Wind turbine site-specific load estimation using artificial neural networks calibrated by means of high-fidelity load simulations

Research output: Contribution to journalConference articleResearchpeer-review

303 Downloads (Pure)

Abstract

Previous studies have suggested the use of reduced-order models calibrated by means of high-fidelity load simulations as means for computationally inexpensive wind turbine load assessments; the so far best performing surrogate modelling approach in terms of balance between accuracy and computational cost has been the polynomial chaos expansion (PCE). Regarding the growing interest in advanced machine learning applications, the potential of using Artificial Neural-Network (ANN) based surrogate models for improved simplified load assessment is investigated in this study. Different ANN model architectures have been evaluated and compared to other types of surrogate models (PCE and quadratic response surface). The results show that a feedforward neural network with two hidden layers and 11 neurons per layer, trained with the Levenberg Marquardt backpropagation algorithm is able to estimate blade root flapwise damage-equivalent loads (DEL) more accurately and faster than a PCE trained on the same data set. Further research will focus on further model improvements by applying different training techniques, as well as expanding the work with more load components.
Original languageEnglish
Article number062027
Book seriesJournal of Physics: Conference Series
Volume1037
Issue number6
Number of pages10
ISSN1742-6596
DOIs
Publication statusPublished - 2018
EventThe Science of Making Torque from Wind 2018 - Politecnico di Milano (POLIMI), Milan, Italy
Duration: 20 Jun 201822 Jun 2018
Conference number: 7
http://www.torque2018.org/

Conference

ConferenceThe Science of Making Torque from Wind 2018
Number7
LocationPolitecnico di Milano (POLIMI)
CountryItaly
CityMilan
Period20/06/201822/06/2018
Internet address

Bibliographical note

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

Cite this

@inproceedings{838a836d04c248bb8aaa8694f1392661,
title = "Wind turbine site-specific load estimation using artificial neural networks calibrated by means of high-fidelity load simulations",
abstract = "Previous studies have suggested the use of reduced-order models calibrated by means of high-fidelity load simulations as means for computationally inexpensive wind turbine load assessments; the so far best performing surrogate modelling approach in terms of balance between accuracy and computational cost has been the polynomial chaos expansion (PCE). Regarding the growing interest in advanced machine learning applications, the potential of using Artificial Neural-Network (ANN) based surrogate models for improved simplified load assessment is investigated in this study. Different ANN model architectures have been evaluated and compared to other types of surrogate models (PCE and quadratic response surface). The results show that a feedforward neural network with two hidden layers and 11 neurons per layer, trained with the Levenberg Marquardt backpropagation algorithm is able to estimate blade root flapwise damage-equivalent loads (DEL) more accurately and faster than a PCE trained on the same data set. Further research will focus on further model improvements by applying different training techniques, as well as expanding the work with more load components.",
author = "Laura Schr{\"o}der and Dimitrov, {Nikolay Krasimirov} and Verelst, {David Robert} and S{\o}rensen, {John Aasted}",
note = "Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.",
year = "2018",
doi = "10.1088/1742-6596/1037/6/062027",
language = "English",
volume = "1037",
journal = "Journal of Physics: Conference Series (Online)",
issn = "1742-6596",
publisher = "IOP Publishing",
number = "6",

}

TY - GEN

T1 - Wind turbine site-specific load estimation using artificial neural networks calibrated by means of high-fidelity load simulations

AU - Schröder, Laura

AU - Dimitrov, Nikolay Krasimirov

AU - Verelst, David Robert

AU - Sørensen, John Aasted

N1 - Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

PY - 2018

Y1 - 2018

N2 - Previous studies have suggested the use of reduced-order models calibrated by means of high-fidelity load simulations as means for computationally inexpensive wind turbine load assessments; the so far best performing surrogate modelling approach in terms of balance between accuracy and computational cost has been the polynomial chaos expansion (PCE). Regarding the growing interest in advanced machine learning applications, the potential of using Artificial Neural-Network (ANN) based surrogate models for improved simplified load assessment is investigated in this study. Different ANN model architectures have been evaluated and compared to other types of surrogate models (PCE and quadratic response surface). The results show that a feedforward neural network with two hidden layers and 11 neurons per layer, trained with the Levenberg Marquardt backpropagation algorithm is able to estimate blade root flapwise damage-equivalent loads (DEL) more accurately and faster than a PCE trained on the same data set. Further research will focus on further model improvements by applying different training techniques, as well as expanding the work with more load components.

AB - Previous studies have suggested the use of reduced-order models calibrated by means of high-fidelity load simulations as means for computationally inexpensive wind turbine load assessments; the so far best performing surrogate modelling approach in terms of balance between accuracy and computational cost has been the polynomial chaos expansion (PCE). Regarding the growing interest in advanced machine learning applications, the potential of using Artificial Neural-Network (ANN) based surrogate models for improved simplified load assessment is investigated in this study. Different ANN model architectures have been evaluated and compared to other types of surrogate models (PCE and quadratic response surface). The results show that a feedforward neural network with two hidden layers and 11 neurons per layer, trained with the Levenberg Marquardt backpropagation algorithm is able to estimate blade root flapwise damage-equivalent loads (DEL) more accurately and faster than a PCE trained on the same data set. Further research will focus on further model improvements by applying different training techniques, as well as expanding the work with more load components.

U2 - 10.1088/1742-6596/1037/6/062027

DO - 10.1088/1742-6596/1037/6/062027

M3 - Conference article

VL - 1037

JO - Journal of Physics: Conference Series (Online)

JF - Journal of Physics: Conference Series (Online)

SN - 1742-6596

IS - 6

M1 - 062027

ER -