Zhenjiang Maritime Safety Administration's intelligent system monitors 8 types of navigation risks around the clock.

Recently, the Long Voyage System Artificial Intelligence Application Skills Competition and the First Youth Innovation and Creativity Competition concluded in Wuhan. Two projects submitted by Zhenjiang Maritime Safety Administration, "AI Empowered VTS Duty Listening, Speaking, Reading, and Writing" and "Predicting Crossing, Intelligent Sentinel — Ship Crossing Control Intelligent Agent," won second place. Since the deployment of the Ship Crossing Control Intelligent Agent, more than 1,400 daily navigation warnings have been triggered in the Zhenjiang section of the Yangtze River, with an effective warning accuracy rate consistently above 85%. The effective reminders of high-risk behaviors in VTS increased from over 700 per day to over 2,100 per day.

The Zhenjiang section of the Yangtze River has a large volume of ship traffic with frequent crossing interactions. Under the traditional VTS supervision model, duty personnel must monitor multiple key water areas simultaneously. To improve assessment efficiency, the Zhenjiang Maritime Safety Administration integrates radar, Automatic Identification System (AIS), electronic fences, manual reporting, and other information, overlays geographic data such as channel boundaries, shallow waters, and restricted crossing areas, and develops a ship traffic management intelligent agent for continuous monitoring and parallel analysis of vessel dynamics in the jurisdiction. Currently, this intelligent agent can automatically identify eight types of navigation risks, including crossing, collision, grounding, and overtaking. This means that regulatory attention is no longer limited to a few key water areas being monitored by duty officers. The system helps personnel promptly identify abnormal behaviors that may be overlooked, ensuring broader risk identification coverage and faster responses.

However, sounding an alarm does not mean the regulatory work is complete. Determining whether the warning is accurate, the severity of the risk, and who should handle it requires further judgment. To address this, the Zhenjiang Maritime Safety Administration has introduced a large AI model to the intelligent agent. After issuing a warning, the system automatically retrieves information such as ship trajectories, nearby ship traffic scenarios, and navigation environments to conduct an initial round of intelligent assessment, aiming to reduce false alarms caused by normal ship traffic activities like docking and departure. Post-assessment risk-level handling involves: major risks being immediately pushed to the traffic control center, risks requiring further checks assigned to the rapid response center for follow-up, and minor risks sent via text message for crew self-correction. The system also continuously tracks vessel dynamics post-warning, adjusts the warning levels based on risk changes, and keeps records for each warning, forming a closed-loop process covering warning issuance, assessment, distribution, handling, and tracking.

Currently, the Zhenjiang Maritime Safety Administration is continually exploring the deep integration of artificial intelligence with VTS operations. The plan is to further refine the full-process intelligent monitoring model of "Listening, Speaking, Reading, and Writing," enabling intelligent technology to better serve dynamic vessel regulation and navigation risk prevention tasks, thereby supporting the safety and smooth operation of the Yangtze River Golden Waterway.