Initial Results on Forecasting Residual Power Demand with Gradient Boosting Trees

カテゴリ: 部門大会

論文No: 186

グループ名: 【B】令和2年電気学会電力・エネルギー部門大会

発行日: 2020/08/28

タイトル(英語): Initial Results on Forecasting Residual Power Demand with Gradient Boosting Trees

著者名: Joao Gari da Silva Fonseca Junior（東京大学），宇田川佑介（構造計画研究所），大関崇（産業技術総合研究所），荻本和彦（東京大学）

著者名(英語): Joao Gari da Silva Fonseca Junior (The University of Tokyo), Yusuke Udagawa (Kozo Keikaku Engineering), Takashi Oozeki (AIST), Kazuhiko Ogimoto (The University of Tokyo)

キーワード: 残余電力需要の予測|勾配ブースティングツリー|太陽光発電の高導入|数値予報データ|Residual Power Demand Forecast|Gradient Boosting Trees|PV Power Large Penetration|Numerical Weather Forecasting Data

要約(英語): The objective of this study is to present the development of a method to forecast directly residual power demand in a scenario of power systems with high penetration of PV power. In the proposed method we train a gradient boosting regression trees model to forecast residual demand using basic weather prediction variables and calendar information directly. The forecast horizon is one day ahead of time, and data simulating the current conditions of Kyushu region in Japan was used. In the study we will discuss the initial results obtained, level of accuracy achieved and issues of the proposed approach.

PDFファイルサイズ: 685 Kバイト