中国机械工程 ›› 2022, Vol. 33 ›› Issue (10): 1142-1152.DOI: 10.3969/j.issn.1004-132X.2022.10.002

• 先进水路交通装备与智能系统专辑(续) • 上一篇    下一篇

基于多模型融合的内河船舶航行轨迹预测方法

张阳1;高曙1;何伟2;蔡菁1   

  1. 1.武汉理工大学计算机与人工智能学院,武汉,430070
    2.闽江学院物理与电子信息工程学院,福州,350108
  • 出版日期:2022-05-25 发布日期:2022-06-09
  • 通讯作者: 高曙(通信作者),女,1967年生,教授。E-mail:gshu418@163.com。
  • 作者简介:张阳,男,1995年生,硕士研究生。研究方向为数据分析与智能交通。
  • 基金资助:
    国家自然科学基金(52172327);
    绿色智能内河船舶创新国家重大科技专项

Navigation Trajectory Prediction Method of Inland Ships Based on Multi-model Fusion

ZHANG Yang1;GAO Shu1;HE Wei2;CAI Jing1   

  1. 1.School of Computer and Artificial Intelligence,Wuhan University of Technology,Wuhan,430070
    2.School of Physics and Electronic Information Engineering,Minjiang University,Fuzhou,350108
  • Online:2022-05-25 Published:2022-06-09

摘要: 内河航运是现代综合运输体系的重要组成部分,实时和高精度的船舶轨迹预测方法能够有效规避水上交通事故、增强船舶自动化与智能化监管能力。针对现有内河船舶轨迹预测方法精度不高的问题,以提高船舶轨迹短期预测精度为目标,综合使用待测船舶近期船舶自动识别系统(AIS)数据和历史AIS数据,基于轨迹与航速和航向间的内在联系以及内河航道特点,构建了面向航速和航向预测的时域卷积网络模型、船舶轨迹动力学方程模型、自适应双隐层径向基函数网络等模型,提出了基于多模型融合的船舶轨迹预测方法。实验结果表明,所提方法轨迹预测精度有明显提高,并能满足实时性要求。

关键词: 时域卷积网络, 径向基函数网络, 多模型融合, 内河船舶, 轨迹预测

Abstract:  Inland waterway navigation was an important part of the modern comprehensive transportation systems. The real-time and high-precision ship trajectory prediction method was helpful to effectively avoid water traffic accidents and enhance the ability of automation and intelligent supervision. Aiming at the problems that the accuracy of the existing inland ship trajectory prediction was not high, in order to improve the short-term prediction accuracy of ship trajectory, comprehensively using the recent AIS (automatic identification system) data and historical AIS data of the ships, and based on the relationship among trajectory and speed, course, and the characteristics of inland waterway, the temporal convolutional network model for speed and course prediction, ship trajectory dynamics equation model and adaptive double-hidden layer RBF network were constructed. The ship trajectory prediction method based on multi-model fusion was proposed. Experimental results show that the proposed method has obvious improvement in trajectory prediction accuracy and may meet the real-time requirements. 

Key words:  , temporal convolutional network, radial basis function(RBF) network, multi-model fusion, inland ship, trajectory prediction

中图分类号: