船舶行为是水上交通研究中的热点，船舶密集程度与交通流量的日益增长使得船舶行为变的更加复杂，船舶碰撞事故发生的几率显著增加，水上交通安全监管的压力日趋繁重。准确表征船舶行为并实现行为预测对掌握船舶航行规律，拓新船舶安全监管方法、提高船舶自主导航与决策能力具有重要意义。

  以长江口南槽交汇水域为研究对象，利用船舶自动识别系统（Automatic Identification System，AIS）提取船舶行为特征参量，考量船舶行为的时间演化特性建立了行为特征序列，提出基于上下文自编码的船舶行为语义表征模型；利用双向长短期记忆网络（Bi-directional Long Short-Term Memory，Bi-LSTM）建立Sequence-to-Sequence（Seq2Seq）框架，结合注意力机制提出了船舶行为预测模型；最后，考量一定距离内两艘船舶间的冲突概率，构建船舶行为冲突检测方法。具体研究工作与成果如下：

（1）基于上下文自编码的船舶行为语义表征。对交汇水域的AIS数据进行解析、插值以及平滑等预处理方法，实现对船舶轨迹数据的补全；分析行为特征参量（经度、纬度、航速和航向）在船舶航行过程中的时间演化特性，构建了船舶的行为特征序列；借助连续词袋模型将行为特征序列划分为中心船舶行为和上下文船舶行为，利用深度自编码网络构建船舶上下文行为的语义表征模型，模型得到的中心船舶行为编码作为表征向量输出，随后通过聚类算法构建船舶行为词典；选取南槽交汇水域数据进行验证与分析，结果表明所提出的模型对不同行为特征参量的表征误差低于0.6，基于多种聚类评价指标构建船舶行为词典，实现了不同行为模式下船舶行为词组合关系的提取。

（2）基于表征学习的船舶行为预测。利用核密度估计方法分析行为特征参量在船舶航行过程中的差异性，提取不同行为的行为特征序列，并根据交汇水域警戒区边界对行为特征序列进行切割；结合Seq2Seq框架与注意力机制构建船舶行为预测模型和算法，将交汇水域未进入警戒区的行为特征序列作为输入，将警戒区内行为特征序列作为输出，对船舶行为预测模型进行有监督训练，实现对警戒区内船舶行为的预测；对左转、直行、右转行为下的场景实例开展实验分析，结果表明所提出模型能够较好的预测警戒区内船舶行为的分布特性与变化趋势，与Bi-LSTM、门控循环单元（Gated Recurrent Unit，GRU）、循环神经网络（Recurrent Neural Network, RNN）三种模型进行对比，提出的模型具有更好的准确性。

（3）交汇水域的船舶行为冲突检测。围绕会遇船舶设计安全隔离区域，计算两艘船舶之间的行为冲突概率，建立船舶行为冲突概率度量模型；利用船舶行为预测模型对船舶行为进行预测，并将预测值作为船舶行为冲突概率度量模型的输入，根据警戒区内行为特征序列计算任意时刻的行为冲突概率，设计船舶行为冲突检测流程；针对不同会遇场景实例进行实验分析，结果表明所提出模型可以利用船舶未进入警戒区的行为信息准确反映未来一段时间内行为冲突概率的变化趋势。

   研究成果可为研制水上交通安全管理系统、船舶助航服务系统与船舶避碰预警与决策系统提供理论方法指导和实验数据参考。

[1] Mou J M , Chen P F , He Y X , et al. Vessel traffic safety in busy waterways: A case study of accidents in western Shenzhen port[J]. Accident Analysis & Prevention, 2016, 123(FEB.):461-468.

[2] Silveira P A M , Teixeira A P , Soares C G . Use of AIS Data to Characterise marine traffic patterns and ship collision risk off the Coast of Portugal[J]. Journal of Navigation, 2013, 66(6):879-898.

[3] Zaman M B, Kobayashi E, Wakabayashi N, et al. Fuzzy FMEA model for risk evaluation of ship collisions in the Malacca Strait: based on AIS data[J]. Journal of Simulation, 2014, 8(1): 91-104.

[4] Zhang W, Goerlandt F, Kujala P, et al. An advanced method for detecting possible near miss ship collisions from AIS data[J]. Ocean Engineering, 2016, 124: 141-156.

[5] Zhang W, Goerlandt F, Montewka J, et al. A method for detecting possible near miss ship collisions from AIS data[J]. Ocean Engineering, 2015, 107: 60-69.

[6] 张树波,唐强荣.基于AIS数据的船舶异常行为检测方法[J]．人工智能与机器人研究，2015,4(4):23-31.

[7] Tu E, Zhang G, Rachmawati L, et al. Exploiting AIS data for intelligent maritime navigation: A comprehensive survey from data to methodology[J]. IEEE Transactions on Intelligent Transportation Systems, 2017, 19(5): 1559-1582.

[8] Wu X , Mehta A L , Zaloom V A , et al. Analysis of waterway transportation in Southeast Texas waterway based on AIS data[J]. Ocean Engineering, 2016, 121:196-209.

[9] Huang Y, Van Gelder P. Time‐Varying Risk Measurement for Ship Collision Prevention[J]. Risk Analysis, 2020, 40(1): 24-42.

[10] Fang Z, Yu H, Ke R, et al. Automatic identification system-based approach for assessing the near-miss collision risk dynamics of ships in ports[J]. IEEE Transactions on Intelligent Transportation Systems, 2018, 20(2): 534-543.

[11] Xiao Z, Ponnambalam L, Fu X, et al. Maritime traffic probabilistic forecasting based on vessels’ waterway patterns and motion behaviors[J]. IEEE Transactions on Intelligent Transportation Systems, 2017, 18(11): 3122-3134.

[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): 4211-4226.

[13] Zhou Y, Daamen W, Vellinga T, et al. Review of maritime traffic models from vessel behavior modeling perspective[J]. Transportation Research Part C: Emerging Technologies, 2019, 105: 323-345.

[14] 潘家财,邵哲平,姜青山.数据挖掘在海上交通特征分析中的应用研究[J].中国航海,2010,33(2):60-62,73.

[15] 朱飞祥, 张英俊, 高宗江. 基于数据挖掘的船舶行为研究[J]. 中国航海, 2012,35(2): 50-54.

[16] 徐武雄,初秀民,刘兴龙.多桥水域航道通过能力仿真研究[J].交通运输系统工程与信息,2015,15(3):127-133.

[17] 陶阳, 毛喆, 盛萍, 等. 基于船舶行为的武汉长江大桥水域货船通航规律研究[J]. 交通信息与安全, 2018, 36(1): 49-56.

[18] Xiao F, Ligteringen H, Van Gulijk C, et al. Comparison study on AIS data of ship traffic behavior[J]. Ocean Engineering, 2015, 95: 84-93.

[19] 甄荣, 邵哲平, 潘家财, 等. 基于 AIS 信息的航道内船舶速度分布统计分析[J]. 集美大学学报 (自然科学版), 2014, 4: 274-278.

[20] Armudi A, Marques W C, Oleinik P H. Analysis of ship behavior under influence of waves and currents[J]. Revista de Engenharia Térmica, 2017, 16(2): 18-26.

[21] 朱姣, 刘敬贤, 陈笑, 等. 基于轨迹的内河船舶行为模式挖掘[J]. 交通信息与安全, 2017, 35(3): 107-116.

[22] Zhou Y , Daamen W , Vellinga T , et al. Ship classification based on ship behavior clustering from AIS data[J]. Ocean Engineering, 2019, 175:176-187.

[23] Gao M, Shi G Y. Ship-handling behavior pattern recognition using AIS sub-trajectory clustering analysis based on the T-SNE and spectral clustering algorithms[J]. Ocean Engineering, 2020, 205: 106919.

[24] Zhao L, Shi G. A trajectory clustering method based on Douglas-Peucker compression and density for marine traffic pattern recognition[J]. Ocean Engineering, 2019, 172: 456-467.

[25] Zhang S, Shi G, Liu Z, et al. Data-driven based automatic maritime routing from massive AIS trajectories in the face of disparity[J]. Ocean Engineering, 2018, 155: 240-250.

[26] 文元桥,张义萌,黄亮,周春辉,肖长诗,张帆.基于语义的船舶行为动态推理机制[J].中国航海,2019,42(03):34-39+50.

[27] Cho Y, Kim J, Kim J. Intent Inference of Ship Collision Avoidance Behavior Under Maritime Traffic Rules[J]. IEEE Access, 2021, 9: 5598-5608.

[28] Wen Y, Sui Z, Zhou C, et al. Automatic ship route design between two ports: A data-driven method[J]. Applied Ocean Research, 2020, 96: 102049.

[29] Chen X, Liu Y, Achuthan K, et al. A ship movement classification based on Automatic Identification System (AIS) data using Convolutional Neural Network[J]. Ocean Engineering, 2020, 218: 108182.

[30] 桑凌志,耿丹阳,张诚,曹德胜,严新平.基于船舶行为特征的武汉大桥水域航线优化[J].中国航海,2017,40(03):58-63.

[31] Riveiro M, Pallotta G, Vespe M. Maritime anomaly detection: A review[J]. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery,2018,8(5):e1266.

[32] Botts C. An Alternative Metric for Detecting Anomalous Ship Behavior Using a Variation of the DBSCAN Clustering Algorithm[J]. arXiv preprint arXiv:2006.01936, 2020.

[33] Pallotta G, Vespe M, Bryan K. Vessel Pattern Knowledge Discovery From AIS Data: a Framework for Anomaly Detection and Route Prediction[J]. Entropy, 2013, 15(6):2288-2315.

[34] 吴建华,吴琛,刘文,郭俊纬.舶舶AIS轨迹异常的自动检测与修复算法[J].中国航海,2017,40(01):8-12+101.

[35] Kowalska K, Peel L. Maritime anomaly detection using Gaussian process active learning[C]//2012 15th International Conference on Information Fusion, Singapore, IEEE,USA, 2012: 1164-1171.

[36] Zhang Z, Suo Y, Yang S. Detection of Complex Abnormal Ship Behavior Based on Event Stream, 2020 Chinese Automation Congress (CAC), Shanghai, China, 2020, pp. 5730-5735.

[37] Mascaro S, Nicholso A E, Korb K B. Anomaly detection in vessel tracks using Bayesian networks[J]. International Journal of Approximate Reasoning, 2014, 55(1): 84-98.

[38] 熊勇,瞿小豪,郭俊维,牟军敏.基于AIS数据的渡轮异常运动模式监测[J].中国安全科学学报,2016,26(01):100-103.

[39] 黄亮,刘益,文元桥,周春辉,张帆.内河渡船异常行为识别[J].大连海事大学学报,2017,43(01):8-13.

[40] 吴睿智. 轨迹数据挖掘关键技术研究[D].电子科技大学,2020.

[41] Gao M, Shi G Y. Ship collision avoidance anthropomorphic decision-making for structured learning based on AIS with Seq-CGAN[J]. Ocean Engineering, 2020, 217: 107922.

[42] Wan C, Zhu Y, Yu J, et al. SMOPAT: Mining semantic mobility patterns from trajectories of private vehicles[J]. Information Sciences, 2018, 429: 12-25.

[43] Andrade T, Cancela B, Gama J. Discovering locations and habits from human mobility data[J]. Annals of Telecommunications, 2020, 75(9): 505-521.

[44] 高崇铭. 轨迹语义表征与地点推荐研究[D]. 成都:电子科技大学,2019.

[45] Sun J, Qi X, Xu Y, et al. Vehicle turning behavior modeling at conflicting areas of mixed-flow intersections based on deep learning[J]. IEEE Transactions on Intelligent Transportation Systems, 2019, 21(9): 3674-3685.

[46] Wang W , Xia F , Nie H , et al. Vehicle Trajectory Clustering Based on Dynamic Representation Learning of Internet of Vehicles[J]. IEEE Transactions on Intelligent Transportation Systems, 2020, PP(99):1-10.

[47] Dabiri S, Lu C T, Heaslip K, et al. Semi-supervised deep learning approach for transportation mode identification using GPS trajectory data[J]. IEEE Transactions on Knowledge and Data Engineering, 2019, 32(5): 1010-1023.

[48] Huang L, Wen Y, Guo W, et al. Mobility pattern analysis of ship trajectories based on semantic transformation and topic model[J]. Ocean Engineering, 2020, 201: 107092.

[49] Besse P C , Guillouet B , Loubes J M , et al. Review and Perspective for Distance-Based Clustering of Vehicle Trajectories[J]. IEEE Transactions on Intelligent Transportation Systems, 2016, 17(11): 3306-3317.

[50] Chu D, Sheets D A, Zhao Y, et al. Visualizing hidden themes of taxi movement with semantic transformation[C]//2014 IEEE Pacific Visualization Symposium. Yokohama, IEEE, USA, 2014: 137-144.

[51] Own C M, Sha F, Tao W. Triplet Decoders Neural Network Ensemble System and T-Conversion for Traffic Speed Sequence Prediction[J]. IEEE Access, 2019, 7: 162070-162082.

[52] Li L, Li X, Yang Y, et al. Indoor tracking trajectory data similarity analysis with a deep convolutional autoencoder[J]. Sustainable Cities and Society, 2019,45: 588-595.

[53] Dotti D, Popa M, Asteriadis S. A hierarchical autoencoder learning model for path prediction and abnormality detection[J]. Pattern Recognition Letters, 2020, 130: 216-224.

[54] Zhang R, Xie P, Jiang H, et al. Clustering noisy trajectories via robust deep attention auto-encoders[C]//2019 20th IEEE International Conference on Mobile Data Management (MDM) , Hong Kong, IEEE, USA, 2019: 63-71.

[55] Boyle J, Nawaz T, Ferryman J. Deep trajectory representation-based clustering for motion pattern extraction in videos[C]//2017 14th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS). Lecce, IEEE, USA, 2017: 1-6.

[56] Zissis D, Xidias E K, Lekkas D. Real-time vessel behavior prediction[J]. Evolving Systems, 2016, 7(1): 29-40.

[57] Suo Y, Chen W, Claramunt C, et al. A Ship Trajectory Prediction Framework Based on a Recurrent Neural Network[J]. Sensors, 2020, 20(18): 5133.

[58] Wang S, Bao Z, Culpepper J S, et al. A Survey on Trajectory Data Management, Analytics, and Learning[J]. arXiv preprint arXiv:2003.11547, 2020.

[59] Wang C, Ma L, Li R, et al. Exploring trajectory prediction through machine learning methods[J]. IEEE Access, 2019, 7: 101441-101452..

[60] Li F, Gui Z, Zhang Z, et al. A hierarchical temporal attention-based LSTM encoder-decoder model for individual mobility prediction[J]. Neurocomputing, 2020, 403: 153-166.

[61] Cai J , Zhang Y , Yang L , et al. A context-augmented deep learning approach for worker trajectory prediction on unstructured and dynamic construction sites[J]. Advanced Engineering Informatics, 2020, 46:101173.

[62] Feng J, Li Y, Zhang C, et al. Deepmove: Predicting human mobility with attentional recurrent networks[C]//Proceedings of the 2018 world wide web conference, Lyon, 2018: 1459-1468.

[63] Zhou F, Yue X, Trajcevski G, et al. Context-aware variational trajectory encoding and human mobility inference[C]//The World Wide Web Conference. San Francisco, 2019: 3469-3475.

[64] Ma L , Qu S . A sequence to sequence learning based car-following model for multi-step predictions considering reaction delay[J]. Transportation Research Part C: Emerging Technologies, 2020, 120: 102785.

[65] Xie Z , Lv W , Huang S , et al. Sequential Graph Neural Network for Urban Road Traffic Speed Prediction[J]. IEEE Access, 2020, 8:63349-63358.

[66] 韩皓, 谢天. 基于注意力 Seq2Seq 网络的高速公路交织区车辆变道轨迹预测[J]. 中国公路学报, 2020, 33(6): 106-118.

[67] Zhang Z, Li M, Lin X, et al. Multistep speed prediction on traffic networks: A deep learning approach considering spatio-temporal dependencies[J]. Transportation research part C: emerging technologies, 2019, 105: 297-322.

[68] 游兰, 韩雪薇, 何正伟, 等. 基于改进 Seq2Seq 的短时 AIS 轨迹序列预测模型[J]. 计算机科学, 47(9): 169-174.

[69] Murray B, Perera L P. A dual linear autoencoder approach for vessel trajectory prediction using historical AIS data[J]. Ocean Engineering, 2020, 209: 107478.

[70] You L, Xiao S, Peng Q, et al. ST-Seq2Seq: A Spatio-Temporal Feature-Optimized Seq2Seq Model for Short-Term Vessel Trajectory Prediction[J]. IEEE Access, 2020, 8: 218565-218574.

[71] Yao D, Zhang C, Zhu Z, et al. Trajectory clustering via deep representation learning[C]. International joint conference on neural network, Anchorage, 2017: 3880-3887.

[72] 谢鹏. 基于深度学习的轨迹数据中移动模式发现方法研究[D]. 武汉理工大学, 2019.

[73] Lin J , Keogh E , Wei L , et al. Experiencing SAX: a novel symbolic representation of time series[J]. Data Mining & Knowledge Discovery, 2007, 15(2):107-144.

[74] Mikolov T, Le Q V, Sutskever I. Exploiting similarities among languages for machine translation[J]. arXiv preprint arXiv:1309.4168, 2013.

[75] Krasnov F, Sen A. The number of topics optimization: Clustering approach[J]. Machine Learning and Knowledge Extraction, 2019, 1(1): 416-426.

[76] Hermes C, Wohler C, Schenk K, et al. Long-term vehicle motion prediction[C]//2009 IEEE intelligent vehicles symposium, Xi’an, IEEE, USA, 2009: 652-657.

[77] Ristic B, La Scala B, Morelande M, et al. Statistical analysis of motion patterns in AIS data: Anomaly detection and motion prediction[C]//2008 11th International Conference on Information Fusion. Cologne, IEEE, USA, 2008: 1-7.

[78] 金宸,李维华,姬晨等.基于双向LSTM神经网络模型的中文分词[J].中文信息学报,2018,32(02):29-37.

[79] Zaremba W, Sutskever I, Vinyals O. Recurrent neural network regularization[J]. arXiv preprint arXiv:1409.2329, 2014.

[80] Hochreiter S , Schmidhuber J . Long Short-Term Memory[J]. Neural Computation, 1997, 9(8):1735-1780.

[81] Cho K, Van Merrienboer B, Gulcehre C, et al. Learning phrase representations using RNN encoder-decoder for statistical machine translation[J]. arXiv preprint arXiv:1406.1078, 2014.

[82] Park J, Kim J. Predictive evaluation of ship collision risk using the concept of probability flow[J]. IEEE Journal of Oceanic Engineering, 2016, 42(4): 836-845.