Advanced Early Warning System for Ship–Bridge Collisions Using Multisource Data Fusion and Maneuverability Assessment

Abstract

As ship sizes continue to increase and traffic density rises, the complexity of inland navigation environments is further exacerbated by bridge construction, leading to a heightened risk of ship–bridge collisions. To address this issue and enhance navigation safety in bridge areas, this paper proposes a novel ship–bridge collision prevention warning system. The system is based on the extraction of ship target feature information using the YOLOv5 object detection algorithm and the DeepSort object tracking algorithm. By integrating visual data and Automatic Identification System (AIS) data through advanced data association techniques, the system accurately perceives the maritime situation around bridges, providing real-time situational awareness. Additionally, a responsive ship motion model is developed to evaluate ship maneuverability, offering critical insights into the potential for collision and supporting timely intervention. Experimental validation of the system is conducted through a case study of the Baijusi Yangtze River Bridge. The results demonstrate the system’s efficacy in providing reliable early warnings, especially in scenarios with multiple ships navigating through bridge areas. Specifically, the proposed system significantly improves the prediction of collision risks, ensuring timely and effective interventions. This research makes a substantial contribution to shore-based ship–bridge collision prevention technology by considering the dynamic behavior of ships in bridge zones. The integration of real-time AIS and visual data, coupled with advanced tracking and modeling techniques, provides a robust foundation for enhancing the safety of both ships and bridges in increasingly congested inland waterways.