系統定数推定機能付き強化学習LFCモデルの提案と実機検証

カテゴリ: 論文誌(論文単位)

グループ名: 【B】電力・エネルギー部門

発行日: 2021/06/01

タイトル(英語): Reinforcement Learning LFC Model with Power-Frequency Constant Estimation Function

著者名: 西村　翔太（東京理科大学），山口　順之（東京理科大学）

著者名(英語): Shota Nishimura (Tokyo University of Science), Nobuyuki Yamaguchi (Tokyo University of Science)

キーワード: 自動発電機制御，負荷周波数制御，系統定数，SARSA，再生可能エネルギー　　automatic generation control，load frequency control，power-frequency constant，SARSA，renewable energy

要約(英語): Nowadays, the introduction of renewable energy (RE) has been progressed from the perspective of considering environmental issues, there is concern that unstable output will increase. So system frequency control by existing generators becomes more important than ever. Even in system frequency control, Load Frequency Control (LFC) that support unpredictable short-period components of load fluctuations is expected to improve performance. Since Power-Frequency Constant of Area Requirement calculation of LFC changes from moment to moment, it is necessary to correctly estimate and update it to the best value in order to improve LFC performance.In this study, we propose a new LFC method that has a function of estimating Power-Frequency Constant for the purpose of improving the performance of LFC even in situations when the introduction of RE has expanded. We use reinforcement learning, which is a machine learning method that does not use training data, to estimate Power-Frequency Constant. For the verification, we use a simulated power system consisting of experimental devices such as simulated generators and loads.

本誌: 電気学会論文誌B（電力・エネルギー部門誌） Vol.141 No.6 （2021）

本誌掲載ページ: 484-491 p

原稿種別: 論文／日本語

電子版へのリンク: https://www.jstage.jst.go.jp/article/ieejpes/141/6/141_484/_article/-char/ja/