Antonino Marvuglia
CRP Henri Tudor/CRTE, Luxembourg
Antonio Messineo
Faculty of Engineering & Architecture, Kore University of Enna, Italy
Ladda ner artikel
http://dx.doi.org/10.3384/ecp110574217Ingår i: World Renewable Energy Congress - Sweden; 8-13 May; 2011; Linköping; Sweden
Linköping Electronic Conference Proceedings 57:22, s. 4217-4224
Publicerad: 2011-11-03
ISBN: 978-91-7393-070-3
ISSN: 1650-3686 (tryckt), 1650-3740 (online)
Improving the performance of prediction algorithms is one of the priorities in the wind energy research agenda of the scientific community. In a very simplistic approach; short-term predictions of wind power production at a given site could be generated by passing forecasts of meteorological variables (namely wind speed) through the so-called wind farm power curve; which links the wind speed to the power that is produced by the whole wind farm. However; the estimation of this conversion function is indeed a challenging task; because it is nonlinear and bounded; in addition to being non-stationary due for example to changes in the site environment and seasonality. Even for a single wind turbine the measured power at different wind speeds is generally different to the rated power; since the operating conditions on site are generally different to the conditions under which the turbine was calibrated (the wind speed on site is not uniform horizontally across the face of the turbine; the vertical wind profile and the air density are different than during the calibration; the wind data available on site are not always measured at the height of the turbine’s hub).
The recent developments in data mining and evolutionary computation (EC) offer promising approaches to modelling the power curves of turbines. In this paper we use a self-supervised neural network called GMR (Generalized Mapping Regressor) to learn the relationship between the wind speed and the generated power in a whole wind farm. GMR is an incremental self-supervised neural network which can approximate every multidimensional function or relation presenting any kind of discontinuity. The approach used is a data driven one; in the sense that the relationship is learned directly from the data; without using any explicit physical or mathematical relationship between input and output space. The model is potentially applicable to any site; provided that a statistically consistent amount of wind and power data is available. The methodology allows the creation of a non-parametric model of the power curve that can be used as a reference profile for on-line monitoring of the power generation process; as well as for power forecasts.
The results obtained with the proposed approach are compared with another state-of-the-art data mining algorithm (namely; a feedforward Multi Layer Perceptron) showing that the algorithm provides fair performances if a suitable pre-processing of the input data is accomplished.
[1] R.M.G. Castro; A.F.M. Ferreira; A comparison between chronological and probabilistic methods to estimate wind power capacity credit; IEEE Trans Pow Syst 16(4); 2001; pp. 904-909.
doi: 10.1109/59.962444.
[2] P. Børre Eriksen; T. Ackermann; H. Abildgaard; P. Smith; W. Winter; J. Rogríguez García; System operation with high wind penetration; IEEE Pow Energ Magazine; November/December 2005.
[3] L. Landberg; Short-term prediction of the power production from wind farms; Journal of Wind Engineering and Industrial Aerodynamics 80(1-2); 1998; pp. 207–220.
doi: 10.1016/S0167-6105(98)00192-510.1016/S0167-6105(98)00192-5.
[4] S. Li; D.C. Wunsch; E. O’Hair; M.G. Giesselmann; Comparative analysis of regression and artificial neural network models for wind turbine power curve estimation; J Solar Energ Engineering 123(4); 2001; 327–332.
doi: 10.1115/1.1413216.
[5] S.L. Goh; M. Chen; D.H. Popovic; K. Aihara; D. Obradovic; D.P. Mandic; Complex-valued estimation of wind profile; Renew Energ 31; 2006; pp. 1733–1750.
doi: 10.1016/j.renene.2005.07.006.
[6] D.P. Mandic; S. Javidi; S.L. Goh; A. Kuhc; K. Aihara; Complex-valued prediction of wind profile using augmented complex statistics; Renew Energ 34; 2009; 196–201.
doi: 10.1016/j.renene.2008.03.022.
[7] T. Üstütas; A.D Sahin; Wind turbine power curve estimation based on cluster center fuzzy logic modelling; Journal of Wind Engineering and Industrial Aerodynamics 96; 2008; pp. 611–620.
doi: 10.1016/j.jweia.2008.02.001.
[8] A. Kusiak; H. Zheng; Z. Song. Models for monitoring wind farm power; Renew Energ 34; 2009; pp. 583–590.
doi: 10.1016/j.renene.2008.05.032.
[9] A. Kusiak; H. Zheng; Z. Song; On-line monitoring of power curves; Renew Energ 34; 2009; pp. 1487–1493.
doi: 10.1016/j.renene.2008.10.022.
[10] M. Lei; L. Shiyan; J. Chuanwen; L. Hongling; Z. Yan; A Review on the Forecasting of Wind Speed and Generated Power; Renew Sust Energ Rev 13(4); 2009; pp. 915-920.
doi: 10.1016/j.rser.2008.02.002.
[11] B. Bell; Individual wind turbine and overall power plant performance verification; San Diego; CA; 2008.
[12] Cirrincione G; Cirrincione M; Lu C; Van Huffel S. A novel neural approach to inverse problems with discontinuities (the GMR neural network); Proceedings of 2003 Int. J Conf Neural Networks (IJCNN’03); Portland; Oregon; pp. 3106–3111.
[13] C. Lu; The Generalised Mapping Regressor (GMR) neural network for inverse discontinuous problems; MSc Thesis; 2000; Katholieke Universiteit Leuven (Belgium).
[14] R.M. Gray; Vector quantization; IEEE ASSP Magazine 4(2); 1984; pp. 4–29.
doi: 10.1109/MASSP.1984.1162229.
[15] A.R. Teixeira; A.M. Tomé; K. Stadlthanner; E.W. Lang. KPCA denoising and the pre-image problem revisited; Digit Signal Process 18; 2008; pp. 568–580.
doi: 10.1016/j.dsp.2007.08.001.
[16] M.T. Hagan; H.B. Demuth; M.H. Beale; Neural network design; PWS Publishing; 1996.
[17] S. Haykin; Neural Networks – A Comprehensive Foundation; Prenctice Hall; Upper Saddle River; 1999.
[18] D.C. Montgomery; Introduction to statistical quality control; John Wiley & Sons; New York; 5th ed. 2005.
