TY - GEN
T1 - A Nonparametric Probability Distribution Model for Short-Term Wind Power Prediction Error
AU - Khorramdel, B.
AU - Khorramdel, H.
AU - Zare, A.
AU - Safari, N.
AU - Sangrody, H.
AU - Chung, C. Y.
N1 - Publisher Copyright:
© 2018 IEEE.
PY - 2018/8/27
Y1 - 2018/8/27
N2 - Accurate wind power prediction error (WPPE) modeling is of high importance in power systems with large scale wind power generation containing high level of uncertainty. Since WPPE cannot be entirely removed, providing its accurate probability distribution model can assist power system operators in mitigating its negative effects on decision making conditions. In this paper, unlike previous related works, a nonparametric model is presented using kernel density estimation (KDE) with an efficient bandwidth (BW) selection technique called 'advanced plug-in' technique. The utilized BW selection technique enables KDE to accurately estimate important features of WPPE distribution, e.g., fat tails, high skewness and kurtosis. The proposed WPPE modeling approach is simulated using one-year time series of real wind power and corresponding predicted values for 1-hour look-ahead time. The efficacy of the proposed WPPE model is depicted using Centennial wind farm dataset in south of Saskatchewan province in Canada. Results show that parametric distribution models like Normal, Stable, and so on may not properly model the uncertainty of WPPE.
AB - Accurate wind power prediction error (WPPE) modeling is of high importance in power systems with large scale wind power generation containing high level of uncertainty. Since WPPE cannot be entirely removed, providing its accurate probability distribution model can assist power system operators in mitigating its negative effects on decision making conditions. In this paper, unlike previous related works, a nonparametric model is presented using kernel density estimation (KDE) with an efficient bandwidth (BW) selection technique called 'advanced plug-in' technique. The utilized BW selection technique enables KDE to accurately estimate important features of WPPE distribution, e.g., fat tails, high skewness and kurtosis. The proposed WPPE modeling approach is simulated using one-year time series of real wind power and corresponding predicted values for 1-hour look-ahead time. The efficacy of the proposed WPPE model is depicted using Centennial wind farm dataset in south of Saskatchewan province in Canada. Results show that parametric distribution models like Normal, Stable, and so on may not properly model the uncertainty of WPPE.
KW - Bandwidth selection technique
KW - Extreme learning machine
KW - Kernel density estimation
KW - Wind power prediction error
UR - http://www.scopus.com/inward/record.url?scp=85053632594&partnerID=8YFLogxK
U2 - 10.1109/CCECE.2018.8447838
DO - 10.1109/CCECE.2018.8447838
M3 - Conference article published in proceeding or book
AN - SCOPUS:85053632594
SN - 9781538624104
T3 - Canadian Conference on Electrical and Computer Engineering
BT - 2018 IEEE Canadian Conference on Electrical and Computer Engineering, CCECE 2018
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2018 IEEE Canadian Conference on Electrical and Computer Engineering, CCECE 2018
Y2 - 13 May 2018 through 16 May 2018
ER -