针对海浪有效波高序列波动性、随机性较强,难以精确预测以及模型无法高效挖掘深层特征间关系的问题,提出一种基于Non stationary-CNN-Transformer模型的海浪有效波高预测方法。首先,使用平稳化模块减弱海浪时序数据的非平稳性;其次,利用一维卷积神经网络（CNN）提取相关数据间的深层特征并构建特征向量;最后,使用含有平稳性注意力的Transformer描述波高序列的时间依赖性捕捉到序列之间的全局关系,通过逆归一化处理后获得有效波高预测结果。该方法可消除海浪时序数据的非平稳性,提升数据的预测效果,并具有优异的特征提取能力且善于处理大规模时间序列数据。在实验中应用澳大利亚的浮标实测数据,通过7组对比实验分别预测0.5、3、6、12和24 h的有效波高,对所提模型进行全方位、多角度的验证。算例研究结果表明,该文所提模型在不同时间段精度有明显提升。
Abstract
Aiming at the problems of wave effective height series volatility and strong randomness, which are difficult to predict accurately, and the model cannot efficiently explore the relationship between deep features, a method of wave effective height prediction based on non-stationary CNN-Transformer model is proposed. Firstly, a stabilization module is used to reduce the non-stationarity of wave time series data. Then, a one-dimensional convolutional neural network is used to extract deep features from the relevant data and create the feature vector. Finally, a Transformer with smoothing attention is used to depict the time-dependent wave height sequences and capture the overall relationship among the sequences. The effective prediction results of the wave height after the inverse normalization process are obtained. The approach eliminates the unstable nature of wave time series data, enhancing data prediction effects. Additionally, it possesses excellent feature extraction capabilities and can effectively handle large-scale time series data. During experiments, seven sets of comparative experiments were conducted with Australian buoys, predicting effective wave heights of 0.5, 3, 6, 12 and 24 hours. The proposed model were thoroughly verified from all angles and in every aspect. The case study findings demonstrate that the precision of the model put forward in this paper is notably enhanced across various time periods.
关键词
海洋能 /
时间序列 /
海浪 /
波高预测 /
非平稳CNN-Transformer /
非平稳Transformer
Key words
ocean energy /
time series /
ocean waves /
wave height prediction /
Non stationary-CNN-Transformer /
non stationary-Transformer
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] LÓPEZ I, ANDREU J, CEBALLOS S, et al. Review of wave energy technologies and the necessary power-equipment[J]. Renewable and sustainable energy reviews, 2013, 27: 413-434.
[2] TAN J M, ZHU J X, LI X Y, et al.OSP-FEAN: optimizing significant wave height prediction with feature engineering and attention network[C]//2022 IEEE International Conference on Systems, Man, and Cybernetics (SMC). Prague, Czech Republic, 2022: 2201-2208.
[3] WU M N, STEFANAKOS C, GAO Z.Multi-step-ahead forecasting of wave conditions based on a physics-based machine learning (PBML) model for marine operations[J]. Journal of marine science and engineering, 2020, 8(12): 992.
[4] 夏海南, 王项南, 李强, 等. 波浪能发电装置现场测试中波浪参数比测分析[J]. 太阳能学报, 2022, 43(6): 251-255.
XIA H N, WANG X N, LI Q, et al.Comparison and analysis of wave parameters in field test of wave energy converters[J]. Acta energiae solaris sinica, 2022, 43(6): 251-255.
[5] YU X J, LIU Y R, SUN Z M, et al.Wavelet-based ResNet: a deep-learning model for prediction of significant wave height[J]. IEEE access, 2022, 10: 110026-110033.
[6] 王春晓, 于华明, 李松霖, 等. 基于海浪再分析数据的波浪能资源分析[J]. 太阳能学报, 2022, 43(9): 430-436.
WANG C X, YU H M, LI S L, et al.Wave energy resource valuation based on sea wave reanalysis data[J]. Acta energiae solaris sinica, 2022, 43(9): 430-436.
[7] 张振全, 李醒飞, 杨少波. 基于AR-SVR模型的有效波高短期预测[J]. 太阳能学报, 2021, 42(7): 15-20.
ZHANG Z Q, LI X F, YANG S B.Short-term prediction of significant wave height based on AR-SVR model[J]. Acta energiae solaris sinica, 2021, 42(7): 15-20.
[8] ZHANG X Y, LI Y Q, GAO S, et al.Ocean wave height series prediction with numerical long short-term memory[J]. Journal of marine science and engineering, 2021, 9(5): 514.
[9] HASSELMANN K, HASSELMANN S.On the nonlinear mapping of an ocean wave spectrum into a synthetic aperture radar image spectrum and its inversion[J]. Journal of geophysical research: oceans, 1991, 96(C6): 10713-10729.
[10] DONELAN M A, DOBSON F W, SMITH S D, et al.On the dependence of sea surface roughness on wave development[J]. Journal of physical oceanography, 1993, 23(9): 2143-2149.
[11] ARDHUIN F, ROGERS E, BABANIN A V, et al.Semiempirical dissipation source functions for ocean waves. part I: definition, calibration, and validation[J]. Journal of physical oceanography, 2010, 40(9): 1917-1941.
[12] KOMEN G J, CAVALERI L, DONELAN M, et al.Dynamics and modelling of ocean waves[M]. Cambridge, UK: Cambridge University Press, 1994.
[13] GROUP T W.The WAM model—a third generation ocean wave prediction model[J]. Journal of physical oceanography, 1988, 18(12): 1775-1810.
[14] ETEMAD-SHAHIDI A, MAHJOOBI J.Comparison between M5’ model tree and neural networks for prediction of significant wave height in Lake Superior[J]. Ocean engineering, 2009, 36(15/16): 1175-1181.
[15] MAHJOOBI J, ADELI MOSABBEB E.Prediction of significant wave height using regressive support vector machines[J]. Ocean engineering, 2009, 36(5): 339-347.
[16] MALEKMOHAMADI I, BAZARGAN-LARI M R, KERACHIAN R, et al. Evaluating the efficacy of SVMs, BNs, ANNs and ANFIS in wave height prediction[J]. Ocean engineering, 2011, 38(2/3): 487-497.
[17] ZHAO K, WANG J C.Significant wave height forecasting based on the hybrid EMD-SVM method[J]. Indian journal of geo-marine sciences, 2019, 48(12): 1957-1962.
[18] FAN S T, XIAO N H, DONG S.A novel model to predict significant wave height based on long short-term memory network[J]. Ocean engineering, 2020, 205: 107298.
[19] GUAN X T.Wave height prediction based on CNN-LSTM[C]//2020 2nd International Conference on Machine Learning, Big Data and Business Intelligence (MLBDBI). Taiyuan, China, 2020: 10-17.
[20] WANG J C, WANG Y, YANG J G.Forecasting of significant wave height based on gated recurrent unit network in the Taiwan strait and its adjacent waters[J]. Water, 2021, 13(1): 86.
[21] 李自立, 蒙素素. 基于LSTM-ResNet模型的定点有效波高预测[J]. 海洋预报, 2022, 39(2): 80-85.
LI Z L, MENG S S.Prediction of significant wave height of a fixed point based on LSTM-ResNet model[J]. Marine forecasts, 2022, 39(2): 80-85.
[22] KIM T, KIM J, TAE Y, et al.Reversible instance normalization for accurate time-series forecasting against distribution shift[C]//International Conference on Learning Representations. Austria ,2021.
[23] LIU Y, WU H X, WANG J M, et al.Non-stationary Transformers: exploring the stationarity in time series forecasting[C] //Advances in Neural Information Processing Systems. New Orleans, USA, 2022.
[24] FUKUSHIMA K.Neocognitron: a self-organizing neural network model for a mechanism of pattern recognition unaffected by shift in position[J]. Biological cybernetics, 1980, 36(4): 193-202.
[25] VASWANI A, SHAZEER N, PARMAR N, et al.Attention is all you need[J]. Advances in neural information processing systems, 2017, 30: 5998-6008.
[26] 王园浩, 朱月尧, 查易艺. 基于Transformer模型的电力负荷预测[J]. 机电信息, 2023(4): 1-6.
WANG Y H, ZHU Y Y, ZHA Y Y.Power load forecasting based on Transformer model[J]. Mechanical and electrical information, 2023(4): 1-6.
[27] BAAK M, KOOPMAN R, SNOEK H, et al.A new correlation coefficient between categorical, ordinal and interval variables with Pearson characteristics[J]. Computational statistics & data analysis, 2020, 152: 107043.
基金
国家自然科学基金项目（62173055; 61673083）; 辽宁省自然科学基金计划（2023-MS-093）
PDF(1371 KB)
