Research on Data Repair of Pile-Type Adjustable Wind Turbine Foundation Monitoring Based on FST-ATTNet

WEI Huanwei; ZHAO Jizhang; ZHENG Xiao; TAN Fang; LIU Cong

doi:10.12454/j.jsuese.202400966

您当前的位置：

首页 >

文章列表页 >

Research on Data Repair of Pile-Type Adjustable Wind Turbine Foundation Monitoring Based on FST-ATTNet

更新时间：2025-05-20

- Research on Data Repair of Pile-Type Adjustable Wind Turbine Foundation Monitoring Based on FST-ATTNet
- Advanced Engineering Sciences Pages: 1-13(2025)
- 作者机构：
  
  1.山东建筑大学土木工程学院，山东济南 250101
  2.山东建筑大学建筑结构加固改造与地下空间工程教育部重点实验室，山东济南 250101
  3.山东建筑大学地铁保护研究所，山东济南 250101
- 作者简介：
- 基金信息：
- DOI：10.12454/j.jsuese.202400966
  CLC： TU476;TP183
- Received：19 November 2024，
  
  Revised：21 April 2025，
  
  Published Online：20 May 2025，
- 稿件说明：
移动端阅览
魏焕卫,赵继璋,郑晓等.基于FST-ATTNet的桩式可调风机基础监测数据修复研究[J].工程科学与技术,

WEI Huanwei,ZHAO Jizhang,ZHENG Xiao,et al.Research on Data Repair of Pile-Type Adjustable Wind Turbine Foundation Monitoring Based on FST-ATTNet[J].Advanced Engineering Sciences,XXXX,XX(XX):1‒13.
魏焕卫,赵继璋,郑晓等.基于FST-ATTNet的桩式可调风机基础监测数据修复研究[J].工程科学与技术, DOI：10.12454/j.jsuese.202400966.

WEI Huanwei,ZHAO Jizhang,ZHENG Xiao,et al.Research on Data Repair of Pile-Type Adjustable Wind Turbine Foundation Monitoring Based on FST-ATTNet[J].Advanced Engineering Sciences,XXXX,XX(XX):1‒13. DOI： 10.12454/j.jsuese.202400966.

摘要

在复杂的监测系统中，数据缺失显著影响分析结果的准确性。为了解决风机可调基础监测数据中的连续缺失问题，本文提出了一种基于频域、时域和空间信息深度融合的Frequency-Spatial-Temporal Attention Network （FST-ATTNet），旨在提升复杂时间序列数据的建模能力并解决监测数据连续缺失的问题。该模型通过有机融合多种注意力机制，分别在频域、时域和空间领域进行特征提取与增强，从而显著提升时间序列模式的捕捉和预测能力，有效修复监测缺失数据。并基于实测数据进行了验证分析，得出以下结论：（1）与LSTM、GRU、BiGRU、TCN和Transformer等传统序列模型相比，FST-ATTNet在数据修复实验中表现最佳，显著优于其他模型。（2）在缺失量测试中，即使在40%的数据监测缺失情况下，FST-ATTNet依然能够保持较高修复精度；在特征缺失测试中，在25%数据缺失情况下，面对全部缺失特征仍有较高的修复精度。（3）通过消融实验，证明了FST-ATTNet结合频域、时域、空间信息和注意力机制，能够显著提升模型在高缺失数据下的修复精度。（4）FST-ATTNet不仅在桩式可调风机基础监测数据修复任务中表现出色，在另一风机基础锚索监测数据缺失率高达50%的情况下，仍展示出优异的效果。该方法可作为风机基础健康监测数据修复的可靠方法，同时为类似问题提供了新的解决思路。

Abstract

Objective

Wind energy plays a crucial role in achieving sustainable energy goals. As a critical structural component

the wind turbine foundation significantly influences the operational stability

safety

and long-term performance of wind turbine systems. However

structural health monitoring (SHM) of wind turbine foundations often faces challenges with data integrity due to environmental factors

sensor malfunctions

or data transmission issues. These missing data can severely impact the accuracy of structural health assessments

thereby affecting maintenance decisions and operational safety. To tackle the persistent data gaps in the monitoring system of adjustable wind turbine foundations

this study introduces a frequency-space-time domain attention network (FST-ATTNet). This model is designed to enhance the modeling capabilities for complex time-series data

improve the accuracy of missing data reconstruction

and ensure the reliability of health monitoring

ultimately guaranteeing wind turbines' safe and efficient operation. Moreover

it presents potential solutions for similar data reconstruction challenges across various engineering disciplines.

Methods

The FST-ATTNet model introduces an innovative data repair framework by integrating features from the frequency

time

and spatial domains. In the frequency domain

the model employs discrete cosine transform (DCT) to extract periodic and global patterns from time series data

effectively mitigating the high-frequency noise caused by the Gibbs phenomenon in traditional discrete Fourier transform (DFT). A frequency-domain attention mechanism is introduced to enhance this process

adaptively assigning weights to frequency components and prioritizing those most relevant for data reconstruction. In the time domain

Bidirectional Gated Recurrent Units (BiGRU) capture both forward and backward dependencies within the time series

ensuring a comprehensive understanding of local sequence patterns. The Kolmogorov-Arnold Network (KAN) incorporates a B-spline activation function

further enhancing the model's ability to capture complex nonlinear temporal changes. In the spatial domain

the Temporal Convolutional Network (TCN) models long-range dependencies by expanding causal convolutions

thereby capturing local and global spatial relationships. The Squeeze-and-Excitation Network (SENet) further boosts spatial feature extraction by dynamically adjusting the importance of different feature channels. By combining these various attention mechanisms

FST-ATTNet successfully integrates frequency

time

and spatial domain features

achieving superior modeling of complex time series patterns and robust reconstruction of missing data. The model is validated using monitoring data of the measured strain on a pile-based adjustable wind turbine foundation

and its performance is evaluated using the coefficient of determination (

²) and mean squared error (

MSE

) metrics.Results and Discussions　Validation experiments based on measured data show that FST-ATTNet has the following advantages: (1) Superior performance compared to traditional models: FST-ATTNet outperforms traditional sequence models in data reconstruction tasks

including Long Short-Term Memory (LSTM)

Gated Recurrent Unit (GRU)

Bidirectional GRU (Bi

GRU)

Temporal Convolutional Network (TCN)

and Transformer

achieving excellent results with

² of 0.98 and

MSE

of 0.66. In contrast

the

² of LSTM and GRU models is only 0.86

and the

MSE

is 2.35 and 2.34

respectively

which are limited by their unidirectional or local feature extraction capabilities. The Transformer model performs the worst

with

² of 0.72 and

MSE

as high as 2.77

likely due to its inability to effectively capture local patterns and frequency domain features in multivariate time series data. FST-ATTNet

through deep integration of frequency

time

and spatial information

can capture complex patterns

including periodicity

local dynamics

and long-term dependencies

significantly improving reconstruction accuracy. (2) Robustness in high missing-rate scenarios: The model excels in handling severe data-missing scenarios. In the continuous data missing test

even with a missing rate as high as 40%

FST-ATTNet maintains high reconstruction accuracy with

² of 0.94 and

MSE

of 0.95. At the 45% missing rate

performance slightly declines

with

² of 0.87 and

MSE

of 1.31

but still outperforms other models. In the feature missing test

when 25% of the monitoring features are entirely missing

the model achieves

² of 0.91 and

MSE

of 0.75

demonstrating its ability to handle complex multi-feature missing scenarios commonly found in actual monitoring systems. (3) Insights from ablation experiments: Ablation experiments provide key insights into the contribution of each component of FST-ATTNet. After removing the frequency-domain enhanced attention mechanism

² decreases to 0.92

and

MSE

increases to 1.13. After removing SENet

² is 0.91

and

MSE

is 1.22

indicating that these attention mechanisms play a crucial role in feature enh

ancement. Removing all attention mechanisms results in further performance degradation

with

² of 0.90 and

MSE

of 1.20

highlighting their importance in the prioritization of selective features. Removing the KAN network results in

² of 0.92 and

MSE

of 1.09

indicating its contribution to modeling complex time series patterns. Using only frequency domain

time domain

or spatial information results in significant performance drops

with

² of 0.75

0.68

and 0.86

respectively

and

MSE

of 1.98

2.44

and 1.33

indicating the necessity of integrating frequency

time

or spatial information. Spatial information is especially critical in high-missing scenarios. (4) Applicability of the model: To evaluate the model's applicability

FST-ATTNet was applied to anchor cable monitoring data with a missing rate of up to 50%

achieving excellent results with

² of 0.92 and

MSE

of 0.80. The model achieved near-perfect reconstruction for datasets with strong periodicity at a 25% missing rate (

² = 0.97

MSE

= 0.40). However

performance slightly declined at a 50% missing rate (

² = 0.92

MSE

= 0.80)

with deviations at the peaks primarily due to the training data not fully covering the entire cycle. Nonetheless

FST-ATTNet demonstrates adaptability across different monitoring scenarios and a unique ability to handle cyclic patterns in periodic data reconstruction.

Conclusions

The FST-ATTNet model offers a reliable and robust solution to the problem of continuous data loss in the health monitoring of pile-type adjustable wind turbine foundations. By deeply integrating frequency

time

and spatial domain information and incorporating advanced attention mechanisms

the model achieves exceptional reconstruction accuracy in high missing-rate scenarios

significantly outperforming traditional sequence models. Furthermore

the successful application of the model to other monitoring datasets (such as anchor cable data) demonstrates its versatility and broad applicability in structural health monitoring. FST-ATTNet not only enhances the reliability of wind turbine foundation monitoring but also provides innovative solutions to similar data repair challenges in other engineering domains

offering crucial support for the safety and efficiency of wind turbine systems.

关键词

Keywords

references

Zhao Yan , Pan Jianing , Huang Zhuye , et al . Analysis of vibration monitoring data of an onshore wind turbine under different operational conditions [J ] . Engineering Structures , 2020 , 205 : 110071 . DOI: 10.1 016/j.engstruct.2019.110071 http://dx.doi.org/10.1016/j.engstruct.2019.110071

Yang Yang , Liang Fayun , Zhu Qingxin , et al . An Overview on Structural Health Monitoring and Fault Diagnosis of Offshore Wind Turbine Support Structures [J ] . Journal of Marine Science and Engineering , 2024 , 12 ( 3 ): 377 . DOI: 10.3390/jmse120 3077 http://dx.doi.org/10.3390/jmse1203077

Bertsimas D , Pawlowski C , Zhuo Y D . From predictive methods to missing data imputation: an optimization approach [J ] . J. Mach. Learn. Res. , 2017 , 18 ( 1 ): 7133 - 7171 . URL: http://jmlr.org/papers/v18/17-073.html http://jmlr.org/papers/v18/17-073.html

Shahbazi Hossein , Karimi Sajjad , Hosseini Vahid , et al . A novel regression imputation framework for Tehran air pollution monitoring network using outputs from WRF and CAMx models [J ] . Atmospheric Environment , 2018 , 187 : 24 - 33 . DOI: 10.1016/j.atmos env.2018.05.055 http://dx.doi.org/10.1016/j.atmosenv.2018.05.055

Chen Juan , Wang Xianyu , Luo Lingling , et al . Comparison of the effectiveness of missing data imputation: Machine learning vs. statistical learning [J ] . Statistics and Decision , 2020 , 36 ( 17 ): 28 - 32 . DOI: 10.13546/j.cnki.tjyjc.2020.17.006 http://dx.doi.org/10.13546/j.cnki.tjyjc.2020.17.006

陈娟 , 王献雨 , 罗玲玲 , 等 . 缺失值填补效果: 机器学习与统计学习的比较 [J ] . 统计与决策 , 2020 , 36 ( 17 ): 28 - 32 . DOI: 10.13546/j.cnki.tjyjc.2020.17.006 http://dx.doi.org/10.13546/j.cnki.tjyjc.2020.17.006

Zhao Wei , Mao Jixin , Guan Jianchao , et al . Comparison and application of the ANN model with piecewise linear interpolation and regression models [J/OL ] . Journal of Sediment Research , 1 - 7 . DOI: 10.1 6239/j.cnki.0468-155x.2024.04.012 http://dx.doi.org/10.1 6239/j.cnki.0468-155x.2024.04.012

赵伟 , 毛继新 , 关见朝 , 等 . ANN模型与分段线性插值及回归模型的比较及应用 [J/OL ] . 泥沙研究 , 1 - 7 DOI: 10.1 6239/j.cnki.0468-155x.2024.04.012 http://dx.doi.org/10.1 6239/j.cnki.0468-155x.2024.04.012

Wei Huanwei , Song Zhixin , Lei Shuli , et al . Research on health monitoring data repair of wind turbine foundations based on residual network [J ] . Acta Energiae Solaris Sinica , 2024 , 45 ( 04 ): 143 - 150 . DOI: 10.19912/j.0254-0096.tynxb.2 022-1884 http://dx.doi.org/10.19912/j.0254-0096.tynxb.2022-1884

魏焕卫 , 宋志鑫 , 雷树立 , 等 . 基于残差网络的风电机组基础健康监测数据修复研究 [J ] . 太阳能学报 , 2024 , 45 ( 04 ): 143 - 150 . DOI: 10.19912/j.0254-0096.tynxb.2 022-1884 http://dx.doi.org/10.19912/j.0254-0096.tynxb.2022-1884

Wijesuriya Rushani , Moreno-Betancur Margarita , Carlin John B. , et al . Evaluation of approaches for accommodating interactions and non-linear terms in multiple imputation of incomplete three-level data [J ] . Biometrical Journal , 2022 , 64 ( 8 ): 1404 - 1425 . DOI: 10. /bimj.202000343 http://dx.doi.org/10./bimj.202000343

Sterne Jonathan AC , White Ian R. , Carlin John B. , et al . Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls [J ] . BMJ , 2009 , 338 . DOI: 10.1136/bmj.b2393 http://dx.doi.org/10.1136/bmj.b2393

Wang Jun , Du Wenjie , Cao Wei , et al . Deep learning for multivariate time series imputation: A survey . arXiv preprint arXiv :2402. 04059 , 2024 . DOI: 10.4855 0/arXiv.2402.04059 http://dx.doi.org/10.48550/arXiv.2402.04059

Xing Cheng , Zhang Chunyu , Zhang Min , et al . Data Augmentation Based on Time-series Generative Adversarial Networks for Failure Prediction in Optical Network [J ] . Journal of Lightwave Technology , 2025 , 43 ( 7 ): 1234 - 1245 . DOI: 10.1109/JLT.2025.3556307 http://dx.doi.org/10.1109/JLT.2025.3556307

Cai Zongwu , Fan Jianqing , Yao Qiwei . Functional-coefficient regression models for nonlinear time series [J ] . Journal of the American Statistical Association , 2000 , 95 ( 451 ): 941 - 956 . DOI: 10.1080/01 621459.2000.10474284 http://dx.doi.org/10.1080/01621459.2000.10474284

Esling Philippe , Agon Carlos . Time-series data mining [J ] . ACM Computing Surveys , 2012 , 45 ( 1 ): 12 . DOI: 10.1145/2379776.2379788 http://dx.doi.org/10.1145/2379776.2379788

Baidya Ranjai , Lee Sang-Woong . Addressing the Non-Stationarity and Complexity of Time Series Data for Long-Term Forecasts [J ] . Applied Sciences , 2024 , 14 ( 11 ): 4436 . DOI: 10.3390/app14114436 http://dx.doi.org/10.3390/app14114436

Ahn Hyun , Sun Kyunghee , Kim K. Pio . Comparison of missing data imputation methods in time series forecasting [J ] . Computers , Materials & Continua, 2022 , 70 ( 1 ): 767 - 779 . DOI: 10.32604/cmc. 2022.019369 http://dx.doi.org/10.32604/cmc.2022.019369

Kazijevs Maksims , Samad Manar D . Deep imputation of missing values in time series health data: A review with benchmarking [J ] . Journal of Biomedical Informatics , 2023 , 104440 . DOI: 10.1016/ j.jbi.2023.104440 http://dx.doi.org/10.1016/j.jbi.2023.104440

Foster J. , Richards F . B. The Gibbs phenomenon for piecewise-linear approximation [J ] . The American Mathematical Monthly , 1991 , 98 ( 1 ): 47 - 49 . DOI: 10.10 80/00029890.1991.11995703 http://dx.doi.org/10.1080/00029890.1991.11995703

Raj Sandeep , Ray Kailash Chandra . ECG Signal Analysis Using DCT-Based DOST and PSO Optimized SVM [J ] . IEEE Transactions on Instrumentation and Measurement , 2017 , 66 ( 3 ): 470 - 478 . DOI: 10.1109/TI M.2016.2642758 http://dx.doi.org/10.1109/TIM.2016.2642758

Jiang Maowei , Zeng Pengyu , Wang Kai , et al . FECAM: Frequency enhanced channel attention mechanism for time series forecasting [J ] . Advanced Engineering Informatics , 2023 , 58 : 102158 . DOI: 10.1 016/j.aei.2023.102158 http://dx.doi.org/10.1016/j.aei.2023.102158

Fu Yang , Ren Zixu , Wei Shurong , et al . Ultra-short-term power prediction for offshore wind farms based on improved LSTM-TCN model [J ] Proceedings of the CSEE 2022 , 42 ( 12 ): 4292 - 4303 . DOI: 10.13334/j.0258-80 13.pcsee.210724 http://dx.doi.org/10.13334/j.0258-8013.pcsee.210724

符杨 , 任子旭 , 魏书荣 , 等 . 基于改进LSTM-TCN模型的海上风电超短期功率预测 [J ] . 中国电机工程学报 , 2022 , 42 ( 12 ): 4292 - 4303 . DOI: 10.13334/j.0258-80 13.pcsee.210724 http://dx.doi.org/10.13334/j.0258-8013.pcsee.210724

Ju Hanwen , Deng Yang , Li Aiqun . Structural anomaly monitoring data recovery method based on GRU neural network [J ] . Vibration and Shock , 2023 , 42 ( 09 ): 328 - 338 . DOI: 10.13465/j.cnki. Jvs.2023.09.038 http://dx.doi.org/10.13465/j.cnki.Jvs.2023.09.038

鞠翰文 , 邓扬 , 李爱群 . 基于GRU神经网络的结构异常监测数据修复方法 [J ] . 振动与冲击 , 2023 , 42 ( 09 ): 328 - 338 . DOI: 10.13465/j.cnki. Jvs.2023.09.038 http://dx.doi.org/10.13465/j.cnki.Jvs.2023.09.038

Liu Ziming , Wang Yixuan , Vaidya Sachin , et al . Kan: Kolmogorov-arnold networks . arXiv preprint arXiv :2404. 19756 , 2024 . DOI: 10.48550/arXiv.2404.1 9756 http://dx.doi.org/10.48550/arXiv.2404.19756

He Kaiming , Zhang Xiangyu , Ren Shaoqing , et al . Deep residual learning for image recognition . In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2016 : 770 - 778 . DOI: 1 0.1109/CVPR.2016.90 http://dx.doi.org/10.1109/CVPR.2016.90

Xing Chen , Zhang Zhaobei . Short-term probabilistic forecasting method of photovoltaic output power based on improved temporal convolutional network [J ] . Acta Energiae Solaris Sinica , 2023 , 44 ( 2 ): 373 - 380 . DOI: 10.19912/j.0254-0096. ty nxb.2021-1033 http://dx.doi.org/10.19912/j.0254-0096.tynxb.2021-1033

邢晨 , 张照贝 . 基于改进时间卷积网络的短期光伏出力概率预测方法 [J ] . 太阳能学报 , 2023 , 44 ( 2 ): 373 - 380 . DOI: 10.19912/j.0254-0096. ty nxb.2021-1033 http://dx.doi.org/10.19912/j.0254-0096.tynxb.2021-1033

Hu Jie , Shen Li , Sun Gang . Squeeze-and-excitation networks . In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2018 : 7132 - 7141 . DOI: 10.1109/CVPR.2 018.00745 http://dx.doi.org/10.1109/CVPR.2018.00745

Wei Huanwei , Zhang Wei , Tan Fang , et al . A pile-type adjustable wind turbine foundation and construction method : CN202111192089.2 [P ] . CN 113 833012A .

魏焕卫 , 张伟 , 谭芳 , 等 . 一种桩式可调风机基础及施工方法 : CN202111192089.2 [P ] . CN 11 3833012A .

Lee Deuk-Hwan , Cheon Enok , Lim Hwan-Hui , Choi Shin-Kyu , Kim Yun-Tae , Lee Seung-Rae . An artificial neural network model to predict debris-flow volumes caused by extreme rainfall in the central region of South Korea [J ] . Engineering Geology , 2021 , 281 : 105979 . DOI: 10.1016/j.enggeo.2020.105979 http://dx.doi.org/10.1016/j.enggeo.2020.105979

Lin Tao , Liu Hangpeng , Zhao Cancan , et al . Short-term wind speed prediction based on SSA-PSO-ANFIS [J ] . Acta Energiae Solaris Sinica , 2021 , 42 ( 03 ): 128 - 134 . DOI: 10.19912/j.02540096. tynxb.2018-1184 http://dx.doi.org/10.19912/j.02540096.tynxb.2018-1184

林涛 , 刘航鹏 , 赵参参 , 等 . 基于SSA-PSO-ANFIS的短期风速预测研究 [J ] . 太阳能学报 , 2021 , 42 ( 03 ): 128 - 134 . DOI: 10.19912/j.02540096. tynxb.2018-1184 http://dx.doi.org/10.19912/j.02540096.tynxb.2018-1184

Wei Huanwei , Lei Shuli , Song Zhixin , et al . Study on the field monitoring of the mechanical performance of rock-anchored reinforced foundations for wind turbines [J/OL ] . Journal of Civil and Environmental Engineering (Chinese-English) , 1 - 9 . DOI: 10.11835/j. issn.2096-6717.2023.089 http://dx.doi.org/10.11835/j.issn.2096-6717.2023.089

魏焕卫 , 雷树立 , 宋志鑫 , 等 . 风电机组岩石锚杆式加固基础受力性能现场监测研究 [J/OL ] . 土木与环境工程学报(中英文) , 1 - 9 . DOI: 10.11835/j. issn.2096-6717.2023.089 http://dx.doi.org/10.11835/j.issn.2096-6717.2023.089

Views

下载量

CNKI被引量

Alert me when the article has been cited

提交

Tools

Publicity Resources

Aircraft Multi-stage Altitude Prediction Under Satellite Signal Loss

Evaluating Rust Grades Based on Attention Mechanism and Bilinear Pooling

Classification and Rating of Scrap Steel Based on Deep Learning

TCS-Net: A Tiny Crack Segmentation Network for Nuclear Containment Vessel

P2P Botnet Detection Method Based on Graph Neural Network

Related Author

Qiang MIAO

Mengchan HUANG

Haifei DONG

Baojia CHEN

Yunpeng YANG

Ruixue PAN

Fafa CHEN

Yunfeng XU

Related Institution

School of Electrical Engineering, Sichuan University

Hubei Key Lab. of Hydroelectric Machinery Design & Maintenance, China Three Gorges Univ.

National Dam Safety Research Center

College of Info. Sci.and Eng., Hebei Univ. of Sci. and Technol.

College of Metallurgy and Energy, North China Univ. of Sci. and Technol. Univ.

⁰