目的 针对导管架平台在复杂海洋环境下的结构响应预测问题,提出一种基于序列到序列（Seq2Seq）架构、交叉注意力机制和门控循环单元（GRU）的混合深度学习预测模型。方法 该模型采用双分支GRU编码器分别提取历史环境特征和历史载荷的时序依赖关系,通过交叉注意力机制实现未来环境序列与历史信息的有效融合,并利用双层GRU解码器生成多步载荷预测。模型采用AdamW优化器和余弦退火学习率调度策略进行端到端训练。结果 所提方法在结构响应预测中取得了较高精度,决定系数R2达0.996 5,预测值与真实值具有较好的一致性。结论 所提的预测方法能够有效捕捉海洋环境-载荷之间的复杂耦合关系,其注意力权重具有良好的可解释性,为导管架平台的运维决策和健康管理提供了可靠的智能预测手段。
Abstract
To address the challenge of structural response prediction for jacket platforms in complex marine environments, the work aims to propose a hybrid deep learning prediction model based on the Sequence-to-Sequence (Seq2Seq) architecture, cross-attention mechanism, and Gated Recurrent Unit (GRU). By employing a dual-branch GRU encoder to extract temporal dependencies between historical marine environmental features and historical load sequences, the model achieved effective integration of future environmental sequences and historical information through the cross-attention mechanism, and utilized a two-layer GRU decoder to generate multi-step load predictions. Then, the model was trained end-to-end with the AdamW optimizer and a cosine annealing learning rate scheduling strategy. The proposed method achieved high prediction accuracy, with a coefficient of determination (R2) of 0.996 5, showing excellent agreement between predicted and actual values. The method effectively captures the complex coupling relationship between the marine environment and structural loads, and its attention weights exhibit good interpretability, providing a reliable intelligent tool for operational decision-making and structural health management of jacket platforms.
关键词
导管架平台 /
结构响应预测 /
序列到序列模型 /
注意力机制 /
门控循环单元 /
深度学习
Key words
jacket platform /
structural response prediction /
sequence-to-sequence model /
attention mechanism /
gated recurrent unit /
deep learning
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] FALTINSEN O M.Sea Loads on Ships and Offshore Structures[M]. Cambridge: Cambridge University Press, 1993.
[2] YAN J, ZHANG Y Y, SU Q, et al.Time Series Prediction Based on LSTM Neural Network for Top Tension Response of Umbilical Cables[J]. Marine Structures, 2023, 91: 103448.
[3] SHI W, HU L H, LIN Z B, et al.Short-Term Motion Prediction of Floating Offshore Wind Turbine Based on Muti-Input LSTM Neural Network[J]. Ocean Engineering, 2023, 280: 114558.
[4] LIM B, ZOHREN S.Time-Series Forecasting with Deep Learning[J]. Philosophical Transactions: Mathematical, Physical and Engineering Sciences, 379(2194): 1-14.
[5] HOCHREITER S, SCHMIDHUBER J.Long Short-Term Memory[J]. Neural Computation, 1997, 9(8): 1735-1780.
[6] WEN Q S, ZHOU T, ZHANG C L, et al.Transformers in Time Series: A Survey[C]//Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence (IJCAI), 2023: 6778-6786.
[7] VASWANI A, SHAZEER N, PARMAR N, et al.Attention is All You Need[C]// Advances in Neural Information Processing Systems 30 (NIPS 2017), 2017: 5998-6008.
[8] ZHOU H Y, ZHANG S H, PENG J Q, et al.Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting[C]// Proceedings of the AAAI Conference on Artificial Intelligence, 2021, 35(12): 11106-11115.
[9] WU H, XU J, WANG J, et al.Autoformer: Decomposition Transformers with Auto-correlation for Long-term Series Forecasting[C]// Advances in Neural Information Processing Systems 34 (NeurIPS 2021), 2021: 22419-22430.
[10] FU D X, WANG L Y, LYU G L, et al.Advances in Dynamic Load Identification Based on Data-Driven Techniques[J]. Engineering Applications of Artificial Intelligence, 2023, 126: 106871.
[11] WAZID M, DAS A K, CHAMOLA V, et al.Uniting Cyber Security and Machine Learning: Advantages, Challenges and Future Research[J]. ICT Express, 2022, 8(3): 313-321.
[12] LIU Y R, WANG L, GU K X.A Support Vector Regression (SVR)-Based Method for Dynamic Load Identification Using Heterogeneous Responses under Interval Uncertainties[J]. Applied Soft Computing, 2021, 110: 107599.
[13] WANG L, LIU Y R, GU K X, et al.A Radial Basis Function Artificial Neural Network (RBF ANN) Based Method for Uncertain Distributed Force Reconstruction Considering Signal Noises and Material Dispersion[J]. Computer Methods in Applied Mechanics and Engineering, 2020, 364: 112954.
[14] LIU Y R, WANG L, GU K X, et al.Artificial Neural Network (ANN) - Bayesian Probability Framework (BPF) Based Method of Dynamic Force Reconstruction under Multi-Source Uncertainties[J]. Knowledge-Based Systems, 2022, 237: 107796.
[15] LUO S Y, JIANG J H, ZHANG F, et al.Distributed Dynamic Load Identification of Beam Structures Using a Bayesian Method[J]. Applied Sciences, 2023, 13(4): 2537.
[16] CUI Z K, WANG C.Multi-Point Vibration Response Prediction Based on Deep Transfer Learning[J]. Journal of Vibration Engineering & Technologies, 2023, 11(3): 1163-1175.
[17] YANG H J, JIANG J H, CHEN G P, et al.A Recurrent Neural Network-Based Method for Dynamic Load Identification of Beam Structures[J]. Materials, 2021, 14(24): 7846.
[18] ZHOU J, DONG L L, GUAN W, et al.Impact Load Identification of Nonlinear Structures Using Deep Recurrent Neural Network[J]. Mechanical Systems and Signal Processing, 2019, 133: 106292.
[19] YANG H J, JIANG J H, CHEN G P, et al.Dynamic Load Identification Based on Deep Convolution Neural Network[J]. Mechanical Systems and Signal Processing, 2023, 185: 109757.
[20] CANDON M, ESPOSITO M, FAYEK H, et al.Advanced Multi-Input System Identification for Next Generation Aircraft Loads Monitoring Using Linear Regression, Neural Networks and Deep Learning[J]. Mechanical Systems and Signal Processing, 2022, 171: 108809.
[21] WANG Y, JIANG W S, WANG C, et al.An Electricity Load Forecasting Model Based on Multilayer Dilated LSTM Network and Attention Mechanism[J]. Frontiers in Energy Research, 2023, 11: 1116465.
[22] BREIMAN L.Random Forests[J]. Machine Learning, 2001, 45(1): 5-32.
[23] FREUND Y, SCHAPIRE R E.A Decision-Theoretic Generalization of On-Line Learning and an Application to Boosting[J]. Journal of Computer and System Sciences, 1997, 55(1): 119-139.
[24] GEURTS P, ERNST D, WEHENKEL L.Extremely Randomized Trees[J]. Machine Learning, 2006, 63(1): 3-42.
[25] KE G L, MENG Q, FINLEY T, et al.LightGBM: A Highly Efficient Gradient Boosting Decision Tree[C]// Neural Information Processing Systems, 2017.
[26] SHI Y, EBERHART R.A Modified Particle Swarm Optimizer[C]// 1998 IEEE International Conference on Evolutionary Computation Proceedings. IEEE World Congress on Computational Intelligence. Anchorage, AK, USA. IEEE, 2002: 69-73.
[27] HE W Y, WANG Y, REN W X.Dynamic Force Identification Based on Composite Trigonometric Wavelet Shape Function[J]. Mechanical Systems and Signal Processing, 2020, 141: 106493.
[28] BAEK S M, PARK J C, JUNG H J.Impact Load Identification Method Based on Artificial Neural Network for Submerged Floating Tunnel under Collision[J]. Ocean Engineering, 2023, 286: 115641.
[29] GUO C Y, JIANG L L, YANG F, et al.Impact Load Identification and Localization Method on Thin-Walled Cylinders Using Machine Learning[J]. Smart Materials and Structures, 2023, 32(6): 065018.
[30] LOSHCHILOV I, HUTTER F.Decoupled Weight Decay Regularization[C]// International Conference on Learning Representations, 2017.
PDF(4161 KB)
