Vessel incidents periodically occur in the waterways of the United States, but some types of commercial vessels have shown a downward trend in the number of incidents in recent years. One of the missions of the United States Coast Guard (USCG) is to develop regulations to mitigate and potentially prevent maritime incidents. In this paper, the USCG gathered data on more than 117,000 incidents that involved U.S.-flag vessels in U.S. waterways for the period 2001 through 2018. We applied the Grey System theory or model and Grey-Markov forecasting model to predict the future number of vessel incidents for four different vessel types from 2019 through 2030. Incident data can vary considerably from year to year and often can be incomplete. The Grey-Markov model, which is a combination of the Grey model and the Markov chain process, is suitable for this purpose because of its predictive ability. From our results, we found that the Grey-Markov model performed exceptionally well and showed the predicted values of the number of incidents to be remarkably similar to the actual values with acceptable mean relative errors ranging from 5.2% to 8.2%. We expect that these results will benefit decision makers in formulating sound policies thereby improving the maritime safety of vessels operating in waterways of the United States.

在美国的水道中定期发生船舶事故，但近年来某些类型的商船事故呈下降趋势。美国海岸警卫队（USCG）的任务之一是制定法规，以减轻和潜在地预防海上事件。在本文中，USCG收集了2001年至2018年期间在美国水道中涉及美国国旗船舶的117,000多起事件的数据。我们使用灰色系统理论或模型以及Grey-Markov预测模型来预测未来的船舶数量从2019年到2030年，四种不同类型的船舶发生事故。每年的事故数据可能会有很大不同，并且通常是不完整的。Grey-Markov模型，是Grey模型和Markov链过程的组合，由于具有预测能力，因此适用于此目的。从我们的结果中，我们发现Grey-Markov模型的表现异常出色，并表明事件数量的预测值与实际值非常相似，平均相对误差在5.2％至8.2％的范围内。我们希望这些结果将有利于决策者制定合理的政策，从而改善在美国水道内航行的船舶的海上安全。