基于视觉多点响应的结构模态识别与振型拼接方法OA

Traditional contact-type sensors suffer from issues such as sparse measuring points and difficulty in obtaining global mode shapes in large-scale structural health monitoring.Single vision devices are prone to mode shape fragmentation phenomenon due to the limited field of view.To solve these problems,a structural modal identification and mode shape stitching method based on visual multi-point responses was proposed.The template matching algorithm based on the zero-mean normalized cross-correlation criterion and the greyscale centroid algorithm were adopted to identify structural vibration displacements.Meanwhile,an inte-grated edge-computing visual monitoring device was developed to realize localized data processing.Through sub-regional measurement,overlapping point selection,amplitude consistency correction and global mode shape reconstruction,the complete stitching of local mode shapes in multiple regions was achieved.The re-sults show that in the clamped-clamped steel beam experiments,the first four orders of modal frequencies ex-tracted by visual monitoring are 10.22,27.46,55.35 and 87.86 Hz,respectively,and all of the relative er-rors between these results and those measured by triaxial sensors are less than 2%.The modal assurance crite-rion values range from 0.840 to 0.997,indicating that the dense-point mode shapes after stitching are highly consistent with those measured by sensors in terms of overall shape.