본문 바로가기

LiDAR-informed transformer framework for spatiotemporal estimation of rotor-effective wind speed beyond Taylor's frozen turbulence hypothesis

문서 유형
연구출판물 · 학술지 논문
등재
Nature IndexSCIEScopus
저널 순위
JCR Q1Scopus Q1
저널
ENERGY366 · #142669 · 2026
피인용
Web of Science 0Scopus 0
JCR IF
JCR 2024 9.4
ISSN
0360-5442

저자

  1. 장멍제1저자Pusan Natl Univ, Sch Mech Engn, Busan 46241, South Korea
  2. 줄루아가오소리오존세바스티안Pusan Natl Univ, Sch Mech Engn, Busan 46241, South Korea
  3. 요시프 무스타파 즈헤이얼게니Pusan Natl Univ, Sch Mech Engn, Busan 46241, South Korea
  4. Choi, JungchulMokpo Natl Univ, Sch Mech & Ocean Engn, Muan 58554, South Korea
  5. 임희창교신저자Pusan Natl Univ, Sch Mech Engn, Busan 46241, South Korea

초록

The accurate estimation of rotor effective wind speed (REWS) is crucial in terms of improving feedforward control and enabling the effective optimization of wind turbine performance. Currently, upstream measurements provided by light detection and ranging (LiDAR) technology have been employed for REWS estimation; however, Taylor's frozen turbulence hypothesis is commonly assumed. Furthermore, effects present in real atmospheric boundary layer flow, e.g., turbulence evolution, are neglected; thus, estimation accuracy is limited. This paper proposes a data-driven approach that utilizes a transformer model to estimate REWS directly from LiDAR measurements. Through its attention mechanism, the transformer model understands the spatiotemporal information of the wind, effectively bypassing Taylor's hypothesis. LiDAR measurements are obtained from high-fidelity large eddy simulations of the atmospheric boundary layer across different wind speeds and preview distances to train and evaluate the proposed framework. The obtained results demonstrate that the transformer estimation exhibits good agreement with the target REWS values, significantly outperforming the REWS estimated with Taylor's hypothesis. The proposed model improved both the time lag and wind speed overestimation present in the LiDAR measurements. At the closest measurement gate, the root mean squared error of the estimation was reduced by 68%, 56%, and 37% for mean wind speeds of 12, 14, and 16 m/s. The findings of this study highlight the application prospects of the proposed framework in real-world wind turbine control.

키워드

Rotor effective wind speedTransformerLiDARWind turbineTURBINE CONTROLPITCH CONTROL

식별자

Scopus EID
2-s2.0-105053992358
WoS UID
001908714300001