A probabilistic approach to the estimation of regional photovoltaic power production

摘要

Forecasting the total photovoltaic (PV) power generated in the control areas of the transmission system operators (TSO) is an important step in the integration of the large amounts of PV energy into the German electricity supply system. A standard approach for evaluating the regional PV power generation from weather forecast consists in upscaling the forecast of a limited set of reference plants to the complete area. Previous studies shown that this method can lead to large errors when the set of reference plants has different characteristics or weather conditions than the set of unknown plants. In this paper, an alternative to the upscaling approach is proposed. In this method, called a probabilistic regional PV model, an average PV model with a very limited number of inputs (two module orientation angles) is used to calculate the power generation of the most frequent module orientation angles. The resulting power values are finally weighted according to their probability of occurrence to estimate the actual power generation. The implementation of this model thus only requires information on the location and peak capacity of the plant installed in a region and no PV plant measurement is necessary. The proposed method has been evaluated against the estimate of the total power generation provided by the German TSOs, which shows that an RMSE ranging from 4.2 to 4.9% can be obtained with this method using on IFS meteorological forecast. The regional power forecasted with the probabilistic approach was also compared to the day-ahead forecast disseminated by the TSO. This analysis shows that,the forecast evaluated with the proposed approach has an RMSE less than 0.5% higher than the reference forecasts. This is considered a promising result given that the forecast evaluated with the probabilistic model is based on one single weather model and that - at the exception of the model calibration - no statistical post-processing method is used to optimize its performance. (C) 2017 Elsevier Ltd. All rights reserved.