21 / 31
Page 93
Innovative Energy & Research | ISSN: 2576-1463 | Volume 7
Renewable Energy and Resources
Energy Materials and Fuel Cell Research
2
nd
International Conference on
&
August 27-28, 2018 | Boston, USA
ASurvey and some new results on machine learning methods for the estimation of the power curve of wind
turbines
Guilherme Barreto
and
Haroldo Maya
Federal University of Ceara (UFC), Brazil
I
n this work, we provide a comprehensive survey of recent machine learning approaches for wind turbine power curve (WTPC)
estimation. Additionally, we revisit the classical polynomial model aiming at improving it by means of an automatic and more
parsimonious design. For this purpose, we propose a methodology based on evolutionary computation which returns the optimal
order of the polynomial as well as selects (by pruning) the relevant terms in this polynomial. A comprehensive performance
comparison is carried out involving the proposed approach and the state of the art in estimating the power curve of wind turbines,
such as the logistic models (with 4 and 5 parameters), artificial neural networks, Takagi-Sugeno fuzzy model, and weighted polynomial
regression. The results clearly indicate that the proposed methodology consistently outperforms the state of the art methods.
gbarreto@ufc.brInnov Ener Res 2018, Volume 7
DOI: 10.4172/2576-1463-C2-006