A Data-Driven Approach for Collision Risk Early Warning in Vessel Encounter Situations Using Attention-BiLSTM

被引:24
|
作者
Ma, Jie [1 ,2 ,3 ]
Jia, Chengfeng [1 ]
Yang, Xin [1 ]
Cheng, Xiaochun [4 ]
Li, Wenkai [1 ]
Zhang, Chunwei [5 ]
机构
[1] Wuhan Univ Technol, Sch Nav, Wuhan 430063, Peoples R China
[2] Hubei Inland Shipping Technol Key Lab, Wuhan 430063, Peoples R China
[3] Natl Engn Res Ctr Water Transportat Safety, Wuhan 430063, Peoples R China
[4] Middlesex Univ, Dept Comp Sci, London NW4 4BE, England
[5] Huawei Technol Co Ltd, Wuhan Res Inst, Wuhan 430200, Peoples R China
基金
中国国家自然科学基金;
关键词
Marine vehicles; Navigation; Task analysis; Accidents; Trajectory; Safety; Time series analysis; Collision risk; vessel encounter; early warning; spatial-temporal model; bidirectional LSTM; attention mechanism; SHORT-TERM-MEMORY; AVOIDANCE; TRACKING; MODEL; NETWORK;
D O I
10.1109/ACCESS.2020.3031722
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Collision risk early warning is critical to sailing safety in vessel encounter situations because it provides ship officers with sufficient time to react to emergencies and take evasive actions in advance. In this study, we take spatiotemporal motion behaviors of encountering vessels into account since vessel motion behaviors have great influences on the occurrence of a dangerous situation. For this purpose, a data-driven approach is proposed to associate the motion behaviors with the future risk and early prediction of risk is achieved through classifying the behaviors into corresponding risk level. Specifically, we first derive a sequence of relative motion features between encountering vessels to characterize the spatial interactions that vary over time. Then a novel deep learning architecture, which combines bidirectional long short-term memory (BiLSTM) and attention mechanism, is developed to capture the spatial-temporal dependences of behaviors as well as their impacts on future risk. In particular, the BiLSTM is able to discover correlations among behaviors and the attention mechanism can emphasize the key information relevant to the risk prediction task. Exploiting the advantages of these two mechanisms makes the risk prediction more reasonable and reliable. Extensive experiments using ship trace data from the Yangtze River Estuary demonstrate that the proposed Attention-BiLSTM approach outperforms conventional LSTM in terms of accuracy and stability. Moreover, the real-time capability of the approach gives it a significant potential for use in predicting collisions at the early stages.
引用
收藏
页码:188771 / 188783
页数:13
相关论文
共 41 条
  • [31] A data-driven structural damage identification approach using deep convolutional-attention-recurrent neural architecture under temperature variations
    Fathnejat, Hamed
    Ahmadi-Nedushan, Behrouz
    Hosseininejad, Sahand
    Noori, Mohammad
    Altabey, Wael A.
    ENGINEERING STRUCTURES, 2023, 276
  • [32] Rainstorm-induced flood risk assessment in developed urban area using a data-driven approach with watershed units
    Zhou, Suhua
    Xu, Zhiwen
    Zhang, Qinshan
    Yu, Peng
    Jiang, Mingyi
    Li, Jinfeng
    Yang, Minghui
    SCIENCE OF THE TOTAL ENVIRONMENT, 2024, 946
  • [33] Impact of geophysical and anthropogenic factors on wildfire size: a spatiotemporal data-driven risk assessment approach using statistical learning
    Masoudvaziri, Nima
    Ganguly, Prasangsha
    Mukherjee, Sayanti
    Sun, Kang
    STOCHASTIC ENVIRONMENTAL RESEARCH AND RISK ASSESSMENT, 2022, 36 (04) : 1103 - 1129
  • [34] Impact of geophysical and anthropogenic factors on wildfire size: a spatiotemporal data-driven risk assessment approach using statistical learning
    Nima Masoudvaziri
    Prasangsha Ganguly
    Sayanti Mukherjee
    Kang Sun
    Stochastic Environmental Research and Risk Assessment, 2022, 36 : 1103 - 1129
  • [35] Quantifying Spatio-temporal risk of Harmful Algal Blooms and their impacts on bivalve shellfish mariculture using a data-driven modelling approach
    Stoner, Oliver
    Economou, Theo
    Torres, Ricardo
    Ashton, Ian
    Brown, Ross
    HARMFUL ALGAE, 2023, 121
  • [36] Data-driven prediction for curved pipe jacking performance during underwater excavation of ancient shipwreck using an attention-based graph convolutional network approach
    Dai, Zeyu
    Li, Peinan
    Liu, Jun
    Liu, Xue
    Rui, Yi
    Zhai, Yixin
    EXPERT SYSTEMS WITH APPLICATIONS, 2024, 236
  • [37] Identifying neurobiological heterogeneity in clinical high-risk psychosis: a data-driven biotyping approach using resting-state functional connectivity
    Tang, Xiaochen
    Wei, Yanyan
    Pang, Jiaoyan
    Xu, Lihua
    Cui, Huiru
    Liu, Xu
    Hu, Yegang
    Ju, Mingliang
    Tang, Yingying
    Long, Bin
    Liu, Wei
    Su, Min
    Zhang, Tianhong
    Wang, Jijun
    SCHIZOPHRENIA, 2025, 11 (01)
  • [38] Early-Stage Form-Finding for a Complex Urban High-Rise using an Informed Data-Driven Design Approach Case study of the Westblaaktoren
    Titulaer, Rick
    Vola, Mathew
    Nijenmanting, Filique
    ECAADE 2023 DIGITAL DESIGN RECONSIDERED, VOL 1, 2023, : 211 - 220
  • [39] Development and validation of a diagnostic model for early differentiation of sepsis and non-infectious SIRS in critically ill children - a data-driven approach using machine-learning algorithms
    Florian Lamping
    Thomas Jack
    Nicole Rübsamen
    Michael Sasse
    Philipp Beerbaum
    Rafael T. Mikolajczyk
    Martin Boehne
    André Karch
    BMC Pediatrics, 18
  • [40] Development and validation of a diagnostic model for early differentiation of sepsis and non-infectious SIRS in critically ill children a data-driven approach using machine-learning algorithms
    Lamping, Florian
    Jack, Thomas
    Ruebsamen, Nicole
    Sasse, Michael
    Beerbaum, Philipp
    Mikolajczyk, Rafael T.
    Boehne, Martin
    Karch, Andre
    BMC PEDIATRICS, 2018, 18