基于CNN和Bi-LSTM的船舶航迹预测  

Prediction Model of Ship Trajectory Based on CNN and Bi-LSTM

作　　者：刘姗姗 马社祥 孟鑫 张启超 LIU Shanshan;MA Shexiang;MENG Xin;ZHANG Qizhao(College of Electrical and Electronic Engineering,Tianjin University of Technology,Tianjin 300384,China)

机构地区：[1]天津理工大学电气电子工程学院,天津3OO384

出　　处：《重庆理工大学学报（自然科学）》2020年第12期196-205,共10页Journal of Chongqing University of Technology：Natural Science

基　　金：国家自然科学基金项目(61601326,61371108)。

摘　　要：船舶自动识别系统(automatic identification system,AIS)数据可以反映出船舶的航行状态特征,并且实时精确地预测船舶未来航行轨迹,能及时避免一些海上交通事故的发生,具有重要的现实意义。提出了一种根据船舶AIS数据训练混合深度学习网络预测船舶航行轨迹的方法。根据船舶AIS数据的航行轨迹特征,构建了基于卷积神经网络(convalutional neural network,CNN)和双向长短期记忆(bidirectional long short term memory,Bi-LSTM)网络的船舶航行轨迹预测混合模型。CNN-Bi-LSTM模型根据船舶AIS数据进行训练,形成期望的输入-输出映射关系,进而预测船舶未来的航行轨迹。实验结果表明,对比传统的预测方法,CNN-Bi-LSTM不仅能更加准确有效地处理序列数据,预测船舶航行轨迹的精确度也更高。The data of Automatic Identification System(AIS)can reflect the navigational characteristics of the ship,and it is of great practical significance to accurately predict the future trajectory of the ship in time to avoid some sea traffic accidents.Therefore,a method for predicting the ship’s navigation trajectory by training a hybrid deep learning network based on the ship’s AIS data is proposed.According to the characteristics of navigation trajectory of AIS data,a hybrid model of ship navigation trajectory prediction based on Convolutional Neural Network(CNN)and Bidirectional Long Short Term Memory(Bi-LSTM)network is proposed.The CNN-Bi-LSTM model is trained based on ship AIS data to form the desired input-output mapping relationship to predict the ship’s future trajectory.The experimental results show that compared with the traditional prediction method,CNN Bi LSTM can not only process the sequence data more accurately and effectively,but also predict the ship’s trajectory with higher accuracy.

关 键 词：AIS信息 卷积神经网络 双向长短期记忆 船舶航迹预测 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]