80. A study on inputting time series meteorological information into area photovoltaic output prediction with Auto-Encoder

Abstract

Under the large penetration of photovoltaics (PV), precise prediction of PV output is necessary to operate power network properly. Authors developed the prediction model using Auto-Encoder which predicts total PV output in the area of Tokyo Electric Power Company (TEPCO) from forecast image of Global Horizontal Irradiance (GHI). The developed model can handle the spatial information of meteorological data.

In this paper, we investigate the accuracy of the prediction model with Auto-Encoder by inputting GHI data at different times to consider time series information. Numerical results of forecast for 4 months in 2018 are reported.