Regional solar power forecasting

Operational Day-Ahead Solar Power Forecasting for Aggregated PV Systems with a Varying Spatial Distribution

“…Furthermore, the results show that a decreasing MAE can be expected as the inter-system distance or the number of aggregated systems grow. Similar results were found for a case study in Italy, where the performance of day-ahead PV power output forecast models improved when several regions were aggregated [3]. On the other hand, in this study the improvement is not observed for the model performance in terms of the ER with respect to the maximum ER.…”

“…The integration of variable renewable energy sources, i.e. solar PV, in the electricity grid poses challenges to grid operators in maintaining grid stability [3]. Moreover, the power output of PV systems may drop or increase by almost 70% within 1 min and can differ by more than 75% from 1 h to the following [4].…”

“…Fluctuations in PV power output are caused by variations in irradiance that is received at the plane of the PV array [3]. This observed irradiance can be described by a deterministic and a stochastic component [9].…”

“…On the other hand, the stochastic component is rather difficult to predict as it is affected by the chaotic nature of the atmosphere, where the formation, evolution and motion of clouds affect the Global Horizontal Irradiance (GHI) received at the surface. In addition, the power output of a PV system is also directly affected by the ambient temperature and wind speed [3].…”

“…Besides, proposed novel methods often gain a marginal performance improvement and are commonly executed using a single case study so that it remains uncertain if the found improvements are universal. This argument is supported by the fact that the forecast accuracy is greatly affected by the local climate conditions and therefore the location of interest [3]. The lack of an extensive comparison raises the question if these advanced models are worth developing and adopting for different users.…”