PREDICTION OF DAILY SOLAR RADIATION WITH ANN MODEL USING INFLUENTIAL INPUTS SELECTED BY FACTOR ANALYSIS

電気学会 電子図書館 Verified independent store

→

カテゴリ: 研究会(論文単位)論文No: PSE22019グループ名: 【B】電力・エネルギー部門 電力系統技術研究会発行日: 2022/01/18タイトル(英語): PREDICTION OF DAILY SOLAR RADIATION WITH ANN MODEL USING INFLUENTIAL INPUTS SELECTED BY FACTOR A NALYSIS 著者名: MPAMBA SHAMBUYI Alain(Gifu University),Takano Hirotaka(Gifu University),Asano Hiroshi(Gifu University(*) / Central Research Institute of Electric Power Industry )著者名(英語): Alain MPAMBA SHAMBUYI(Gifu University),Hirotaka Takano(Gifu University),Hiroshi Asano(Gifu University(*) / Central Research Institute of Electric Power Industry )キーワード: Artificial Neural Networks | Photovoltaic Generation Systems|Solar Radiation|Meteorological Variables|Factor Analysis|Artificial Neural Networks | Photovoltaic Generation Systems|Solar Radiation|Meteorological Variables|Factor Analysis要約(日本語): For both standalone and grid-connected photovoltaic generation systems (PVs), it is necessary to collect solar radiation information beforehand as it is used for different purposes. In this article, an Artificial Neural Network (ANN)-based prediction mode要約(英語): For both standalone and grid-connected photovoltaic generation systems (PVs), it is necessary to collect solar radiation information beforehand as it is used for different purposes. In this article, an Artificial Neural Network (ANN)-based prediction model is proposed. In this model, meteorological variables are used as inputs. The most influential inputs are selected through the Factor Analysis (FA) method. From nine inputs available in our dataset for numerical simulations, only four inputs are selected as a result of application of FA to the dataset. The number of neurons in the ANN’s hidden layer is calculated using an experimental formula. Numerical simulations show us the lower neurons’ number to use for the model. Consequently, the proposed model has a simple structure as compared to many existing models because fewer neurons and only influential inputs are used. A simple model is easy to implement and requires lower computer capacity, which is economically beneficial本誌: 2022年1月21日電力系統技術研究会本誌掲載ページ: 95-100 p原稿種別: 英語PDFファイルサイズ: 1,137 Kバイト Show more