Probabilistic Solar Irradiance Forecasting by Conditioning Joint Probability Method and Its Application to Electric Power Trading

Abstract

This study presents a method for probabilistic forecasting of solar irradiance based on the joint probability distribution function (PDF) of irradiance predicted by numerical weather prediction (NWP) and irradiance observed. Multidimensional kernel density estimation was used to construct this joint PDF. The probabilistic forecast is obtained by deriving a conditional PDF given a current NWP by using the Bayes rule. The proposed method can naturally handle the nonlinear nature of the relation between observed and predicted irradiance. The method showed better statistical performance in terms of the continuous ranked probability score, reliability, and sharpness than an existing probabilistic forecasting method based on ensemble NWP. Simulation of solar power trading based on pricing of an actual electric power market also confirmed that the proposed method results in greater profit by suppressing penalties imposed due to overbidding.