下载全文： 
202305014.pdf
文章编号： 1672-6987（2023）05-0115-07； DOI： 10.16351/j.1672-6987.2023.05.014
李海涛， 苏美玲， 张俊虎(青岛科技大学 信息科学技术学院，山东 青岛 266061）
摘要： 渔船轨迹预测的研究对于海上航行安全以及海洋资源保护都有重要的研究意义和价值。针对渔船航行影响因素复杂、传统模型预测难度大的问题，提出了注意力机制(attention mechanism，AM)和长短期记忆神经网络(long short-term memory，LSTM)结合的渔船轨迹预测模型(AM-LSTM)。以渔船位置的经度、纬度和渔船的速度、航向、自身属性作为模型输入参数，建立基于经度和纬度的渔船航迹经度预测模型和纬度预测模型。利用大西洋区域内的AIS数据进行实验验证，结果表明，与未加入注意力的LSTM模型对比，AM-LSTM的预测精度明显高于LSTM模型;加入渔船自身数据的模型预测精度明显高于未加入渔船自身数据的模型预测精度。
关键词： 注意力机制; LSTM; 轨迹预测; AIS数据
中图分类号： TP 183文献标志码： A
引用格式： 李海涛， 苏美玲， 张俊虎. 一种渔船轨迹预测的AM-LSTM算法［J］. 青岛科技大学学报(自然科学版), 2023, 44(5): 115-121.
LI Haitao, SU Meiling, ZHANG Junhu. An AM-LSTM algorithm for predicting the trajectory of fishing vessels［J］. Journal of Qingdao University of Science and Technology（Natural Science Edition), 2023， 44（5）： 115-121.
An AM-LSTM Algorithm for Predicting the Trajectory of Fishing Vessels
LI Haitao, SU Meiling, ZHANG Junhu
(College of Information Science and Technology, Qingdao University of Science and Technology, Qingdao 266061, China)
Abstract: The research of fishing vessel trajectory prediction is of great significance and value for the safety of maritime navigation and the protection of marine resources. In view of the complex influencing factors of fishing vessel navigation and the difficulty of traditional model prediction, a fishing vessel trajectory prediction model based on attention model and Long Short-Term Memory (AM-LSTM) is proposed. The longitude and latitude prediction model and latitude prediction model of fishing vessel track based on longitude and latitude are established by taking the longitude and latitude of fishing vessel position and the speed, heading and self-attribute of fishing vessel as the input parameters of the model. Experimental verification using AIS data in the Atlantic region, and the results show that, compared with the LSTM model without attention, the prediction accuracy of the AM-LSTM model is significantly higher than that of the LSTM model, and the prediction accuracy of the model with the fishing vessel′s own data is significantly higher than that of the model without the fishing vessel′s own data.
Key words: attention mechanism; LSTM; trajectory prediction; AIS data
收稿日期： 2022-11-06
基金项目： 国家自然科学基金项目(61806107)；青岛市创新创业领军人才项目(15-07-03-0030)；农业部水产养殖数字建设试点项目(2017-A2131-130209-K0104-004).
作者简介： 李海涛(1978—),男,副教授.
