针对复杂场景中雷达和船舶自动识别系统（Automatic Identification System，AIS）由于船舶航迹密集导致错误关联增多、关联精度低的问题，提出一种基于Transformer的雷达AIS水面多目标轨迹关联算法。该算法通过结合Siamese架构并引入多头注意力和交叉注意力对Transformer模型进行改进，在此基础上利用改进模型提取雷达图像和AIS数据中目标轨迹的特征并进行匹配，实现了雷达图像和AIS数据中船舶目标轨迹与轨迹之间的关联。使用真实的内河航道场景数据制作关联数据集并进行实验，结果显示本文提出算法的关联精度较传统方法提高了11.42%，对比同类型的深度学习方法提升了5.41%，这表明该关联方法在密集航行场景下具有显著优势，验证了该方法的有效性。
To address the issues of increased false associations and low association accuracy caused by dense vessel tracks in complex scenarios involving radar and the AIS(Automatic Identification System), this paper proposes a radar-AIS surface multi-target trajectory association algorithm based on the Transformer architecture. This algorithm improves the Transformer model by combining the Siamese architecture with multi-head attention and cross-attention. Based on this improved model, features of target trajectories in radar images and AIS data are extracted and matched, enabling the association of vessel target trajectories between radar images and AIS data. An association dataset was created using real inland waterway scenario data, and experiments were conducted. The results showed that the association accuracy of the proposed algorithm improved by 11.42% compared to traditional methods and by 5.41% compared to similar deep learning methods, indicating that this association method has significant advantages in dense navigation scenarios and validating its effectiveness.
2026,48(8): 114-121 收稿日期：2025-7-28
DOI：10.3404/j.issn.1672-7649.2026.08.018
分类号：U675.74
基金项目：国家自然科学基金资助项目（52371374，51979210）
作者简介：王睿承(2001-),男,硕士研究生,研究方向为智能船舶技术
参考文献：
[1] 封波. 智能船舶设计方法体系研究[J]. 船舶工程, 2020, 42(S1): 1-4 FENG B. Research on intelligent ship design methodology system[J]. Ship Engineering, 2020, 42(S1): 1-4
[2] YU H, FANG Z, FU X, et al. Literature review on emission control-based ship voyage optimization[J]. Transportation Research Part D Transport and Environment, 2021, 93(1): 102768
[3] RØ JAN. Maritime unmanned navigation through intelligence in networks: the MUMIN project[J]. Unmanned Ship, 2013(3): 177-183
[4] 顾一清. 智能船舶技术体系研究及其对航运生态的影响[J]. 船舶工程, 2020, 42(3): 21-24 GU Y Q. Research on intelligent ship technology systems and their impact on the shipping industry[J]. Ship Engineering, 2020, 42(3): 21-24
[5] 严新平, 柳晨光. 智能航运系统的发展现状与趋势[J]. 智能系统学报, 2016, 11(6): 807-817 YAN X P, LIU C G. Current status and trends in the development of intelligent shipping systems[J]. CAAI Transactions on Intelligent Systems, 2016, 11(6): 807-817
[6] LEE M J, KIM J E, RYU B H, et al. Robust maritime target detector in short dwell time[J]. Remote Sensing, 2021, 13(7): 1319
[7] 毛勇峰. 基于海杂波动态特性的航海雷达目标检测改进技术研究[D]. 哈尔滨: 哈尔滨工程大学, 2024.
[9] 吴孝贤, 纪永刚. 基于Kuhn-Munkres算法的船载地波雷达与自动识别系统船只目标航迹关联[J]. 中国海洋大学学报(自然科学版), 2024, 54(11): 142-150 WU X X, JI Y G. Shipborne ground wave radar and AIS ship target track association based on the Kuhn-Munkres algorithm[J]. Periodical of Ocean University of China, 2024, 54(11): 142-150
[10] WANG C M, LI Y A, MIN L X, et al. Intelligent marine area supervision based on AIS and radar fusion[J]. Ocean Engineering, 2023, 285: 115373
[11] LEI J, SUN Y, WU Y, et al. Association of AIS and radar data in intelligent navigation in inland waterways based on trajectory characteristics[J]. Journal of Marine Science and Engineering, 2024, 12(6): 890
[12] 刘根旺, 刘永信, 纪永刚, 等. 基于模糊双门限的高频地波雷达与AIS目标航迹关联方法[J]. 系统工程与电子技术, 2016, 38(3): 557-562 LIU G W, LIU Y X, JI Y G, et al. HF ground wave radar and AIS target track association method based on fuzzy double threshold[J]. Systems Engineering and Electronics, 2016, 38(3): 557-562
[13] 赵力, 韩晟, 刘峰. 一种基于模糊数学的灰色航迹关联方法[J]. 电脑编程技巧与维护, 2023(5): 135-137 ZHAO L, HAN S, LIU F. A grey track association method based on fuzzy mathematics[J]. Computer Programming Skills & Maintenance, 2023(5): 135-137
[14] 唐裕峰. 基于机器学习的雷达与AIS航迹关联方法研究[D]. 镇江: 江苏科技大学, 2023.
[15] 黄善稳. 基于神经网络的AIS和雷达的目标信息融合研究[D]. 重庆: 重庆交通大学, 2023.
[16] KIM K T. Focusing of high range resolution profiles of moving targets using stepped frequency wave-forms[J]. Iet Radar Sonar & Navigation, 2010, 4(4): 564-575
[17] YANG Y, YANG F, SUN L, et al. Multi-target association algorithm of AIS-radar tracks using-graph matching-based deep neural network[J]. Ocean Engineering, 2022, 266: 112208
[18] VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[C]//Advances in Neural Information Processing Systems, 2017, 30: 5998–6008.
[19] 池哲瀛, 张毅, 陈恒强, 等. 船舶AIS和海事雷达信息融合感知方法研究[J]. 中国造船, 2025, 66(1): 36-49 CHI Z Y, ZHANG Y, CHEN H Q, et al. Research on information fusion perception methods for ship AIS and maritime radar[J]. Shipbuilding of China, 2025, 66(1): 36-49
