Probabilistic wind power forecasting combining deep learning architectures

Eric Lacoa Arends; Simon J. Watson; Sukanta Basu; Bedassa Cheneka

doi:10.1109/EEM49802.2020.9221929

Probabilistic wind power forecasting combining deep learning architectures

Eric Lacoa Arends, Simon J. Watson, Sukanta Basu, Bedassa Cheneka

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

61 Downloads (Pure)

Abstract

A series of probabilistic models were bench-marked during the European Energy Markets forecasting Competition 2020 to assess their relative accuracy in predicting aggregated Swedish wind power generation using as input historic weather forecasts from a numerical weather prediction model. In this paper, we report the results of one of these models which uses a deep learning approach integrating two architectures: (a) Convolutional Neural Network (CNN) LeNet-5 based architectrure; (b) Multi-Layer Perceptron (MLP) architecture -with two hidden layers-. These are concatenated into the Smooth Pinball Neural Network (SPNN) framework for quantile regression. Hyperparameters were optimised to produce the best model for every region. When tuned, the re-forecasts from the model performed favorably compared to other machine learning approaches and showed significant improvement on the original competition results, though failed to fully capture spatial patterns in certain cases when compared to other methods.

Original language	English
Title of host publication	2020 17th International Conference on the European Energy Market, EEM 2020
Publisher	IEEE
Number of pages	6
ISBN (Electronic)	9781728169194
DOIs	https://doi.org/10.1109/EEM49802.2020.9221929
Publication status	Published - 2020
Event	17th International Conference on the European Energy Market, EEM 2020 - Stockholm, Sweden Duration: 16 Sept 2020 → 18 Sept 2020

Publication series

Name	International Conference on the European Energy Market, EEM
Volume	2020-September
ISSN (Print)	2165-4077
ISSN (Electronic)	2165-4093

Conference

Conference	17th International Conference on the European Energy Market, EEM 2020
Country/Territory	Sweden
City	Stockholm
Period	16/09/20 → 18/09/20

Bibliographical note

Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care
Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

Keywords

convolutional neural network
multilayer perceptron
numerical weather prediction
smooth pinball neural network
wind power forecasting

Access to Document

10.1109/EEM49802.2020.9221929

09221929Final published version, 489 KB

Cite this

@inproceedings{152addeac35e42379062a107c721658e,

title = "Probabilistic wind power forecasting combining deep learning architectures",

abstract = "A series of probabilistic models were bench-marked during the European Energy Markets forecasting Competition 2020 to assess their relative accuracy in predicting aggregated Swedish wind power generation using as input historic weather forecasts from a numerical weather prediction model. In this paper, we report the results of one of these models which uses a deep learning approach integrating two architectures: (a) Convolutional Neural Network (CNN) LeNet-5 based architectrure; (b) Multi-Layer Perceptron (MLP) architecture -with two hidden layers-. These are concatenated into the Smooth Pinball Neural Network (SPNN) framework for quantile regression. Hyperparameters were optimised to produce the best model for every region. When tuned, the re-forecasts from the model performed favorably compared to other machine learning approaches and showed significant improvement on the original competition results, though failed to fully capture spatial patterns in certain cases when compared to other methods.",

keywords = "convolutional neural network, multilayer perceptron, numerical weather prediction, smooth pinball neural network, wind power forecasting",

author = "Arends, {Eric Lacoa} and Watson, {Simon J.} and Sukanta Basu and Bedassa Cheneka",

note = "Green Open Access added to TU Delft Institutional Repository {\textquoteleft}You share, we take care!{\textquoteright} – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.; 17th International Conference on the European Energy Market, EEM 2020 ; Conference date: 16-09-2020 Through 18-09-2020",

year = "2020",

doi = "10.1109/EEM49802.2020.9221929",

language = "English",

series = "International Conference on the European Energy Market, EEM",

publisher = "IEEE",

booktitle = "2020 17th International Conference on the European Energy Market, EEM 2020",

address = "United States",

}

Arends, EL, Watson, SJ, Basu, S & Cheneka, B 2020, Probabilistic wind power forecasting combining deep learning architectures. in 2020 17th International Conference on the European Energy Market, EEM 2020., 9221929, International Conference on the European Energy Market, EEM, vol. 2020-September, IEEE, 17th International Conference on the European Energy Market, EEM 2020, Stockholm, Sweden, 16/09/20. https://doi.org/10.1109/EEM49802.2020.9221929

Probabilistic wind power forecasting combining deep learning architectures. / Arends, Eric Lacoa; Watson, Simon J.; Basu, Sukanta et al.
2020 17th International Conference on the European Energy Market, EEM 2020. IEEE, 2020. 9221929 (International Conference on the European Energy Market, EEM; Vol. 2020-September).

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

TY - GEN

T1 - Probabilistic wind power forecasting combining deep learning architectures

AU - Arends, Eric Lacoa

AU - Watson, Simon J.

AU - Basu, Sukanta

AU - Cheneka, Bedassa

N1 - Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

PY - 2020

Y1 - 2020

N2 - A series of probabilistic models were bench-marked during the European Energy Markets forecasting Competition 2020 to assess their relative accuracy in predicting aggregated Swedish wind power generation using as input historic weather forecasts from a numerical weather prediction model. In this paper, we report the results of one of these models which uses a deep learning approach integrating two architectures: (a) Convolutional Neural Network (CNN) LeNet-5 based architectrure; (b) Multi-Layer Perceptron (MLP) architecture -with two hidden layers-. These are concatenated into the Smooth Pinball Neural Network (SPNN) framework for quantile regression. Hyperparameters were optimised to produce the best model for every region. When tuned, the re-forecasts from the model performed favorably compared to other machine learning approaches and showed significant improvement on the original competition results, though failed to fully capture spatial patterns in certain cases when compared to other methods.

AB - A series of probabilistic models were bench-marked during the European Energy Markets forecasting Competition 2020 to assess their relative accuracy in predicting aggregated Swedish wind power generation using as input historic weather forecasts from a numerical weather prediction model. In this paper, we report the results of one of these models which uses a deep learning approach integrating two architectures: (a) Convolutional Neural Network (CNN) LeNet-5 based architectrure; (b) Multi-Layer Perceptron (MLP) architecture -with two hidden layers-. These are concatenated into the Smooth Pinball Neural Network (SPNN) framework for quantile regression. Hyperparameters were optimised to produce the best model for every region. When tuned, the re-forecasts from the model performed favorably compared to other machine learning approaches and showed significant improvement on the original competition results, though failed to fully capture spatial patterns in certain cases when compared to other methods.

KW - convolutional neural network

KW - multilayer perceptron

KW - numerical weather prediction

KW - smooth pinball neural network

KW - wind power forecasting

UR - http://www.scopus.com/inward/record.url?scp=85094853887&partnerID=8YFLogxK

U2 - 10.1109/EEM49802.2020.9221929

DO - 10.1109/EEM49802.2020.9221929

M3 - Conference contribution

AN - SCOPUS:85094853887

T3 - International Conference on the European Energy Market, EEM

BT - 2020 17th International Conference on the European Energy Market, EEM 2020

PB - IEEE

T2 - 17th International Conference on the European Energy Market, EEM 2020

Y2 - 16 September 2020 through 18 September 2020

ER -

Probabilistic wind power forecasting combining deep learning architectures

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this