基于概率有向图模型的船舶进出港航路规划  

Route planning for ship entering and leaving port based on probability directed graph model

作　　者：唐皇[1] 魏立桥 尹勇[1] 漆映宏 TANG Huang;WEI Liqiao;YIN Yong;QI Yinghong(Navigation College,Dalian Maritime University,Dalian 116026,Liaoning China;Business Administration College,Chongqing Three Gorges University,Chongqing 404020,China)

机构地区：[1]大连海事大学航海学院,辽宁大连116026 [2]重庆三峡学院工商管理学院,重庆404020

出　　处：《华中科技大学学报（自然科学版）》2020年第5期105-109,共5页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：工信部智能船舶综合测试与验证研究资助项目(2018-473);中央高校基本科研业务费专项资金资助项目(3132019312)。

摘　　要：为降低港口水域内的船舶航行风险,提高港口生产效率,提出基于概率有向图模型的船舶进出港航路规划算法.采用基于密度的聚类方法,对船舶进出港的历史轨迹进行分类,从而提取出船舶在港内的正常航行模式,并采用基于网格划分的方法对轨迹进行离散.以网格为节点及相邻网格为有向边,建立概率有向图模型.给定航路的起始点和目的地,动态搜索船舶可能的航行路径进行航路规划.分别选取100条进港轨迹和100条出港轨迹对模型进行验证,结果表明:模型输出的进港规划航路与船舶真实航迹间的平均航路偏差为0.18,出港规划航路与真实航迹间的航路偏差为0.28.本模型生成的航路结合了多数船舶的进出港航行规律,其适航性能满足要求.To reduce the risk of ship navigation in port waters and improve the port production efficiency,a route planning algorithm for ship entering and leaving port was proposed based on probability directed graph model.The density-based clustering method was used to classify the historical trajectories of ships entering and leaving the port,so as to extract the normal navigation mode of the ship in the port,and the grid-based method was used to discretize the trajectories.The probability directed graph model was established with grid as node and adjacent grids as directed edge.Possible navigation paths of ships were searched dynamically for route planning by specifying the starting point and destination of the routes.100 inbound trajectories and 100 outbound trajectories were selected to verify the model.Results show that the average deviation between the planned inbound route and the real track of the ship is 0.18,and the deviation between the outbound planned route and the real track is 0.28.The route generated by the model combines the entry and departure rules of most ships,and the seaworthiness could meet the requirements.

关 键 词：航路规划 轨迹聚类 模式识别 概率有向图模型 船舶进出港 自动识别系统(AIS)数据 

分 类 号：U675.79[交通运输工程—船舶及航道工程]