摘 要

近年来海上运输量日渐增多，海事管理部门的管理压力也越来越大。但受船舶识别技术的限制，在船舶行为的自动识别方面，尤其是船舶的停泊行为识别，依旧存在频繁误报或错报的情况。因此船舶停泊行为的识别研究已经成为了海上交通研究的热点问题。对船舶停泊行为进行深度的研究，不但可以对锚地和码头的使用率进行评估，为日后锚地和码头的重新规划提供必要的数据支持，而且可以更准确地掌握船舶的停泊行为，有助于更加高效的运输，增加公司效益，还可以及时获知船舶的非正常停泊，可以对事故船舶及时进行救助或整治船舶的违法行为，保护船舶及船上人员的生命财产安全。

目前对于船舶停泊行为识别的研究还比较少。传统的船舶停泊行为识别方法是依靠船舶AIS系统上传的船舶动态来进行识别，这种识别方式的准确度依赖船舶主动上传信息的准确性，依赖性高且船舶状态的更新存在滞后性。随着技术的不断发展，出现了不少关于船舶停泊行为识别的方法，但这些方法多是采用密度聚类以及船舶停泊时在速度、时间、位移等方面的特征来识别船舶的停泊轨迹或停泊水域，缺乏对船舶停泊行为特征进行分析和船舶停泊行为种类的识别。

为了细化船舶停泊行为的识别并提高船舶停泊行为识别的精度，减少船舶调度的时间间隔，提高船舶进出港效率以及加强船舶管理。本文提出了一种基于AIS信息的船舶停泊行为识别与分类方法。本文的主要工作包括以下两方面；一方面根据船舶停泊行为的速度特征，利用速度阈值划分船舶停泊行为识别的范围，再用DBSCAN密度聚类对船舶轨迹进行筛选，最后利用速度阈值补全船舶停泊行为轨迹；另一方面，根据船舶在长江下游和海上等船舶停泊受风流作用影响较大的水域停泊时，不同的停泊行为在停泊前轨迹和停泊时轨迹所表现出来的不同特征，识别船舶停泊行为的种类。

在选定的水域内，采用本文提出的船舶停泊行为识别与分类方法对随机的船舶轨迹进行识别，实验结果证明该船舶停泊行为识别与分类方法，可以根据AIS信息有效地识别船舶停泊行为及其种类。

关键词：停泊行为；速度阈值；DBSCAN聚类；行为分类

Abstract

In recent years, the maritime transportation volume has been increasing day by day, and the management pressure of the maritime management department is also increasing. However, due to the limitation of ship identification technology, in the automatic identification of ship behavior, especially the identification of the ship's berthing behavior, there are still frequent false positives or false positives. Therefore, the research on the identification of ship's berthing behavior has become a hot issue in marine traffic research. In-depth research on the ship's berthing behavior can not only evaluate the utilization rate of the anchorage and wharf, provide necessary data support for the future re-planning of the anchorage and wharf, but also can more accurately grasp the ship's berthing behavior, which helps Efficient transportation increases the company's benefits, and it can also be informed of the abnormal berthing of the ship in time. It can help the accident ship in time or rectify the illegal behavior of the ship, and protect the life and property of the ship and the personnel on board.

At present, there are few researches on the identification of ship's berthing behavior. The traditional method for identifying berthing behavior is to rely on the ship's dynamics uploaded by the ship's AIS system for identification. The accuracy of this identification method depends on the accuracy of the ship's active uploading information, which is highly dependent and there is a lag in the update of the ship's status. With the continuous development of technology, there have been many methods for identifying the mooring behavior of ships, but most of these methods use density clustering and the characteristics of speed, time, displacement, etc. when the ship is moored to identify the mooring trajectory or mooring of the ship In the waters, lack of analysis of the characteristics of the ship's berthing behavior and identification of the types of ship's berthing behavior.

In order to refine the identification of the ship's berthing behavior and improve the accuracy of the ship's berthing behavior identification, reduce the time interval of ship scheduling, improve the efficiency of ship entry and exit and strengthen ship management. This paper presents a method for identifying and classifying ship's berthing behavior based on AIS information. The main work of this paper includes the following two aspects; on the one hand, according to the speed characteristics of the ship's berthing behavior, the speed threshold is used to divide the range of the ship's berthing behavior recognition, and then the DBSCAN density clustering is used to filter the ship's trajectory, and the speed threshold is used to complete the ship's berth Behavior trajectory; on the other hand, based on the different characteristics of different berthing behaviors before the berth and trajectory of the berth when the ship is moored in the waters where the ship is moored in the lower Yangtze River and the sea by the wind, the ship's berth is identified Kind of behavior.

In the selected water area, the ship berthing behavior recognition and classification method proposed in this paper is used to identify random ship trajectories. The experimental results prove that the ship berthing behavior recognition and classification method can effectively identify the ship berthing behavior and its berth kind.

Key words：Parking behavior;Speed threshold; DBSCAN clustering; Behavior classification

目 录

第1章 绪论 1

1.1 研究背景及意义 1

1.2 国内外研究现状 2

1.2.1 基于AIS的船舶移动行为识别 2

1.2.2 基于AIS的船舶停泊行为识别 3

1.3 主要研究内容 3

1.4 技术路线 4

第2章 船舶停泊行为识别方法 6

2.1 相关概念 6

2.2 数据选择和处理方法 6

2.2.1 数据选择 6

2.2.2 数据处理 7

2.3 船舶停泊行为识别 8

2.3.1 利用速度阈值划定停泊行为识别范围 8

2.3.2 基于DBSCAN算法识别船舶停泊行为 9

2.3.3 利用速度特征补全船舶停泊轨迹 9

2.3.4 利用速度阈值划定船舶停泊状态 10

2.4 实验与分析 11

2.4.1 试验水域概况 11

2.4.2 划定停泊行为识别范围的实验与分析 12

2.4.3 停泊行为识别的实验与分析 12

2.5 本章小节 20

第3章 船舶停泊行为分类方法 21

3.1 锚泊行为和靠泊行为特征分析 21

3.1.1 锚泊行为特征分析 21

3.1.2 靠泊行为特征分析 22

3.2 船舶停泊行为分类 23

3.2.1 根据停泊前轨迹的分类方法 23

3.2.2 轨迹点集面积计算方法 24

3.2.3 船舶停泊行为分类方法 24

3.2.4 挖掘船舶停泊水域方法 25

3.3 实验与分析 26

3.3.1 轨迹点集面积计算的实验与分析 26

3.3.2 船舶停泊行为分类的实验与分析 30

3.3.3 船舶停泊水域挖掘的实验与分析 32

3.4 本章小节 34

第4章 结论与展望 35

4.1 研究结论 35

4.2 研究展望 35

参考文献 37

致 谢 40

绪论

船舶运输是世界各国之间进行交流和贸易的重要手段。随着经济全球化进程的不断加快，国家之间的贸易往来越来越普遍，海上运输量日渐增多，海事管理部门的管理压力也越来越大。虽然随着船载AIS系统的普及以及全国各地TVS中心的陆续建成，增强了海事管理部门对海上交通的监控力度，降低了海事管理部门的管理压力。但受船舶识别技术的限制，在船舶行为的自动识别方面，尤其是船舶的停泊行为的识别，依旧存在着频繁的误报或错报的情况。因此船舶的停泊行为的识别研究已经成为了海上交通研究的热点问题。下面，首先对本文的选题背景及意义、主要研究内容和技术路线逐一进行说明，并就当前学界关于船舶行为检测的研宄现状进行详细述评。

1.1 研究背景及意义