Wind and Structures

Techno-Press (테크노프레스)
권호 표지
DOI QR코드

DOI QR Code

Simulation combined transfer learning model for missing data recovery of nonstationary wind speed

  • Qiushuang Lin (College of Architecture and Civil Engineering, Xinyang Normal University) ;
  • Xuming Bao (Institute of Structural Engineering, Zhejiang University) ;
  • Ying Lei (School of Architecture and Civil Engineering, Xiamen University) ;
  • Chunxiang Li (School of Mechanics and Engineering Science, Shanghai University)
  • Received : 2023.02.28
  • Accepted : 2023.08.27
  • Published : 2023.11.25
⟨ Previous Next ⟩

Abstract

In the Structural Health Monitoring (SHM) system of civil engineering, data missing inevitably occurs during the data acquisition and transmission process, which brings great difficulties to data analysis and poses challenges to structural health monitoring. In this paper, Convolution Neural Network (CNN) is used to recover the nonstationary wind speed data missing randomly at sampling points. Given the technical constraints and financial implications, field monitoring data samples are often insufficient to train a deep learning model for the task at hand. Thus, simulation combined transfer learning strategy is proposed to address issues of overfitting and instability of the deep learning model caused by the paucity of training samples. According to a portion of target data samples, a substantial quantity of simulated data consistent with the characteristics of target data can be obtained by nonstationary wind-field simulation and are subsequently deployed for training an auxiliary CNN model. Afterwards, parameters of the pretrained auxiliary model are transferred to the target model as initial parameters, greatly enhancing training efficiency for the target task. Simulation synergy strategy effectively promotes the accuracy and stability of the target model to a great extent. Finally, the structural dynamic response analysis verifies the efficiency of the simulation synergy strategy.

Keywords

Acknowledgement

This study is supported by the National Natural Science Foundation of China (Grant No. 51778354).

References

  1. Akcay, H. and Filik, T. (2017), "Short-term wind speed forecasting by spectral analysis from long-term observations with missing values", Appl. Energ., 191, 653-662. https://doi.org/10.1016/j.apenergy.2017.01.063.
  2. Bao, Y., Yu, Y., Li, H., Ma, X., Jiao, W., Zou, Z. and Ou, J. (2015), "Compressive sensing-based lost data recovery of fast-moving wireless sensing for structural health monitoring", Struct. Control. Hlth., 22: 433-448. http://dx.doi.org/10.1002/stc.1681.
  3. Bao, X. and Li, C. (2019), "Fast simulation of non-stationary wind velocity based on time-frequency interpolation", J. Wind. Eng. Ind. Aerod., 193, 1-11, 103982. https://doi.org/10.1016/j.jweia.2019.103982.
  4. Bao, X. and Li, C. (2020), "Application of time-frequency interpolation and proper orthogonal decomposition in nonstationary wind field simulation", J. Eng. Mech., 146(5), 04020034. https://doi.org/10.1061/(ASCE)EM.1943-7889.0001761.
  5. Bulte, C., Kleinebrahm, M., Yilmaz, H.u . and Gomez-Romero, J. (2023), "Multivariate time series imputation for energy data using neural networks", Energy AI, 13, 100239. https://doi.org/10.1016/j.egyai.2023.100239.
  6. Chen, X. (2008), "Analysis of along wind tall building response to transient nonstationary winds", J. Struct. Eng., 134(5), 782-791. http://dx.doi.org/10.1061/(ASCE)0733-9445(2008)134:5(782).
  7. Chen, J., Michael, C.H.H. and Xu, Y. (2007), "A comparative study of stationary and non-stationary wind models using field measurements", Bound-Lay Meteorol., 122, 105-121. https://doi.org/10.1007/s10546-006-9085-1.
  8. Chen, L. and Letchford, C.W. (2006), "Multi-scale correlation analyses of two lateral profiles of full-scale downburst wind speeds", J. Wind. Eng. Ind. Aerod., 94, 675-696. https://doi.org/10.1016/j.jweia.2006.01.021.
  9. Chen, X., Zhang, Z., Li, A., Hu, L., Liu, X., Fan, Z. and Sun, P. (2021), "Field measurement-based wind-induced response analysis of multi-tower building with tuned mass damper", Wind. Struct., 32(2), 143-159. https://doi.org/10.12989/was.2021.32.2.143.
  10. Fan, G., Li, J. and Hao, H. (2019), "Lost data recovery for structural health monitoring based on convolutional neural networks", Struct. Control. Hlth., 26(10), 1-21, e2433. https://doi.org/10.1002/stc.2433.
  11. Gast, K.D. (2003), "A comparison of extreme wind events as sampled in the 2002 thunderstorm outflow experiment", Master's Thesis, Texas Tech University, Lubbock, TX.
  12. Guo, Z., Wan, Y. and Ye, H. (2019), "A data imputation method for multivariate time series based on generative adversarial network", Neurocomputing, 360, 185-197. https://doi.org/10.1016/j.neucom.2019.06.007
  13. Jeong, S., Max Ferguson, M., Hou, R., Lynch, J.P., Sohn, H. and Law, K.H. (2019), "Sensor data reconstruction using bidirectional recurrent neural network with application to bridge monitoring", Adv. Eng. Inform., 42, 100991. https://doi.org/10.1016/j.aei.2019.100991.
  14. Jing, B., Pei, Y., Qian, Z., Wang, A., Zhu, S. and An, J. (2022), "Missing wind speed data reconstruction with improved context encoder network", Energy Rep., 8, 3386-3394. https://doi.org/10.1016/j.egyr.2022.02.177
  15. Ketkar, N. (2017), Convolutional Neural Networks. Springer International Publishing.
  16. Li, J., Li, C., He, L. and Shen, J. (2015), "Extended modulating functions for simulation of wind velocities with weak and strong nonstationarity", Renew. Energ. 83, 384-397. https://doi.org/10.1016/j.renene.2015.04.044.
  17. Li, Y., Bao, T., Chen, H., Zhang, K., Shu, X., Chen, Z. and Hu, Y. (2021), "A large-scale sensor missing data imputation framework for dams using deep learning and transfer learning strategy", Measurement, 109377. https://doi.org/10.1016/j.measurement.2021.109377.
  18. Lin, Q. and Li, C. (2020), "Kriging based sequence interpolation and probability distribution correction for gaussian wind field data reconstruction", J. Wind. Eng. Ind. Aerod., 205, 1-15, 104340. https://doi.org/10.1016/j.jweia.2020.104340.
  19. Lin, Q., Bao, X. and Li, C. (2022), "Deep learning based missing data recovery of non-stationary wind velocity", J. Wind. Eng. Ind. Aerod., 224, 1-13, 104962. https://doi.org/10.1016/j.jweia.2022.104962.
  20. Liu, J., Zhang, Q., Li, X., Li, G., Liu, Z., Xie, Y., Li, K. and Liu, B. (2021), "Transfer learning-based strategies for fault diagnosis in building energy systems", Energ. Buildings., 111256. https://doi.org/10.1016/j.enbuild.2021.111256
  21. Lombardo, F. T., Smith, D. A., Schroeder, J. L. and Mehta, K. C. (2014), "Thunderstorm characteristics of importance to wind engineering", J. Wind. Eng. Ind. Aerod., 125, 121-132. https://doi.org/10.1016/j.jweia.2013.12.004.
  22. Luengo, M. M., Shafiee, M. and Kolios, A. (2019), "Data management for structural integrity assessment of offshore wind turbine support structures: data cleansing and missing data imputation." Ocean Eng., 173, 867-883. https://doi.org/10.1016/j.oceaneng.2019.01.003.
  23. Luo, J., Liang, D. and Weiss, C. (2015), "Reconstruction of a near-surface tornado wind field from observed building damage", Wind. Struct., 20(3), 389-404. http://dx.doi.org/10.12989/was.2015.20.3.389 389.
  24. Luo, Y., Huang, G., Han, Y. and Cai, C.S. (2021), "Evaluation of full-order method for extreme wind effect estimation considering directionality", Wind. Struct., 32(3), 193-204. https://doi.org/10.12989/was.2021.32.3.193.
  25. Ma, C., Li, Z., Meng, F., Liao, H. and Wang, J. (2021a), "Wind-induced vibrations and suppression measures of the Hong Kong-Zhuhai-Macao Bridge", Wind. Struct., 32(3), 179-191. https://doi.org/10.12989/was.2021.32.3.179.
  26. Ma, Z., Yun, C. B., Wan, H. P., Shen, Y., Feng Yu, F. and Luo, Y. (2021b), "Probabilistic principal component analysis-based anomaly detection for structures with missing data", Struct. Control. Hlth., 1-25, e2698. http://dx.doi.org/10.1002/stc.2698.
  27. Nagayama, T., Sim, S. H., Miyamori, Y. and Spencer, B. F. Jr. (2007), "Issues in structural health monitoring employing smart sensors", Smart. Struct. Syst., 3(3), 299-320. https://doi.org/10.12989/sss.2007.3.3.299.
  28. Nagayama, T. and Spencer, B.F. Jr. (2008), "Structural health monitoring using smart sensors", Proceedings of SPIE - The International Society for Optical Engineering, 1-15.
  29. Nayer, G. D., Breuer, M. and Boulbrachene, K. (2021), "FSI simulations of wind gusts impacting an air-inflated flexible membrane at Re = 100,000", J. Fluid. Struct., 28.
  30. Oh, B.K., Glisic, B., Kim, Y. and Park, H.S. (2020), "Convolutional neural network-based data recovery method for structural health monitoring", Struct. Health. Monit., 1-18. https://doi.org/10.1177/1475921719897571.
  31. Pan, Y., Ventura, C.E. and Li, T. (2022), "Sensor placement and seismic response reconstruction for structural health monitoring using a deep neural network", B. Earthq. Eng., 20, 4513-4532. https://doi.org/10.1007/s10518-021-01266-y.
  32. Pan, S.J. and Yang, Q. (2009), "A survey on transfer learning", IEEE Trans. Knowl. Data Eng., 22, 1345-1359. http://refhub.elsevier.com/S0952-1976(21)00091-9/sb20
  33. Paola, M.D. (1998), "Digital simulation of wind field velocity", J. Wind. Eng. Ind. Aerod., 74-76(2), 91-109. https://doi.org/10.1016/S0167-6105(98)00008-7.
  34. Priestly, M.B. (1965), "Evolutionary spectra and non-stationary processes", J. Royal Statistic. Soc., 27(2), 204-237. https://doi.org/10.2307/2984191.
  35. Razavi-Far, R. and Saif, M. (2015), "Imputation of missing data for diagnosing sensor faults in a wind turbine", IEEE Int. Conf. Syst., Man, Cybern. 99-104. https://doi.org/10.1109/SMC.2015.30.
  36. Santos, T.M.O., Silva, I.N. and Bessani, M. (2022), "Evolving dynamic bayesian networks by an analytical threshold for dealing with data imputation in time series dataset", Big Data Res., 28, 100316. https://doi.org/10.1016/j.bdr.2022.100316.
  37. Su, Y., Huang, G., Liu, R. and Zeng, Y. (2021), "Efficient buffeting analysis under non-stationary winds and application to a mountain bridge", Wind. Struct., 32(2), 89-104. https://doi.org/10.12989/was.2021.32.2.89.
  38. Tian J., Gurley, K.R., Diaz, M.T., Fernandez-Caban, P.L., Masters, F.J. and Fang, R. (2020), "Low-rise gable roof buildings pressure prediction using deep neural networks", J. Wind Eng. Ind. Aerod., 196, 1-17, 104026.
  39. von Karman, T. (1948), "Progress in the statistical theory of turbulence", Proceedings of the National Academy of Sciences of the United States of America, 34(11), 530-539. 10.1073/pnas.34.11.530.
  40. Wang, D., Zhao, Z., Ou, T., Xin, Z., Wang, M. and Zhang, Y. (2021), "Wind resistance performance of a continuous welding stainless steel roof under static ultimate wind loading with testing and simulation methods", Wind. Struct., 32(1), 55-69. https://doi.org/10.12989/was.2021.32.1.55.
  41. Wellenzohn, K., Bohlen, M.H.A., Dignos, A., Gamper, J. and Mitterer, H. (2017), "Continuous imputation of missing values in streams of pattern-determining time series", in: EDBT: 330-341.
  42. Weng, Y. and Paal, S.G. (2022), "Machine learning-based wind pressure prediction of low-rise non-isolated buildings", Eng. Struct., 258, 114148. https://doi.org/10.1016/j.engstruct.2022.114148.
  43. Wesonga, R. (2015), "On multivariate imputation and forecasting of decadal wind speed missing data", Springer Plus, 4, 12, 12-20. https://doi.org/10.1186/s40064-014-0774-9.
  44. Wu, H., Xian, J., Wang, J., Khandge, S. and Mohapatra, P. (2018), "Missing data recovery using reconstruction in ocean wireless sensor networks", Comput. Commun., 132, 1-9. https://doi.org/10.1016/j.comcom.2018.09.007
  45. Yang, Y. and Nagarajaiah, S. (2016), "Harnessing data structure for recovery of randomly missing structural vibration responses time history: Sparse representation versus low-rank structure", Mech. Syst. Signal. Pr., 74, 165-182. https://doi.org/10.1016/j.ymssp.2015.11.009.
  46. Zapata-Sierra, A.J., Cama-Pinto, A., Montoya, F.G., Alcayde, A. and Manzano-Agugliaro, F. (2019), "Wind missing data arrangement using wavelet based techniques for getting maximum likelihood", Energ. Convers. Manage., 185, 552-561. https://doi.org/10.1016/j.enconman.2019.01.109.
  47. Zhang, J., Mu, X., Fang, J. and Yang, Y. (2019), "Time series imputation via integration of revealed information based on the residual shortcut connection", IEEE Access. 7, 102397-102405. https://doi.org/10.1109/ACCESS.2019.2928641
  48. Zhang, Y.D., Zhou, B., Cai, X., Guo, W., Ding, X. and Yuan, X. (2021), "Missing value imputation in multivariate time series with end-to-end generative adversarial networks", Inform. Sci., 551, 67-82. https://doi.org/10.1016/j.ins.2020.11.035
  49. Zhu, H., Yang, B., Zhang, Q., Pan, L. and Sun, S. (2021a), "Wind engineering for high-rise buildings: A review", Wind. Struct., 32(3), 249-265. https://doi.org/10.12989/was.2021.32.3.249.
  50. Zhu, J., Wang, X., Kou, L., Zheng, L. and Zhang, H. (2021b), "Application of combined transfer learning and convolutional neural networks to optimize plasma spraying", Appl. Surf. Sci., 563, 1-11, 150098. https://doi.org/10.1016/j.apsusc.2021.150098.