Building a smart power grid is essential to supporting advanced city
infrastructure. A smart grid can be defined as a computer-driven power grid that
integrates efficiently and safely the actions of all connected entities: power plants and
consumers. The modern concept of smart grid management targets a real-time balance
of the inherent variability in the power production from renewable sources.
Consequently, short-term forecasting of power production became a key task in
providing smartness to the grid. Accurate forecasts allow computers to take control
actions to balance the grid. In this context, this chapter focuses on intra-hour
forecasting of photovoltaic (PV) power. A brief introduction to solar irradiance
variability is presented firstly. Then, a survey on the performance of solar irradiance
forecasting models is conducted. It is motivated by the commonsense observation that
the accuracy of forecasting the output power of a PV plant is highly conditioned by the
accuracy of forecasting the solar resource. The second part of the chapter includes a
review of several models for short-term forecasting the output power of a PV plant. A
critical survey on the metrics used for measuring the accuracy of the forecast is
presented as well. The chapter ends with a case study on short-term forecasting of PV
power, namely on the specific climate of southeastern Pannonia Plain. The study is
conducted with high-quality data measured on the Solar Platform of the West
University of Timisoara, Romania.
Keywords: Accuracy metrics, Intra-hour forecasting, PV power, Solar irradiance.
