针对单一模型在预测无人水面艇（Unmanned Surface Vessel，USV）运动姿态时精度不高的问题，提出一种基于卷积神经网络（CNN）和门控循环单元（GRU）的USV姿态多步预测模型。首先，使用滑动窗口法构造运动姿态数据集作为模型输入；然后，使用CNN模块挖掘时序数据的局部特征；最后，使用GRU模型进行多步预测。使用实测USV运动姿态数据进行预测实验，实验结果表明，该模型比XGBoost模型、单一LSTM模型和单一GRU模型具有更高的预测精度，各项评价指标表现更佳，具有重要的应用价值。
Aiming at the problem of low accuracy of single model in predicting the ship motion of Unmanned Surface Vehicle (USV), a multi-step prediction model based on Convolutional Neural Network (CNN) and Gate Recurrent Unit (GRU) is proposed. Firstly, the sliding window method is used to construct the motion data set as the model input. Then, the CNN module is used to mine the local features of time series data. Finally, the GRU network is used for multi-step prediction. The experimental results show that the model has higher prediction accuracy than XGBoost model, single LSTM model and single GRU model, and its performance of each evaluation index is better, which has important application value.
2024,46(13): 132-136 收稿日期：2023-09-01
DOI：10.3404/j.issn.1672-7649.2024.13.023
分类号：U661.32
基金项目：青岛市关键技术攻关及产业化示范类资助项目（23-1-3-hygg-11-hy）；三亚崖州湾科技城科研项目（SKJC 2022 01 001）
作者简介：宋大雷(1987-),男,教授,研究方向为智能感知与控制
参考文献：
[1] 张晓东, 刘世亮, 刘宇, 等. 无人水面艇收放技术发展趋势探讨[J]. 中国舰船研究, 2018(6): 50–57.
ZHANG Xiaodong, LIU Shiliang, LIU Yu, et al. Review on development tendency of launch and recovery technology for USV[J]. Chinese Journal of Ship Research, 2018(6): 50–57.
[2] KAPLAN P. A study of prediction techniques for aircraft carrier motions at sea[J]. Journal of Hydronautics. 1969, 3(3): 121–131.
[3] TRIANTAFYLLOU M, ATHANS M. Real time estimation of the heaving and pitching motions of a ship using a Kalman filter[C]// IEEE, Boston MA, 1981, 1090–1095.
[4] TRIANTAFYLLOU M, BODSON M. Real time prediction of marine vessel motion using Kalman filtering techniques[C]// IEEE, Houston, Texas, 1982,159–165.
[5] 刘煜城. 基于深度学习的船舶运动极短期预报方法研究[D]. 哈尔滨: 哈尔滨工程大学, 2019.
[6] 詹可, 朱仁传. 一种CNN-LSTM船舶运动极值预报模型[J/OL]. 上海交通大学学报: 1–9[2023–07–17].
[7] 谷达京, 施哲源, 陈根良, 等. 基于神经网络的船体运动位姿预测方法[J]. 舰船科学技术, 2022, 44(15): 55-59.
GU Dajing, SHI Zheyuan, CHEN Genliang, et al. Deep learning for predicting ship motion from images, pitches and rolls[J]. Ship Science and Technology, 2022, 44(15): 55-59.
[8] 严春满, 王铖. 卷积神经网络模型发展及应用[J]. 计算机科学与探索, 2021(1): 27-46.
[9] 杨丽, 吴雨茜, 王俊丽, 等. 循环神经网络研究综述[J]. 计算机应用, 2018, (S2): 1–6.
[10] HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780.
[11] CHO K, VAN MERRIENBOER B, GULCEHRE C, et al. Learning phrase representations using RNN Encoder-Decoder for statistical machine translation[J]. arXiv Preprint, 2014, 2014(1406): 1078.
[12] CHUNG J, GULCEHRE C, CHO K, et al. Empirical evaluation of gated recurrent neural networks on sequence modeling[J]. arXiv Preprint, 2014, 2014(1412): 3555.
[13] 姚程文, 杨苹, 刘泽健. 基于CNN-GRU混合神经网络的负荷预测方法[J]. 电网技术. 2020, (9): 3416-3424.
[14] CHEN T, GUESTRIN C. XGBoost: a scalable tree boosting system[C]// ACM SIGKDD International Conference on Knowledge & Data Mining, New York, 2016.
