Short-Term Dni Forecasting Using Cloud History Patterns Matching

Huahui Zhang,Chongyang Zhang,Wei Li

DOI: https://doi.org/10.1109/icist.2017.7926492

2017-01-01

Abstract:In this work, one novel short-term Direct Normal Irradiance (DNI) forecasting method using cloud history patterns matching (CHPM) is proposed. Unlike existing persistent and linear DNI prediction schemes that use cloud coverage only, the proposed CHPM method considers both the cloud coverage and the Influence of other Atmospheric Particulates (IoAP), such as the absorbing, reflecting and scattering of DNI by aerosols, particles and water vapor column. Considering the temporal similarity in a short time, the IoAP can be predicted using historical data by CHPM. Our experimental results using TSI (Total Sky Imager) images show that the proposed method outperforms the existing persistent and linear prediction approaches in term of root mean square error (RMSE).