Just In Timeモデリングとコレントロピーを用いた外れ値を考慮した配電系統状態推定へのディペンダブルな並列複数集団型Modified Brain Storm Optimizationの適用

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: 2020/02/01

タイトル(英語): Dependable Parallel Multi-population Modified Brain Storm Optimization for Distribution State Estimation considering Outliers using Just-In-Time Modeling and Correntropy

著者名: 東　大智（明治大学大学院 先端数理研究科 ネットワークデザイン専攻），福山　良和（明治大学大学院 先端数理研究科 ネットワークデザイン専攻），大井　章弘（富士電機（株）技術開発本部），神通川　亨（富士電機（株）技術開発本部），藤本　久（富士電機（株）技術開発本部）

著者名(英語): Daichi Azuma (Graduate School of Advanced Mathematical Science, Meiji University), Yoshikazu Fukuyama (Graduate School of Advanced Mathematical Science, Meiji University), Akihiro Oi (R&D Headquarters, Fuji Electric, Co. Ltd.), Toru Jintsugawa (R&D Headquarters, Fuji Electric, Co. Ltd.), Hisashi Fujimoto (R&D Headquarters, Fuji Electric, Co. Ltd.)

キーワード: 配電系統，負荷推定型状態推定，ジャストインタイムモデリング，コレントロピー，ディペンダビリティ，並列複数集団型モディファイドブレインストームオプティミゼイション　　distribution system，load adjustment distribution state estimation，just in time modeling，correntropy，dependability，parallel multi-population modified brain storm optimization

要約(英語): This paper proposes dependable parallel multi-population modified brain storm optimization for load adjustment distribution state estimation (DSE) considering outliers using just in time (JIT) modeling and correntropy. If the outliers are measured at the measurement points, estimation results at the measurement points only using correntropy are affected by the outliers. The challenge is solved by JIT modeling in the proposed method. Evolutionary computation techniques have been applied to the DSE problem because practical equipment causes a nonlinear characteristic of the objective function in distribution systems. However, it is required to obtain high and stable estimation quality in a short time considering uncertain outputs of renewable energies. The challenge is solved by multi-population, and parallel and distributed computing. Moreover, it is also required to keep appropriate estimation quality using dependable calculation even if some of the calculation results of distributed processes cannot be obtained due to various faults of the processes. The proposed method is verified to estimate distribution system conditions more accurately regardless of the outliers, realize faster computation, obtain higher and stabler estimation results, and keep higher quality of solutions even with high fault probabilities than the conventional methods.

本誌: 電気学会論文誌B（電力・エネルギー部門誌） Vol.140 No.2 （2020）　特集：令和元年電力・エネルギー部門大会

本誌掲載ページ: 68-77 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/140/2/140_68/_article/-char/ja/