针对双向神经网络结构复杂度高、鲁棒性不足的问题，本文提出一种非对称双向长短期记忆门控循环单元模型（Asymmetric Bidirectional Long Short-Term Memory neural network- Gated Recurrent Units network, AB-LGRU），该模型采用长短期记忆神经网络和门控循环单元网络处理正反向信息，分别捕获前后船舶轨迹特征信息。首先，预处理船舶轨迹数据，获取弯曲轨迹和直航轨迹；然后，训练本文模型并开展轨迹预测，与4种轨迹预测模型进行对比。采用均方误差作为损失函数对比分析5种模型的精度，并采用测试集预测结果对比分析模型预测效果。实验结果表明，AB-LGRU在训练集和验证集上均表现出最高精度，测试集预测结果均具有误差小、精确度高的优点。本文研究成果能够为船舶轨迹预测模型提供新的理论方法，预测的轨迹数据为水上交通管理决策提供指导。
In response to the problems of high structural complexity and limited robustness inherent in bidirectional neural networks by proposing an Asymmetric Bidirectional Long Short-Term Memory-Gated Recurrent Unit (AB-LGRU) network, which employs Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) networks to separately process forward and backward information, effectively capturing sequential ship trajectory features. Initially, ship trajectory data are preprocessed to categorize them into curved and straight trajectories. Subsequently, the proposed AB-LGRU model is trained, and its trajectory predictions are compared against four other trajectory prediction models. The mean square error serves as the loss function, facilitating an analysis of prediction accuracy among the five models using the test set results. Experimental outcomes indicate that AB-LGRU achieves superior accuracy in both training and validation datasets, demonstrating minimal prediction errors and enhanced accuracy on the test set. This research offers a novel theoretical approach for ship trajectory prediction modeling, and the resultant predictive data provide valuable guidance for decision-making in maritime traffic management.
2026,48(2): 151-156 收稿日期：2025-4-29
DOI：10.3404/j.issn.1672-7649.2026.02.025
分类号：U694
基金项目：国家留学基金委创新型人才国际合作项目（CXXM2209260070）
作者简介：张杰(1985-),男,博士,副教授,研究方向为海上交通规划与管理、船舶运输保障技术
参考文献：
[1] XIN X, LIU K, LI H, et al. Maritime traffic partitioning: an adaptive semi-supervised spectral regularization approach for leveraging multi-graph evolutionary traffic interactions[J]. Transportation Research Part C: Emerging Technologies, 2024, 164: 104670.
[2] WANG W, HE N, CHEN M, et al. Freight rate index forecasting with prophet model based on multi-dimensional significant events[J]. Expert Systems with Applications, 2024, 249: 123451.
[3] 何静. 利用卡尔曼滤波预测船舶航行轨迹异常行为[J]. 舰船科学技术, 2017, 39(2): 16-18.
HE J. Prediction of ship navigation trajectory abnormal behavior by Kalman filter[J]. Ship Science and Technology, 2017, 39(1Z): 16-18.
[4] 杜志强, 谭玉琪, 仇林遥. 基于卡尔曼滤波的船舶轨迹异常行为快速检测方法[J]. 地理信息世界, 2021, 28(4): 112-118.
DU Z Q, TAN Y Q, QIU L Y. On the rapid detection of abnormal ship trajectories by kalman filter[J]. Geomatics World, 2021, 28(4): 112-118.
[5] RONG H, TEIXEIRA A P, GUEDES SOARES C. Ship trajectory uncertainty prediction based on a gaussian process model[J]. Ocean Engineering, 2019, 182: 499-511.
[6] 闫佩玉. 舰船航行非线性轨迹预测的数学模型[J]. 舰船科学技术, 2020, 42(18): 25-27.
YAN P Y. Mathematical model design for nonlinear trajectory prediction of ship navigation[J]. Ship Science and Technology, 2020, 42(18): 25-27.
[7] DING X, BIAN H, MA H, et al. Ship trajectory generator under the interference of wind, current and waves[J]. Sensors, 2022, 22(23): 9395.
[8] TU E, ZHANG G, MAO S, et al. Modeling historical AIS data for vessel path prediction: a comprehensive treatment[M]. arXiv, 2020.
[9] 高大为, 朱永生, 张金奋, 等. 基于AIS数据的船舶航迹多维预测方法[J]. 中国航海, 2021, 44(3): 56-63.
GAO D W, ZHU Y S, ZHANG J F, et al. MuIti-dimensional prediction of ship track with AIS data augmentation[J]. Navigation of China, 2021, 44(3): 56-63.
[10] AGARAP A F M. A neural network architecture combining gated recurrent unit (GRU) and support vector machine (SVM) for intrusion detection in network traffic data[C]//Proceedings of the 2018 10th International Conference on Machine Learning and Computing. Macau China: ACM, 2018: 26-30.
[11] LI H, XING W, JIAO H, et al. Bi-directional information fusion-driven deep network for ship trajectory prediction in intelligent transportation systems[J]. Transportation Research Part E: Logistics and Transportation Review, 2024, 192: 103770.
[12] GAO M, SHI G, LI S. Online prediction of ship behavior with automatic identification system sensor data using bidirectional long short-term memory recurrent neural network[J]. Sensors, 2018, 18(12): 4211-4227.
[13] 马全党, 张丁泽, 王群朋, 等. 双向门控循环单元在船舶轨迹预测中的应用[J]. 安全与环境学报, 2024, 24(1): 83-91.
MA Q D, ZHANG D Z, WANG Q P, et al. Application of Bi-GRU for ship trajectory prediction[J]. Journal of Safety and Environment, 2024, 24(1): 83-91.
[14] 谢海波, 乔冠洲, 代程, 等. 基于NGO-Bi-GRU的船舶轨迹预测模型[J]. 舰船科学技术, 2025, 47(4): 14-20.
XIE H B, QIAO G Z, DAI C, et al. Ship trajectory prediction model based on NGO-Bi-GRU[J]. Ship Science and Technology, 2025, 47(4): 14-20.
[15] 徐洪敏. LSTM网络在船舶航行轨迹预测中的应用[J]. 舰船科学技术, 2021, 43(8): 37-39.
XU H M. Application of LSTM network in ship trajectory prediction[J]. Ship Science and Technology, 2021, 43(8): 37-39.
[16] 杨家轩, 吴长胜, 赵时雨. 基于相似度和密度的抗噪声船舶轨迹聚类方法[J]. 舰船科学技术, 2025, 47(2): 178-184.
YANG J X, WU C S, ZHAO S Y. The method of ship traffic trajectory clustering based on degree of similarity and density with noise[J]. Ship Science and Technology, 2025, 47(2): 178-184.
[17] WANG Y, WATANABE D, HIRATA E, et al. Real-time management of vessel carbon dioxide emissions based on automatic identification system database using deep learning[J]. Journal of Marine Science and Engineering, 2021, 9(8): 871.
[18] WANG J, GUO Y, WANG Y. A sequential random forest for short-term vessel speed prediction[J]. Ocean Engineering, 2022, 248: 110691.
