View in EDS HTML Full Text PDF Full Text

Load-Predictive Pitch Control Strategy for Wind Turbines Under Turbulent Wind Conditions Based on Long Short-Term Memory Neural Networks.

Saved in:
Bibliographic Details
Title: Load-Predictive Pitch Control Strategy for Wind Turbines Under Turbulent Wind Conditions Based on Long Short-Term Memory Neural Networks.
Authors: Bao, Daorina1 (AUTHOR), Li, Peng1,2 (AUTHOR), Zhang, Jun1,3 (AUTHOR), Shi, Zhongyu1,4 (AUTHOR), Luo, Yongshui1,2 (AUTHOR), Ao, Xiaohu2,3 (AUTHOR), Cui, Ruijun1,3 (AUTHOR), Guo, Xiaodong4 (AUTHOR) guoxiaodong@sztu.edu.cn
Source: Energies (19961073). May2026, Vol. 19 Issue 9, p2044. 18p.
Subject Terms: *Long short-term memory, *Predictive control systems, *Wind turbines, *Turbulent flow, *Fatigue cracks
Abstract: Under turbulent wind conditions, rapid wind speed fluctuations can markedly increase the fatigue loads borne by wind turbine blades and towers. In practice, conventional PID pitch control based on speed feedback often struggles to deliver satisfactory load mitigation, mainly because the wind turbine system is highly nonlinear, strongly coupled, and subject to time-delay effects. To overcome these limitations, this paper proposes a load-predictive pitch control strategy built on a Long Short-Term Memory (LSTM) network. Specifically, the LSTM model is first employed to predict the hub-fixed tilt and yaw moments ahead of time. These predicted values are then introduced as feedforward signals and combined with the conventional speed-based pitch control signal as well as a proportional feedback term. After that, the inverse Coleman transformation is used to generate the individual pitch commands for each blade. To verify the effectiveness of the proposed method, co-simulations were carried out in FAST and MATLAB/Simulink on a 5000 KW distributed pitch-controlled wind turbine under IEC Kaimal spectrum wind conditions, with a mean wind speed of 18 m/s and Class B turbulence intensity. The results show that the LSTM prediction model achieves an R2 of 0.998 on the test dataset, with an RMSE as low as 0.0051. Compared with the conventional pitch-based power control strategy, the proposed approach maintains the same average power output while significantly reducing fatigue loads, thereby contributing to a longer service life for the wind turbine. [ABSTRACT FROM AUTHOR]
Database: Energy & Power Source
Full text is not displayed to guests.
FullText Links:
  – Type: pdflink
Text:
  Availability: 1
Header DbId: enr
DbLabel: Energy & Power Source
An: 193715940
AccessLevel: 6
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Load-Predictive Pitch Control Strategy for Wind Turbines Under Turbulent Wind Conditions Based on Long Short-Term Memory Neural Networks.
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Bao%2C+Daorina%22">Bao, Daorina</searchLink><relatesTo>1</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Li%2C+Peng%22">Li, Peng</searchLink><relatesTo>1,2</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Zhang%2C+Jun%22">Zhang, Jun</searchLink><relatesTo>1,3</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Shi%2C+Zhongyu%22">Shi, Zhongyu</searchLink><relatesTo>1,4</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Luo%2C+Yongshui%22">Luo, Yongshui</searchLink><relatesTo>1,2</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Ao%2C+Xiaohu%22">Ao, Xiaohu</searchLink><relatesTo>2,3</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Cui%2C+Ruijun%22">Cui, Ruijun</searchLink><relatesTo>1,3</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Guo%2C+Xiaodong%22">Guo, Xiaodong</searchLink><relatesTo>4</relatesTo> (AUTHOR)<i> guoxiaodong@sztu.edu.cn</i>
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="JN" term="%22Energies+%2819961073%29%22">Energies (19961073)</searchLink>. May2026, Vol. 19 Issue 9, p2044. 18p.
– Name: Subject
  Label: Subject Terms
  Group: Su
  Data: *<searchLink fieldCode="DE" term="%22Long+short-term+memory%22">Long short-term memory</searchLink><br />*<searchLink fieldCode="DE" term="%22Predictive+control+systems%22">Predictive control systems</searchLink><br />*<searchLink fieldCode="DE" term="%22Wind+turbines%22">Wind turbines</searchLink><br />*<searchLink fieldCode="DE" term="%22Turbulent+flow%22">Turbulent flow</searchLink><br />*<searchLink fieldCode="DE" term="%22Fatigue+cracks%22">Fatigue cracks</searchLink>
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: Under turbulent wind conditions, rapid wind speed fluctuations can markedly increase the fatigue loads borne by wind turbine blades and towers. In practice, conventional PID pitch control based on speed feedback often struggles to deliver satisfactory load mitigation, mainly because the wind turbine system is highly nonlinear, strongly coupled, and subject to time-delay effects. To overcome these limitations, this paper proposes a load-predictive pitch control strategy built on a Long Short-Term Memory (LSTM) network. Specifically, the LSTM model is first employed to predict the hub-fixed tilt and yaw moments ahead of time. These predicted values are then introduced as feedforward signals and combined with the conventional speed-based pitch control signal as well as a proportional feedback term. After that, the inverse Coleman transformation is used to generate the individual pitch commands for each blade. To verify the effectiveness of the proposed method, co-simulations were carried out in FAST and MATLAB/Simulink on a 5000 KW distributed pitch-controlled wind turbine under IEC Kaimal spectrum wind conditions, with a mean wind speed of 18 m/s and Class B turbulence intensity. The results show that the LSTM prediction model achieves an R2 of 0.998 on the test dataset, with an RMSE as low as 0.0051. Compared with the conventional pitch-based power control strategy, the proposed approach maintains the same average power output while significantly reducing fatigue loads, thereby contributing to a longer service life for the wind turbine. [ABSTRACT FROM AUTHOR]
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=enr&AN=193715940
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.3390/en19092044
    Languages:
      – Code: eng
        Text: English
    PhysicalDescription:
      Pagination:
        PageCount: 18
        StartPage: 2044
    Subjects:
      – SubjectFull: Long short-term memory
        Type: general
      – SubjectFull: Predictive control systems
        Type: general
      – SubjectFull: Wind turbines
        Type: general
      – SubjectFull: Turbulent flow
        Type: general
      – SubjectFull: Fatigue cracks
        Type: general
    Titles:
      – TitleFull: Load-Predictive Pitch Control Strategy for Wind Turbines Under Turbulent Wind Conditions Based on Long Short-Term Memory Neural Networks.
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Bao, Daorina
      – PersonEntity:
          Name:
            NameFull: Li, Peng
      – PersonEntity:
          Name:
            NameFull: Zhang, Jun
      – PersonEntity:
          Name:
            NameFull: Shi, Zhongyu
      – PersonEntity:
          Name:
            NameFull: Luo, Yongshui
      – PersonEntity:
          Name:
            NameFull: Ao, Xiaohu
      – PersonEntity:
          Name:
            NameFull: Cui, Ruijun
      – PersonEntity:
          Name:
            NameFull: Guo, Xiaodong
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 01
              M: 05
              Text: May2026
              Type: published
              Y: 2026
          Identifiers:
            – Type: issn-print
              Value: 19961073
          Numbering:
            – Type: volume
              Value: 19
            – Type: issue
              Value: 9
          Titles:
            – TitleFull: Energies (19961073)
              Type: main
ResultId 1