A monocular vision-based targetless modal identification method for membrane structures

Abstract On-orbit modal identification of space membrane structures is challenging due to difficulties in sensor placement and potential obstruction of observation targets. To this end, this paper proposes a monocular vision-based targetless modal identification method using on-orbit monocular vibration videos to identify dynamic parameters. The method leverages the pyramid Lucas–Kanade optical flow algorithm with feature point relocation to track feature positions and capture structural vibration without sensors or observation targets. A biaxial tensioned membrane modal test system, utilizing a monocular camera and running the improved optical flow algorithm, is designed to validate the approach. Pixel-level displacement time-history curves are obtained, and modal identification is performed using fast Fourier transform and subspace state space identification methods. Laser displacement sensor array data serves as ground truth for verification. The results show that the method accurately identifies the structure’s frequency and mode shape, with high robustness in frequency identification and sensitivity to shooting angle for mode shape identification. This research offers key technical insights for on-orbit modal identification of space membrane structures.