Computerimplementiertes Verfahren zur Vorhersage eines Vertikalen Windprofils an einer Windturbine

Details

Veröffentlichungs-Nr.

EP4807179

Anmeldetag

14. März 2025

Veröffentlichung

16. September 2026

Rechtsraum

EP

IPC

F03D17/00

Abstract

The invention relates to a computer-implemented method for predicting a vertical wind profile (WP) present at a wind turbine (1), where the method processes bending moment vectors (BV) at consecutive time points (t) derived from measurements, each bending vector (BV) comprising for one or more bending directions (d1, d2) a bending moment (bm1, bm2) for each blade (5) present at a predetermined location along the respective blade (5) at a respective time point (t), the method comprising the following steps: a) deriving from the plurality of bending moment vectors (BV) several spanwise load estimates (lesp) for each blade (5) and each bending direction (d1, d2), where each spanwise load estimate (lesp) is associated with a different horizontal portion (A1, A2, .., A6) of the circumferential area (CA) covered by the rotation of the blades (5),; b) feeding at least the spanwise load estimates (lesp) to one or more first trained machine learning models (MLa, MLb, MLc) each predicting an intermediate vertical wind profile (WPIa, WPIb, WPIc); c) deriving from the plurality of bending moment vectors (BV) several sectorwise load estimates (lese) for each blade (5) and each bending direction (d1, d2), where each sectorwise load estimate (lese) is associated with a different angular sector (AS) of the circumferential area (CA) covered by the rotation of the blades (5); d) feeding at least the sectorwise load estimates (lese) to one or more second trained machine learning models (MLa', MLb', MLc') each predicting a second intermediate vertical wind profile (WPIa', WPIb', WPIc'); e) feeding the first and second intermediate vertical wind profiles (WPIa, WPIb, WPIc, WPIa', WPIb', WPIc') as an input to a third trained machine learning model (MLe) predicting the vertical wind profile (WP) as an output.

Anmelder

Firma

Siemens Gamesa Renewable Energy A/S

Land

🇩🇰 Dänemark

Fachgebiete