moving Average (ARMA) model and Enhanced Frequency Domain Decomposition (EFDD) for assessing structural integrity. The research comprises two main components: (i) the development of a vibration monitoring system with industrial cameras and open-source image processing techniques, and (ii) the application of specialized image processing software. Key aspects of the study include the use of displacement sensors with template matching for laboratory and field measurements, a comprehensive vision sensor system with high-grade hardware, and camera calibration to correlate camera images with real-world measurements. The methodology focuses on target tracking through optical flow estimation which is crucial in calculating displacements and analyzing structural movements. Experimental verification was conducted on three models such as a single-story steel planar frame, a single-story space shear frame, and a real-scale steel footbridge. The results show good agreement between the frequencies obtained from video camera-based measurements and those retrieved from numerical models, which validates the experimental approach's reliability. The study emphasizes the potential of computer vision in SHM especially for inaccessible or restricted structures. It highlights the cost-effectiveness and efficiency of the considered non-contact methods. Overall, the research demonstrates significant advancements in remote, nondestructive structural analysis, offering promising implications for predictive maintenance and the safety of engineering structures.