An Advanced Method For Detecting Possible Near Miss Ship Collisions From AIS Data
An Advanced Method For Detecting Possible Near Miss Ship Collisions From AIS Data
复制标题
DOI：
10.1016/j.oceaneng.2016.07.059
复制
发表时间：
2016-09
影响因子：
5
通讯作者：
Weibin Zhang;F. Goerlandt;P. Kujala;Yinhai Wang
中科院分区：
文献类型：
--
作者：
Weibin Zhang;F. Goerlandt;P. Kujala;Yinhai Wang
海上事故有可能造成重大的经济损失、人身伤害和环境破坏。调查海上安全的一种方法是把重点放在“未遂事故”上，即没有导致事故但侥幸避免事故的情况。基于交通冲突技术（通过冲突严重程度等级对交通冲突进行排序）的原理，本文提出了一种新的海上交通数据筛选模型，用于近靶船-船碰撞，特别是在公海和沿海限制海域。与以前的方法相比，所提出的方法具有更大的特异性，在情境化的交通环境中，导航专家评估的可能的近靶情况更少。这是通过船舶域包括船舶尺寸的影响，以及通过与早期提出的模型相比，通过最小碰撞距离概念更好地考虑碰撞方向的临界性来实现的。模型中包含的因素及其关系是基于专家判断和使用以往研究的知识。模型参数来源于参考遭遇情况数据集的AIS数据点。所开发的模型已应用于波罗的海北部的交通数据。该模型进行了一系列有效性测试，结果表明，该模型足以对遭遇进行排序和优先排序，以便在专家判断阶段进行进一步评估，以确定险些脱险的情况。因此，建立了一种方法，使后续研究近靶信息的有效性，以作出与碰撞事故有关的海上安全声明。
Maritime accidents have the potential to cause significant financial loss, injury, and damage to the environment. One approach to investigating maritime safety is to focus on near misses, that is, situations which did not lead to an accident but where an accident was narrowly avoided. Based on the principles of the traffic conflict technique, which ranks traffic encounters through a conflict severity hierarchy, this paper proposes a novel model for screening maritime traffic data for near miss ship-ship encounters, particularly for open sea and coastal restricted sea areas. Compared to previous methods, the proposed method has a greater specificity, leaving fewer possible near miss cases to be assessed by navigational experts in a contextualised traffic setting. This is achieved by including the effect of ship size through a ship domain, and by better accounting for the criticality of the encounter direction through the Minimum Distance To Collision concept compared to earlier proposed models. The factors included in the model and their relation are based on expert judgments and using knowledge from previous studies. Model parameters are derived from AIS data points from a reference encounter situation dataset. The developed model has been applied to traffic data from the Northern Baltic Sea. The model is subjected to a number of validity tests, the results of which suggest that the model is adequate for ranking and prioritizing encounters for further assessment in an expert judgment phase to identify near misses. Thus, it establishes a method to enable subsequent research into the validity of near miss information to make statements of maritime safety in relation to collision accidents.