Deep Concatenated Residual Network with Bidirectional LSTM for One-Hour-Ahead Wind Power Forecasting

Min Seung Ko; Kwangsuk Lee; Jae Kyeong Kim; Chang Woo Hong; Zhao Yang Dong; Kyeon Hur

doi:10.1109/TSTE.2020.3043884

Deep Concatenated Residual Network with Bidirectional LSTM for One-Hour-Ahead Wind Power Forecasting

Min Seung Ko, Kwangsuk Lee, Jae Kyeong Kim, Chang Woo Hong, Zhao Yang Dong, Kyeon Hur

Department of Electrical and Electronic Engineering

Research output: Contribution to journal › Article › peer-review

Abstract

This paper presents a deep residual network for improving time-series forecasting models, indispensable to reliable and economical power grid operations, especially with high shares of renewable energy sources. Motivated by the potential performance degradation due to the overfitting of the prevailing stacked bidirectional long short-term memory (Bi-LSTM) layers associated with its linear stacking, we propose a concatenated residual learning by connecting the multi-level residual network (MRN) and DenseNet. This method further integrates long and short Bi-LSTM networks, ReLU, and SeLU for its activating function. Rigorous studies present superior prediction accuracy and parameter efficiency for the widely used temperature dataset as well as the actual wind power dataset. The peak value forecasting and generalization capability, along with the credible confidence range, demonstrate that the proposed model offers essential features of a time-series forecasting, enabling a general forecasting framework in grid operations. The source code of this paper can be found in https://github.com/MinseungKo/DRNet.git.

Original language	English
Article number	9290065
Pages (from-to)	1321-1335
Number of pages	15
Journal	IEEE Transactions on Sustainable Energy
Volume	12
Issue number	2
DOIs	https://doi.org/10.1109/TSTE.2020.3043884
Publication status	Published - 2021 Apr

Bibliographical note

Funding Information:
Manuscript received May 13, 2020; revised September 3, 2020 and November 4, 2020; accepted December 2, 2020. Date of publication December 10, 2020; date of current version March 22, 2021. This work was supported in part by the Framework of International Cooperation Program Managed by National Research Foundation of Korea under Grant 2017K1A4A3013579 and in part by the “Human Resources Program in Energy Technology” of the Korea Institute of Energy Technology Evaluation and Planning (KETEP) Granted Financial Resource from the Ministry of Trade, Industry & Energy, Republic of Korea under Grant 20194030202420. Paper no. TSTE-00500-2020. (Corresponding author: Kyeon Hur.) Min-Seung Ko, Chang Woo Hong, and Kyeon Hur are with the School of Electrical and Electronic Engineering, Yonsei University, Seoul 03722, Korea (e-mail: kms4634500@yonsei.ac.kr; spearw@yonsei.ac.kr; khur@yonsei.ac.kr).

Publisher Copyright:
© 2010-2012 IEEE.

All Science Journal Classification (ASJC) codes

Renewable Energy, Sustainability and the Environment

UN SDGs

This output contributes to the following Sustainable Development Goal(s)

Access to Document

10.1109/TSTE.2020.3043884

Fingerprint Dive into the research topics of 'Deep Concatenated Residual Network with Bidirectional LSTM for One-Hour-Ahead Wind Power Forecasting'. Together they form a unique fingerprint.

Cite this

@article{42f81857edf44d079cd6b9c236f1b25a,

title = "Deep Concatenated Residual Network with Bidirectional LSTM for One-Hour-Ahead Wind Power Forecasting",

abstract = "This paper presents a deep residual network for improving time-series forecasting models, indispensable to reliable and economical power grid operations, especially with high shares of renewable energy sources. Motivated by the potential performance degradation due to the overfitting of the prevailing stacked bidirectional long short-term memory (Bi-LSTM) layers associated with its linear stacking, we propose a concatenated residual learning by connecting the multi-level residual network (MRN) and DenseNet. This method further integrates long and short Bi-LSTM networks, ReLU, and SeLU for its activating function. Rigorous studies present superior prediction accuracy and parameter efficiency for the widely used temperature dataset as well as the actual wind power dataset. The peak value forecasting and generalization capability, along with the credible confidence range, demonstrate that the proposed model offers essential features of a time-series forecasting, enabling a general forecasting framework in grid operations. The source code of this paper can be found in https://github.com/MinseungKo/DRNet.git. ",

author = "Ko, {Min Seung} and Kwangsuk Lee and Kim, {Jae Kyeong} and Hong, {Chang Woo} and Dong, {Zhao Yang} and Kyeon Hur",

note = "Funding Information: Manuscript received May 13, 2020; revised September 3, 2020 and November 4, 2020; accepted December 2, 2020. Date of publication December 10, 2020; date of current version March 22, 2021. This work was supported in part by the Framework of International Cooperation Program Managed by National Research Foundation of Korea under Grant 2017K1A4A3013579 and in part by the “Human Resources Program in Energy Technology” of the Korea Institute of Energy Technology Evaluation and Planning (KETEP) Granted Financial Resource from the Ministry of Trade, Industry & Energy, Republic of Korea under Grant 20194030202420. Paper no. TSTE-00500-2020. (Corresponding author: Kyeon Hur.) Min-Seung Ko, Chang Woo Hong, and Kyeon Hur are with the School of Electrical and Electronic Engineering, Yonsei University, Seoul 03722, Korea (e-mail: kms4634500@yonsei.ac.kr; spearw@yonsei.ac.kr; khur@yonsei.ac.kr). Publisher Copyright: {\textcopyright} 2010-2012 IEEE.",

year = "2021",

month = apr,

doi = "10.1109/TSTE.2020.3043884",

language = "English",

volume = "12",

pages = "1321--1335",

journal = "IEEE Transactions on Sustainable Energy",

issn = "1949-3029",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

number = "2",

}

TY - JOUR

T1 - Deep Concatenated Residual Network with Bidirectional LSTM for One-Hour-Ahead Wind Power Forecasting

AU - Ko, Min Seung

AU - Lee, Kwangsuk

AU - Kim, Jae Kyeong

AU - Hong, Chang Woo

AU - Dong, Zhao Yang

AU - Hur, Kyeon

N1 - Funding Information: Manuscript received May 13, 2020; revised September 3, 2020 and November 4, 2020; accepted December 2, 2020. Date of publication December 10, 2020; date of current version March 22, 2021. This work was supported in part by the Framework of International Cooperation Program Managed by National Research Foundation of Korea under Grant 2017K1A4A3013579 and in part by the “Human Resources Program in Energy Technology” of the Korea Institute of Energy Technology Evaluation and Planning (KETEP) Granted Financial Resource from the Ministry of Trade, Industry & Energy, Republic of Korea under Grant 20194030202420. Paper no. TSTE-00500-2020. (Corresponding author: Kyeon Hur.) Min-Seung Ko, Chang Woo Hong, and Kyeon Hur are with the School of Electrical and Electronic Engineering, Yonsei University, Seoul 03722, Korea (e-mail: kms4634500@yonsei.ac.kr; spearw@yonsei.ac.kr; khur@yonsei.ac.kr). Publisher Copyright: © 2010-2012 IEEE.

PY - 2021/4

Y1 - 2021/4

N2 - This paper presents a deep residual network for improving time-series forecasting models, indispensable to reliable and economical power grid operations, especially with high shares of renewable energy sources. Motivated by the potential performance degradation due to the overfitting of the prevailing stacked bidirectional long short-term memory (Bi-LSTM) layers associated with its linear stacking, we propose a concatenated residual learning by connecting the multi-level residual network (MRN) and DenseNet. This method further integrates long and short Bi-LSTM networks, ReLU, and SeLU for its activating function. Rigorous studies present superior prediction accuracy and parameter efficiency for the widely used temperature dataset as well as the actual wind power dataset. The peak value forecasting and generalization capability, along with the credible confidence range, demonstrate that the proposed model offers essential features of a time-series forecasting, enabling a general forecasting framework in grid operations. The source code of this paper can be found in https://github.com/MinseungKo/DRNet.git.

AB - This paper presents a deep residual network for improving time-series forecasting models, indispensable to reliable and economical power grid operations, especially with high shares of renewable energy sources. Motivated by the potential performance degradation due to the overfitting of the prevailing stacked bidirectional long short-term memory (Bi-LSTM) layers associated with its linear stacking, we propose a concatenated residual learning by connecting the multi-level residual network (MRN) and DenseNet. This method further integrates long and short Bi-LSTM networks, ReLU, and SeLU for its activating function. Rigorous studies present superior prediction accuracy and parameter efficiency for the widely used temperature dataset as well as the actual wind power dataset. The peak value forecasting and generalization capability, along with the credible confidence range, demonstrate that the proposed model offers essential features of a time-series forecasting, enabling a general forecasting framework in grid operations. The source code of this paper can be found in https://github.com/MinseungKo/DRNet.git.

UR - http://www.scopus.com/inward/record.url?scp=85097958346&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85097958346&partnerID=8YFLogxK

U2 - 10.1109/TSTE.2020.3043884

DO - 10.1109/TSTE.2020.3043884

M3 - Article

AN - SCOPUS:85097958346

VL - 12

SP - 1321

EP - 1335

JO - IEEE Transactions on Sustainable Energy

JF - IEEE Transactions on Sustainable Energy

SN - 1949-3029

IS - 2

M1 - 9290065

ER -

Deep Concatenated Residual Network with Bidirectional LSTM for One-Hour-Ahead Wind Power Forecasting

Abstract

Bibliographical note

All Science Journal Classification (ASJC) codes

UN SDGs

Access to Document

Other files and links

Cite this