Weekly Peak Load Forecasting for 104 Weeks Using Deep Learning Algorithm
- Authors
- Kwon B.-S.; Park R.-J.; Song K.-B.
- Issue Date
- Dec-2019
- Publisher
- IEEE Computer Society
- Keywords
- Deep learning; Long short-term memory; Mid-term load forecasting; Weekly peak load forecasting
- Citation
- Asia-Pacific Power and Energy Engineering Conference, APPEEC, v.2019-December
- Journal Title
- Asia-Pacific Power and Energy Engineering Conference, APPEEC
- Volume
- 2019-December
- URI
- http://scholarworks.bwise.kr/ssu/handle/2018.sw.ssu/35758
- DOI
- 10.1109/APPEEC45492.2019.8994442
- ISSN
- 2157-4839
- Abstract
- Load forecasting is one of important issue in the power system. Especially, accurate mid-term load forecasting plays an essential role for maintenance of generators and transmission systems. Because load has time-series characteristic and is closely related to weather and economic conditions, it is necessary to develop forecasting method considering these non-linear characteristics of loads. To develop a good forecasting method that effectively reflects these non-linear features of loads, load forecasting method using deep learning algorithm is proposed. Proposed forecasting method forecasts the weekly peak load over the next 104 weeks. In order to forecast weekly peak load, forecasting method is constructed by converting the input data, such as load, temperature and GDP into weekly units. Comparing the forecasting accuracy of multiple regression method with the forecasting accuracy of proposed forecasting method, the proposed forecasting method shows good forecasting performance than multiple regression method. © 2019 IEEE.
- Files in This Item
- There are no files associated with this item.
- Appears in
Collections - College of Engineering > School of Electrical Engineering > 1. Journal Articles
![qrcode](https://api.qrserver.com/v1/create-qr-code/?size=55x55&data=https://scholarworks.bwise.kr/ssu/handle/2018.sw.ssu/35758)
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.