A CNN-LSTM Ship Motion Extreme Value Prediction Model

doi:10.16183/j.cnki.jsjtu.2022.089

Abstract

Abstract:

Aimed at the short-term extreme value prediction of ship motion, a sliding window method based on motion spectrum information is proposed to extract feature data, based on which, a series prediction model of convolutional neural networks (CNN) and long short-term memory (LSTM) is built. The CNN module aims at the local correlation characteristics of the input data, and the LSTM module aims at the time dimension characteristics of the data. The simulation test results of S175 ship show that the model has a good prediction effect on the motion extremum information in the next 1 and 2 cycles, and the evaluation indexes are significantly better than those of LSTM and gate recurrent unit (GRU) models, which has an important application value.

Key words: local extremum, short-term prediction, convolutional neural network (CNN), long short-term memory (LSTM) networks

CLC Number:

U664.82

ZHAN Ke, ZHU Renchuan. A CNN-LSTM Ship Motion Extreme Value Prediction Model[J]. Journal of Shanghai Jiao Tong University, 2023, 57(8): 963-971.

Figures/Tables 13

Fig.1

Fig.2

Fig.3

Fig.4

Tab.1

Fig.5

Fig.6

Fig.7

Fig.8

Tab.2

Fig.9

Tab.3

Fig.10

References 17

[1]	喻元根, 马雪泉, 季盛. 基于AR的船舶运动极短期预报摇摆平台试验[J]. 上海船舶运输科学研究所学报, 2016, 39(4): 4-7.
	YU Yuangen, MA Xuequan, JI Sheng. Short time prediction of ship motion based on AR model and stewart platform experiment[J]. Journal of Shanghai Ship and Shipping Research Institute, 2016, 39(4): 4-7.
[2]	范海平. 基于卡尔曼滤波技术的船舶横摇预测方法研究[D]. 哈尔滨: 哈尔滨工程大学, 2008.
	FAN Haiping. Study on the prediction and estimation of ship roll motion based on Kalman filter[D]. Harbin: Harbin Engineering University, 2008.
[3]	ZHOU B, SHI A. Empirical mode decomposition based LSSVM for ship motion prediction[C]//International Symposium on Neural Networks. Berlin, Heidelberg, Germany: Springer, 2013.
[4]	PENA F L, GONZALEZ M M, CASAS V D, et al. An ANN based system for forecasting ship roll motion[C]//Computational Intelligence and Virtual Environments for Measurement Systems and Applications. Milan, Italy: IEEE Instrumentation & Measurement Magazine, 2013.
[5]	HOCHREITER S, SCHMIDHUBER R A, et al. Long short-term memory[J]. Neural Computation, 1997, 9: 1735-1780. doi: 10.1162/neco.1997.9.8.1735 pmid: 9377276
[6]	WANG Y, WANG H, ZHOU B, et al. Multi-dimensional prediction method based on Bi-LSTMC for ship roll[J]. Ocean Engineering, 2021, 242: 110106. doi: 10.1016/j.oceaneng.2021.110106 URL
[7]	WANG Y, WANG H, ZOU D, et al. Ship roll prediction algorithm based on Bi-LSTM-TPA combined model[J]. Journal of Marine Science and Engineering, 2021, 9(4): 387-1-16. doi: 10.3390/jmse9040387 URL
[8]	ZHANG T, ZHENG X Q, LIU M X. Multiscale attention-based LSTM for ship motion prediction[J]. Ocean Engineering, 2021, 230(13): 109066. doi: 10.1016/j.oceaneng.2021.109066 URL
[9]	张彪, 彭秀艳, 高杰. 基于ELM-EMD-LSTM组合模型的船舶运动姿态预测[J]. 船舶力学, 2020, 24(11): 1413-1421.
	ZHANG Biao, PENG Xiuyan, GAO Jie. Ship motion attitude prediction based on ELM-EMD-LSTM integrated model[J]. Journal of Ship Mechanics, 2020, 24(11): 1413-1421.
[10]	STOICA P, MOSES R. Spectral analysis of signals[M]. New Jersey, USA: Prentice Hall, 2005.
[11]	KETKAR N, MOOLAYIL J. Deep Learning with Python[M]. Berkeley, USA: Apress, 2021: 197-242.
[12]	赵建利, 白格平, 李英俊, 等. 基于CNN-LSTM的短期风电功率预测[J]. 自动化仪表, 2020, 41(5): 37-41.
	ZHAO Jianli, BAI Geping, LI Yingjun, et al. Short-term wind power prediction based on CNN-LSTM[J]. Process Automation Instrumentation, 2020, 41(5): 37-41.
[13]	王国栋. 基于LSTM的舰船运动姿态短期预测及仿真研究[D]. 镇江: 江苏科技大学, 2017.
	WANG Guodong. Short-term prediction and simulation of ship's motion based on LSTM[D]. Zhenjiang: Jiangsu University of Science and Technology, 2017.
[14]	刘秀丽, 徐小力. 基于特征金字塔卷积循环神经网络的故障诊断方法[J]. 上海交通大学学报, 2022, 56(2): 182-190.
	LIU Xiuli, XU Xiaoli. Fault diagnosis method based on feature pyramid CRNN network[J]. Journal of Shanghai Jiao Tong University, 2022, 56(2): 182-190.
[15]	TRISTAN P, FOSSEN T I. A MATLAB toolbox for parametric identification of radiation-force models of ships and offshore structures[J]. Modeling, Identification and Control, 2009, 30(1): 1-15. doi: 10.4173/mic.2009.1.1 URL
[16]	朱仁传, 缪国平. 船舶在波浪上的运动理论[M]. 上海: 上海交通大学出版社, 2019: 185-186.
	ZHU Renchuan, MIAO Guoping. The theory of ship motion in waves[M]. Shanghai: Shanghai Jiao Tong University Press, 2019: 185-186.
[17]	MWL A, DYX A, JING G A, et al. A hybrid approach for forecasting ship motion using CNN-GRU-AM and GCWOA[J]. Applied Soft Computing, 2022, 114: 108084. doi: 10.1016/j.asoc.2021.108084 URL

层	核数	尺寸	步长	隐藏节点数	失活率
卷积层1	96	(2, 1)	1	—	—
池化层1	—	(3, 1)	1	—	—
卷积层2	64	(1, 1)	1	—	—
池化层2	—	—	—	—	—
LSTM层1	—	—	—	200	—
LSTM层2	—	—	—	150	—
LSTM层3	—	—	—	100	—
Dropout层	—	—	—	—	0.5

模型	垂荡				横摇				纵摇
模型	e_MSE/ m²	e_RMSE/ m	e_MAE/ m	r	e_MSE/ (°)²	e_RMSE/ (°)	e_MAE/ (°)	r	e_MSE/ (°)²	e_RMSE/ (°)	e_MAE/ (°)	r
CNN-LSTM	0.067 2	0.259 3	0.215 1	0.963 8	0.038 2	0.195 3	0.166 5	0.953 4	0.190 8	0.436 8	0.355 5	0.975 1
LSTM	0.151 1	0.388 7	0.318 3	0.831 8	0.078 2	0.269 9	0.214 3	0.873 1	0.362 5	0.602 1	0.487 0	0.937 2
GRU	0.155 8	0.394 7	0.323 7	0.829 9	0.072 2	0.208 8	0.212 0	0.874 1	0.361 7	0.601 4	0.488 0	0.937 3

预报步数	垂荡					横摇					纵摇
预报步数	e_MSE/ m²	e_RMSE/ m	e_MAE/ m	r		e_MSE/ (°)²	e_RMSE/ (°)	e_MAE/ (°)	r		e_MSE/ (°)²	e_RMSE/ (°)	e_MAE/ (°)	r
提前2步	0.177 2	0.421 0	0.345 4	0.818 0		0.067 0	0.258 9	0.203 6	0.900 7		0.407 2	0.638 2	0.518 9	0.946 1