LSTM 船舶航迹预测模型-《计算机技术与发展》

文章信息/Info

Author(s):: LI Yong; CHENG Meng-ya; Faculty of Information Technology,Beijing University of Technology,Beijing 100124,China

Keywords:: track prediction; AIS data; long short-term memory network; maritime safety; deep learning

摘要:: 在复杂庞大的海洋环境下行驶,精确的航迹预测应是确保船舶在海上安全的重要基础。船舶在海上航行,精准的定位还决定了船舶航行的工作效率。根据船舶航向的复杂特性以及船舶轨迹预测的精度和实时性的需求, 从神经网络入手,提出了基于神经网络的船舶航迹预测方法,充分探索船舶时间序列数据背后的运动规律,进而实现航迹预测。将船舶的经度,纬度,速度和航向作为模型的输入,而经度和纬度作为输出以建立船舶的航迹预测模型。使用 AIS 数据训练模型,并使用均方误差 MSE 作为评估指标,以预测船舶未来的航行位置。实验结果表明,该模型可以利用船舶运动状态的变化规律来预测未来某时间内的船舶运动状态,证明了所提方法的有效性,能够比较准确地预测出船舶轨迹。

Abstract:: When driving in a complex and huge ocean environment, accurate trajectory prediction should be an important basis for ensuring the safety of ships at sea. When a ship is sailing at sea,precise positioning also determines the efficiency of ship navigation. According to the complex characteristics of ship heading and? ?the demand for accuracy and real-time of ship trajectory prediction,starting with neural network,we propose? ?a neural network-based ship trajectory prediction method to fully explore the laws of motion behind ship time series data,and then realize trajectory prediction. The longitude,latitude,speed,and heading of the ship are used as the input of the model,and the longitude and latitude are used as the output to build the ship’s track prediction model. AIS data is used to train the model,and the mean square error ( MSE) is used as an evaluation indicator to predict the future navigation position of the ship. The experiment shows that this model can use the changing law of the ship’s motion state to predict the ship’s motion state in a certain time in the future,which proves the effectiveness of the proposed method and can predict the trajectory of the ship more accurately.