Effective vibration length identification and cable tension estimation of stay cables based on UAV video: Integrating multi-order mode-shape consistency constraints and SSD-CEEMDAN drift separation

Cable tension is a key indicator for safety assessment and maintenance decision-making of cable-supported bridges. Effective vibration length identification is essential for improving the accuracy of vibration-based cable tension estimation; however, existing methods rely on multi-point accelerometer measurements, resulting in high deployment costs and low field efficiency. To address this issue, this study proposes a UAV-photography-based framework for effective vibration length identification and high-accuracy cable tension estimation, establishing a complete procedure consisting of video acquisition, cable segmentation, vibration extraction, UAV drift suppression, multi-order mode-shape-consistency-constrained fitting, effective vibration length identification, and cable tension calculation and validation. First, a semantic segmentation model is used to extract the cable and marker regions, and the marker centroids are then tracked to generate multi-point displacement time histories. Subsequently, a spectral-morphology-driven adaptive drift separation method, namely SSD–CEEMDAN, is adopted to suppress UAV drift interference while enhancing the distinguishability of dominant spectral peaks. Under the condition that only amplitude information is available, a joint least-squares model constrained by multi-order mode-shape consistency is constructed to identify the effective vibration length of the cable. Field tests on four cables of an actual bridge show that the relative errors of cable tension calculation using the identified effective vibration length are 1.69%, 1.07%, 3.70%, and 2.85%, respectively. Compared with the results obtained using the geometric cable length, the errors were reduced by 7.36, 6.79, 9.18, and 5.88 percentage points, respectively. The results verify the necessity of effective vibration length identification and the engineering feasibility of UAV-based non-contact measurement, providing a new approach for rapid, efficient, and high-accuracy monitoring of bridge cable tension.