基于贝叶斯网络的船舶引航风险预警

Risk Warning for Ship Pilotage Based on Bayesian Networks

船舶引航是水上交通运输的重要环节,它对于船舶安全进表出港、港口正常作业、环境保护和提升国家港口形象等方面具有重要意义。船舶引航是一个复杂的过程,它与人、船舶和环境等各种因素密切相关,研究各种因素对引航安全的影响以及这些因素之间的相互关系,动态识别船舶引航风险,有利于及时采取措施,确保船舶安全进出港。本文提出一种用于船舶引航风险预警的贝叶斯网络模型,通过文献调研和对专家进行深度访谈,利用专家知识和文献资料的信息确定网络的拓扑结构和相关参数。用SamIam软件建模并对一些船舶引航案例进行分析,结果表明本文的贝叶斯网络能够对船舶引航过程的相关风险做出正确的预警,具有实际应用的价值。

Ship pilotage is an important process of marine traffic transportation, as it is critical to in-and-out port ships, port operations, environmental protection and the image of national port. The pilotage is a complicated process, which is involved with human, vessel and environment. Investigating the elements that influence the safety of ship pilotage, and the relationship among these elements will help to take appropriate measures for safety pilotage. In this study, we developed a Bayesian net-work for pilotage risk warning based on literature investigation and deep interview with experts. The structure and parameters of the network were determined from expert knowledge. The vali-dation experiments were conducted on SamIam and a dozen of real pilotage cases were used to test our network. The experimental results show that the proposed network can correctly predict the risk in case of dangers and is promising for practical application.