Open AccessConstruction of Forecast Model for Power Demand and PV Power Generation Using Tensor Product Spline Function
Author(s) -
Takuji Matsumoto,
Yuji Yamada
Publication year - 2021
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.179
H-Index - 26
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/812/1/012001
Subject(s) - smoothing , robustness (evolution) , econometrics , spline (mechanical) , demand forecasting , computer science , photovoltaic system , sample (material) , smoothing spline , electricity , mathematical optimization , operations research , economics , mathematics , engineering , spline interpolation , biochemistry , chemistry , electrical engineering , structural engineering , chromatography , bilinear interpolation , computer vision , gene
Forecasting power demand and photovoltaic (PV) power generation is indispensable for the economic operation of electric utility businesses. Herein, robust demand and PV power forecast models are proposed that enable electricity companies including new entrants to efficiently forecast from small sample size data by using only published weather forecasts. We further enhance the previously proposed forecast models by using the tensor product spline function to impose smoothing conditions, not only in the seasonal direction but also in the hour direction, thereby constructing effective models that can incorporate multiple explanatory variables. The empirical results of the estimated two-dimensional trends are consistent with the intuitive interpretation, and the validations of the out-of-sample forecast error obtained from the data of nine different areas ensure the high robustness of the proposed model.