基于长短期记忆网络的大型漂浮式风力发电机平台运动极短期预报方法

doi:10.16183/j.cnki.jsjtu.2023.S1.27

上海交通大学学报 ›› 2023, Vol. 57 ›› Issue (S1): 37-45.doi: 10.16183/j.cnki.jsjtu.2023.S1.27

基于长短期记忆网络的大型漂浮式风力发电机平台运动极短期预报方法

卫慧¹, 陈鹏²^,³, 张芮菡², 程正顺²^,³()

1.上海勘测设计研究院有限公司,上海 200335
2.上海交通大学海洋工程国家重点实验室, 上海 200240
3.上海交通大学三亚崖州湾深海科技研究院,海南三亚 570025

收稿日期:2023-07-02 修回日期:2023-07-23 接受日期:2023-08-21 出版日期:2023-10-28 发布日期:2023-11-10
通讯作者: 程正顺 E-mail:zhengshun.cheng@sjtu.edu.cn.
作者简介:卫慧,高级工程师,从事海上风电数字化研究.
基金资助:
国家自然科学基金(42176210);国家自然科学基金(52201330)

Ultra-Short-Term Platform Motion Prediction Method of Large Floating Wind Turbines Based on LSTM Network

WEI Hui¹, CHEN Peng²^,³, ZHANG Ruihan², CHENG Zhengshun²^,³()

1. Shanghai Investigation, Design and Research Institute Co., Ltd., Shanghai 200335, China
2. State Key Laboratory of Ocean Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
3. Sanya Yazhou Bay Institute of Deepsea Science and Technology, Shanghai Jiao Tong University, Sanya 570025, Hainan, China

Received:2023-07-02 Revised:2023-07-23 Accepted:2023-08-21 Online:2023-10-28 Published:2023-11-10
Contact: CHENG Zhengshun E-mail:zhengshun.cheng@sjtu.edu.cn.

摘要/Abstract

摘要：

大型漂浮式风力发电机平台运动响应的超前预报是实现主动调载系统控制和智慧运维监测的关键技术.然而,漂浮式风力发电机复杂的工作环境使得仅依靠物理模型和数值仿真方法的极短期预报具有非常大的挑战.因此,提出一种创新的基于长短期记忆神经网络的漂浮式风力发电机平台运动极短期预报方法,并利用实测数据开展了浮式平台纵荡运动的验证与不确定性分析.结果表明,该极短期预报方法可以获得较好的精度,超前60 s预报工作状态下纵荡运动的均方误差最大仅约为1%.该大型漂浮式风力发电机极短期运动响应预报能够为未来漂浮式风电场的智慧运维提供扎实的技术支撑.

关键词: 大型漂浮式风力发电机, 极短期预报, 长短期记忆网络, 不确定性

Abstract:

The motion prediction of large floating wind turbine platforms is the key technology to realize the control of active ballast systems and intelligent operation and maintenance monitoring. However, the complex environment of floating wind turbines makes ultra-short-term predictions that only rely on physical models and numerical simulation methods very challenging. Therefore, this paper proposes an innovative ultra-short-term prediction method for floating wind turbine platform motion based on the long-short-term memory (LSTM) neural network. Measured data have been used to verify the feasibility and uncertainty of this method in terms of surge motion. The results show that the ultra-short-term prediction method proposed in this paper can obtain a better accuracy. For example, the maximum mean square error of surge motion prediction in the 60 s under working condition is only about 1%. The ultra-short-term motion prediction of large floating wind turbines proposed in this paper provides solid technical support for future intelligent operation and maintenance of floating wind farms.

Key words: large floating wind turbines, ultra-short-term prediction, long-short-term memory (LSTM) network, uncertainty

中图分类号:

TP319
TK 89

卫慧, 陈鹏, 张芮菡, 程正顺. 基于长短期记忆网络的大型漂浮式风力发电机平台运动极短期预报方法[J]. 上海交通大学学报, 2023, 57(S1): 37-45.

WEI Hui, CHEN Peng, ZHANG Ruihan, CHENG Zhengshun. Ultra-Short-Term Platform Motion Prediction Method of Large Floating Wind Turbines Based on LSTM Network[J]. Journal of Shanghai Jiao Tong University, 2023, 57(S1): 37-45.

图/表 11

图1

图2

图3

图4

表1

表2

图5

图6

图7

表3

图8

参考文献 18

[1]	温斌荣, 田新亮, 李占伟, 等. 大型漂浮式风电装备耦合动力学研究: 历史、进展与挑战[J]. 力学进展, 2022, 52(4): 731-808.
	WEN Binrong, TIAN Xinliang, LI Zhanwei, et al. Coupling dynamics of floating wind turbines: History, progress and challenges[J]. Advances in Mechanics, 2022, 52(4): 731-808.
[2]	CHENG Z S, MADSEN H A, GAO Z, et al. A fully coupled method for numerical modeling and dynamic analysis of floating vertical axis wind turbines[J]. Renewable Energy, 2017, 107: 604-619. doi: 10.1016/j.renene.2017.02.028 URL
[3]	HEGSETH J M, BACHYNSKI E E. A semi-analytical frequency domain model for efficient design evaluation of spar floating wind turbines[J]. Marine Structures, 2019, 64: 186-210. doi: 10.1016/j.marstruc.2018.10.015 URL
[4]	CHENG P, HUANG Y, WAN D C. A numerical model for fully coupled aero-hydrodynamic analysis of floating offshore wind turbine[J]. Ocean Engineering, 2019, 173: 183-196. doi: 10.1016/j.oceaneng.2018.12.021 URL
[5]	陈曦. 海上浮式风机系统在波浪上的运动响应分析[D]. 上海: 上海交通大学, 2018.
	CHEN Xi. Motion response analysis of offshore floating fan system on waves[D]. Shanghai: Shanghai Jiao Tong University, 2018.
[6]	朱仁传, 缪国平, 范菊, 等. 海上浮式风力机及其动力学问题[J]. 应用数学和力学, 2013, 34(10): 1110-1118.
	ZHU Renchuan, MIAO Guoping, FAN Ju, et al. Offshore floating wind turbines and related dynamic problems[J]. Applied Mathematics and Mechanics, 2013, 34(10): 1110-1118.
[7]	盖晓娜, 杨建民, 田新亮. 基于小波-SVR模型的浮体极短期运动预报方法[J]. 舰船科学技术, 2018, 40(11): 66-70.
	GAI Xiaona, YANG Jianmin, TIAN Xinliang. A method for very short-term motion prediction of floaters based on Wavelet-SVR model[J]. Ship Science and Technology, 2018, 40(11): 66-70.
[8]	ZHANG G Q, EDDY PATUWO B, HU M Y. Forecasting with artificial neural networks:[J]. International Journal of Forecasting, 1998, 14(1): 35-62. doi: 10.1016/S0169-2070(97)00044-7 URL
[9]	郝立柱, 韩阳, 潘子英. 循环神经网络方法预报船舶操纵运动研究[C]\\第三十一届全国水动力学研讨会论文集(下册). 上海: 海洋出版社, 2020: 1017-1029.
	Hao Lizhu, Han Yang, Pan Ziying. A study on predicting ship handling movements using recurrent neural network method[C]\\Proceedings of the 31 st National Symposium on Hydrodynamics (Volume II). Shanghai, China: China Ocean Press, 2020: 1017-1029.
[10]	ZENG Z G, CHEN G H. Multi-step predictions for generalized heave motion of wave compensating platform based on ELMAN neural network[C]\\2009 Third International Symposium on Intelligent Information Technology Application Workshops. Nanchang, China: IEEE, 2009: 460-463.
[11]	HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780. doi: 10.1162/neco.1997.9.8.1735 pmid: 9377276
[12]	CHO K, VAN MERRIENBOER B, GULCEHRE C, et al. Learning phrase representations using RNN encoder-decoder for statistical machine translation[C]\\Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing. Stroudsburg, PA, USA: Association for Computational Linguistics, 2014: 1724-34.
[13]	刘飞飞. 浮式风机水动力响应与系泊系统疲劳损伤及深度学习短期运动预报研究[D]. 大连: 大连理工大学, 2022.
	LIU Feifei. Study on hydrodynamic response of floating fan and fatigue damage of mooring system and short-term motion prediction of deep learning[D]. Dalian: Dalian University of Technology, 2022.
[14]	QIAO D S, LI P, MA G, et al. Realtime prediction of dynamic mooring lines responses with LSTM neural network model[J]. Ocean Engineering, 2021, 219: 108368. doi: 10.1016/j.oceaneng.2020.108368 URL
[15]	GAO D W, ZHU Y S, ZHANG J F, et al. A novel MP-LSTM method for ship trajectory prediction based on AIS data[J]. Ocean Engineering, 2021, 228: 108956. doi: 10.1016/j.oceaneng.2021.108956 URL
[16]	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. doi: 10.1016/j.oceaneng.2023.114558 URL
[17]	GERS F A, SCHMIDHUBER J, CUMMINS F. Learning to forget: Continual prediction with LSTM[J]. Neural Computation, 2000, 12(10): 2451-2471. pmid: 11032042
[18]	CHEN P, JIA C J, NG C, et al. Application of SADA method on full-scale measurement data for dynamic responses prediction of Hywind floating wind turbines[J]. Ocean Engineering, 2021, 239: 109814. doi: 10.1016/j.oceaneng.2021.109814 URL

参变量	取值	参变量	取值
窗口长度	20	批量大小	50
网络层数	2	网络节点数	64
最大训练轮数	1000	最小训练轮数	100
初始学习率	0.01	C₀	0.3

编号	时间周期	H_s/m	T_p/s	风速/(m·s^-1)	流速/(m·s^-1)	浪向/(°)	风向/(°)	流向/(°)
1	2018-01-06T08:00	4.4	10.9	13.7	0.21	17	11	19
2	2018-01-09T09:40	3.2	9.3	13.9	0.09	131	146	303
3	2018-01-14T15:40	4.2	8.7	20.4	0.32	165	174	24
4	2018-02-13T01:20	2.1	6.5	15.5	0.27	201	174	187
5	2018-03-26T23:30	2.2	10.6	8.5	0.24	14	172	16
6	2018-04-14T00:40	2.1	10.5	5.2	0.32	107	171	195
7	2018-05-02T04:00	2.3	6.5	15.6	0.12	185	175	233
8	2018-07-29T04:00	3.0	7.9	16.6	0.33	161	179	44
9	2018-02-24T04:50	2.5	7.3	14.1	0.17	164	161	150

预报时长/s	测试集	e_MSE×10²		e_MAPE×10²		R² ×10²		e_ONE×10²
10	工作	4.2±	0.0006	5.59±	0.0043	17.34±	0.1200	3.30±	0.0105
	停机	8.6±	0.0005	14.34±	0.0055	53.20±	0.0266	5.51±	0.0155
30	工作	5.7±	0.0007	6.67±	0.0044	-15.69±	0.1457	4.36±	0.0079
	停机	7.4±	0.0014	12.32±	0.0120	62.73±	0.0849	5.47±	0.0175
60	工作	10.9±	0.0022	9.25±	0.0099	-119.88±	0.4546	7.19±	0.0133
	停机	16.6±	0.0021	19.39±	0.0114	-0.846±	0.1514	10.18±	0.0120

基于长短期记忆网络的大型漂浮式风力发电机平台运动极短期预报方法

Ultra-Short-Term Platform Motion Prediction Method of Large Floating Wind Turbines Based on LSTM Network

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 11

参考文献 18

相关文章 15

编辑推荐

Metrics

本文评价

[1]	陈熠, 王晗, 徐潇源, 胡友琳, 严正, 曾丹, 冯凯. 省间-省内两级市场协调下两阶段分布鲁棒经济调度模型[J]. 上海交通大学学报, 2023, 57(9): 1114-1125.
[2]	王精, 邢海军, 王华昕, 彭思佳. 考虑电动汽车及负荷聚合商参与的综合能源系统优化调度[J]. 上海交通大学学报, 2023, 57(7): 814-823.
[3]	胡博, 程欣, 邵常政, 黄威, 孙悦, 谢开贵. 计及建筑热负荷弹性的综合能源系统调度方法[J]. 上海交通大学学报, 2023, 57(7): 803-813.
[4]	尚凡成, 李传庆, 詹可, 朱仁传. 改进LSTM神经网络在极短期波浪时序预报中的应用[J]. 上海交通大学学报, 2023, 57(6): 659-665.
[5]	李琰, 肖龙飞, 魏汉迪, 寇雨丰. 基于长短期记忆网络的半潜平台波浪爬升预报[J]. 上海交通大学学报, 2023, 57(2): 161-167.
[6]	邵萍, 杨之乐, 李慷, 朱晓东. 基于用户意愿的电动汽车备用容量多目标优化[J]. 上海交通大学学报, 2023, 57(11): 1501-1511.
[7]	黄琬迪, 张沈习, 程浩忠, 陈丹, 翟晓萌, 吴霜. 考虑地区发展阶段不确定性的电网投资决策鲁棒优化[J]. 上海交通大学学报, 2023, 57(11): 1455-1464.
[8]	. 行人轨迹预测的动作感知编码器–解码器网络[J]. J Shanghai Jiaotong Univ Sci, 2023, 28(1): 20-27.
[9]	方晓涛, 严正, 王晗, 徐潇源, 陈玥. 考虑概率电压不平衡度越限风险的共享储能优化运行方法[J]. 上海交通大学学报, 2022, 56(7): 827-839.
[10]	姜淏予, 王沛伦, 葛泉波, 徐今强, 罗朋, 姚刚. 漂浮式光伏网格对海上天气突变的感知方法[J]. 上海交通大学学报, 2022, 56(12): 1584-1597.
[11]	张俊涛, 刘晓晶, 张滕飞, 柴翔. 子通道程序对PSBT空泡分布实验计算的不确定性量化分析[J]. 上海交通大学学报, 2022, 56(10): 1420-1426.
[12]	冯静，马建. 云动力过程及其辐射效应将如何影响未来全球变暖？[J]. 上海交通大学学报, 2021, 55(Sup.1): 69-71.
[13]	武光利, 郭振洲, 李雷霆, 王成祥. 融合FCN和LSTM的视频异常事件检测[J]. 上海交通大学学报, 2021, 55(5): 607-614.
[14]	钟光耀, 邰能灵, 黄文焘, 李然, 傅晓飞, 纪坤华. 基于多维聚类的配变负荷注意力短期预测方法[J]. 上海交通大学学报, 2021, 55(12): 1532-1543.
[15]	姜宇迪, 胡晖, 殷跃红. 基于无监督迁移学习的电梯制动器剩余寿命预测[J]. 上海交通大学学报, 2021, 55(11): 1408-1416.