Multirate UKF Damage Identification Based on Computer Vision Monitoring of Ship–Bridge Collisions

Abstract

When a ship–bridge collision occurs, prompt assessment of substructure damage is crucial. This study presents a novel approach for ship–bridge collision damage identification, addressing challenges inherent in traditional monitoring systems. The method overcomes issues such as complex installation, low efficiency, and high costs through a unique combination of the unscented Kalman filter (UKF) and computer vision technique. The approach exerts the structural equation of motion to derive a multirate UKF in the impact process, thereby identifying the stiffness of structures. Displacement and acceleration are fused to enhance the sampling rate of vision-measured displacement. Firstly, it monitors low sampling rate displacements on piers using computer vision, complemented by high-rate accelerometer data at the collision point. Secondly, displacement and acceleration data are integrated using a multirate UKF, addressing the challenge of image storage pressure associated with vision-based measurements. Finally, validation using finite-element and experimental models confirms the effectiveness of the approach in identifying substructure stiffness and recovering lost vibration characteristics. In experiment validation, the influence of computer vision algorithms and camera shooting distance on displacement monitoring and stiffness identification is also discussed separately. This approach provides a cost-effective and efficient solution for ship–bridge collision damage identification, contributing to advancements in the field of ship–bridge collision monitoring.