为解决极端环境传感器故障导致监测数据不充足问题,该文采用长短期记忆神经网络（LSTM）智能算法进行数据重构。鉴于海上风电监测数据稀缺,该文的研究依托数值仿真结果开展。基于作者团队开发的“气动-水动-结构-桩土-智能控制”一体化耦合分析软件Zwind,首先开展10 MW大型风力机全工况仿真分析,并通过提取多个高度处的加速度和倾角响应构建数据库,用以模拟多种风力机塔筒传感器故障导致的数据丢失状况。然后基于LSTM建立风力机塔筒加速度和倾角的数据重构模型,训练并验证所构建的数据重构模型的精度。最后在数个未布置传感器的位点上检验LSTM数据重构模型的泛化性能。结果表明：构建的LSTM故障传感器数据重构模型,可基于有限位点的正常服役传感器的监测数据高精度地重构故障传感器以及未测位点的塔筒响应数据;此外,基于倾角响应的重构结果比基于加速度响应的重构结果精度更高。
Abstract
To solve the problem of insufficient monitoring data caused by sensor failures in extreme environments, this study proposes an AI-driven offshore wind power monitoring data reconstruction method. In view of the scarcity of offshore wind turbine monitoring data, this study is conducted using numerical simulations results. Based on the aero-hydro-elasto-servo integrated design software Zwind, developed by the authors' team, this study first carries out numerical simulations of a 10MW wind turbine considering all loading cases, and extracts accelerations and inclinations to form the database for a series of data loss situations caused by sensor failures. Then, a data reconstruction model is proposed based on LSTM using acceleration response or inclination response of wind turbine tower, following by its training and verification. Finally, the generalization performance of the LSTM data reconstruction model is investigated on the locations without sensors. The results show that 1） the proposed data reconstruction model can accurately reconstruct the data of fault sensors on wind turbine tower and predict the responses of unmeasured locations using the monitoring data at limited locations; 2） the reconstruction results based on inclination response are more accurate than acceleration response.
关键词
海上风力机 /
结构健康监测 /
深度学习 /
故障传感器 /
数据重构 /
倾角响应
Key words
offshore wind turbines /
structural health monitoring /
deep learning /
sensor failure /
data reconstruction /
inclination response
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 王立忠, 洪义, 高洋洋, 等. 近海风电结构台风环境动力灾变与控制[J]. 力学学报, 2023, 55(3): 567-587.
WANG L Z, HONG Y, GAO Y Y, et al.Dynamic catastrophe and control of offshore wind power structures in typhoon environment[J]. Chinese journal of theoretical and applied mechanics, 2023, 55(3): 567-587.
[2] 姜贞强, 何奔, 单治钢, 等. 黄海海域极端荷载下海上风力机结构累积变形及疲劳性状: 3种典型基础对比研究[J]. 太阳能学报, 2021, 42(4): 386-395.
JIANG Z Q, HE B, SHAN Z G, et al.Cumulative deformation and fatigue behaviour of offshore wind turbine structure subjected under extreme loading in Yellow Sea: a comparative study between three typical foundations[J]. Acta energiae solaris sinica, 2021, 42(4): 386-395.
[3] 金晓航, 许壮伟, 孙毅, 等. 基于SCADA数据分析和稀疏自编码神经网络的风电机组在线运行状态监测[J]. 太阳能学报, 2021, 42(6): 321-328.
JIN X H, XU Z W, SUN Y, et al.Online condition monitoring for wind turbines based on SCADA data analysis and sparse auto-encoder neural network[J]. Acta energiae solaris sinica, 2021, 42(6): 321-328.
[4] 杨海艳. 测风仪在风力发电场中的应用[J]. 传感器世界, 2017, 23(6): 35-38.
YANG H Y.Application of anemometers in wind farms[J]. Sensor world, 2017, 23(6): 35-38.
[5] 张运洲, 胡泊. “三北” 地区风电开发、输送及消纳研究[J]. 中国电力, 2012, 45(9): 1-6.
ZHANG Y Z, HU B.Study on wind power exploitation, transmission and accommodation in northern region of China[J]. Electric power, 2012, 45(9): 1-6.
[6] ZAHER A, MCARTHUR S D J, INFIELD D G, et al. Online wind turbine fault detection through automated SCADA data analysis[J]. Wind energy, 2009, 12(6): 574-593.
[7] FAN G, LI J, HAO H, et al.Data driven structural dynamic response reconstruction using segment based generative adversarial networks[J]. Engineering structures, 2021, 234: 111970.
[8] 张立奎, 段大猷, 王佐才. 基于LSTM神经网络的多源数据融合桥梁变形重构方法[J]. 土木与环境工程学报(中英文), 2022, 44(3): 37-43.
ZHANG L K, DUAN D Y, WANG Z C.A multi-source data fusion method for bridge displacement reconstruction based on LSTM neural network[J]. Journal of civil and environmental engineering, 2022, 44(3): 37-43.
[9] XUE M Z, LI F N, YU X S, et al.Research and application of EMD-BiLSTM in bridge data reconstruction[C]//2022 China Automation Congress (CAC). Xiamen, China, 2022: 5502-5507.
[10] JING B, PEI Y, QIAN Z, et al.Missing wind speed data reconstruction with improved context encoder network[J]. Energy reports, 2022, 8: 3386-3394.
[11] NABIYAN M S, KHOSHNOUDIAN F, MOAVENI B, et al.Mechanics-based model updating for identification and virtual sensing of an offshore wind turbine using sparse measurements[J]. Structural control and health monitoring, 2021, 28(2): e2647.
[12] 安利强, 孙阳, 王鹏, 等. 考虑随机参数的风电机组台风载荷特征研究[J]. 太阳能学报, 2021, 42(4): 417-423.
AN L Q, SUN Y, WANG P, et al.Study on typhoon load characteristics of wind turbines considering random parameters[J]. Acta energiae solaris sinica, 2021, 42(4): 417-423.
[13] DIMITROV N, GÖÇMEN T. Virtual sensors for wind turbines with machine learning-based time series models[J]. Wind energy, 2022, 25(9): 1626-1645.
[14] HOCHREITER S, SCHMIDHUBER J.Long short-term memory[J]. Neural computation, 1997, 9(8): 1735-1780.
[15] BAK C, ZAHLE F, BITSCHE R, et al.The DTU 10-MW reference wind turbine[C]//Danish Wind Power Research 2013, Trinity, Fredericia, Denmark, 2013.
[16] LAI Y Q, WANG L Z, ZHANG Y H, et al.Site-specific soil reaction model for monopiles in soft clay based on laboratory element stress-strain curves[J]. Ocean engineering, 2021, 220: 108437.
[17] 廖菲, 邓华, 曾琳, 等. 南海北部海面风速概率分布特征[J]. 海洋学报, 2018, 40(5): 37-47.
LIAO F, DENG H, ZENG L, et al.The probability distribution of sea surface wind speeds over the northern South China Sea[J]. Haiyang Xuebao, 2018, 40(5): 37-47.
[18] ISHIHARA T, WANG L L.A study of modal damping for offshore wind turbines considering soil properties and foundation types[J]. Acta oceanologica sinica, 2019, 22(12): 1760-1778.
[19] WANG L, ISHIHARA T.A new FounDyn module in OpenFAST to consider foundation dynamics of monopile supported wind turbines using a site-specific soil reaction framework[J]. Ocean engineering, 2022, 266: 112692.
[20] HONG Y, CHEN X, WANG L, et al.A bounding-surface based cyclic ‘p-y+M-θ'model for unified description of laterally loaded piles with different failure modes in clay[J]. Canadian geotechnical journal, 2024,61(10): 2104-2123.
[21] HONG Y, YAO M, WANG L.A multi-axial bounding surface py model with application in analyzing pile responses under multi-directional lateral cycling[J]. Computers and geotechnics. 2023, 157: 105301.
基金
国家自然科学基金（52238008; 52122906; 51939010）; 海南省财政科技计划（ZDKJ202019）
PDF(1280 KB)
