著者
小松 大起 黒岩 丈介 小高 知宏 諏訪 いずみ 白井 治彦
出版者
福井大学大学院工学研究科
雑誌
福井大学 学術研究院工学系部門 研究報告 = Memoir of Faculty of Engineering, University of Fukui
巻号頁・発行日
vol.70, pp.29-36, 2021-10

In this paper, we investigate how to apply deep learnings for multimodal inputs. The target problem is the prediction of power consumption, which enable us to control the operating time of each power plant in the short term and adapt the necessary amount of fossil fuels and other resources in the long term. In this paper, therefore, we perform the prediction of power consumption by using LSTM, which is a model that can handle time-series data. We employ the combination of temperature,precipitation and/or weather as multimodal inputs, which should be meteorological factors for the power consumption. The prediction is depend on the model structure and the combination of data.