Structure modal identification based on computer vision technology

Abstract

Mobile phones have the potential to become useful tool in structural modal identification. In this paper, shaking table test videos of a 10-story steel structure captured by mobile phone is processed using computer vision theory and then the modal parameters are identified. A signal processing method based on variational mode decomposition (VMD) is used to improve the accuracy of identification. Using optical flow algorithm, the vibration data is extracted from the video, and then the response of the structure is obtained from the vibration data of selected feature points. Then, the vibration data is processed by VMD and structural modal parameters (mode frequency and mode shapes) are identified using FFD. Finally, the identification results obtained from mobile phone and professional sensors are compared to verify feasibility and accuracy of the proposed modal identification method.