Weather dependent performance: researchers want to improve PV yield forecasts
15.12.2023 As the number of photovoltaic systems continues to rise, it is becoming increasingly important to be able to make accurate yield forecasts. A group of researchers is pursuing this goal in a four-year SNSF project.
The amount of electricity generated by photovoltaic systems depends heavily on the weather. Forecasts of the power output are therefore of foremost importance for grid operators, who must react with foresight to try to prevent problems in the grid that arise when the solar power input is different than expected. Power traders are also interested in good forecasts, which allow them to evaluate price trends on the electricity exchanges. The intraday solar power forecasts that are commonly used today are mainly based on weather forecast models and tend to lead to important forecast errors, especially in fluctuating and cloudy weather conditions. In order to enable the cost-efficient operation of solar plants and reduce the costs incurred by grid operators and utility companies for their standby and storage capacities and sub-optimal supply strategies on electricity markets, more accurate intraday forecasts of the solar power production and a better understanding and quantification of forecast uncertainty are needed.
Reliable and accurate intraday forecasts
A group of researchers led by Angela Meyer, professor of Applied Machine Learning and Industrial Analytics at Bern University of Applied Sciences BFH, has taken up this challenge in a project funded by the Swiss National Science Foundation. Their aim is to develop a framework that enables more accurate and reliable intraday forecasts of solar radiation over the Swiss Plateau, where a major part of current and additional future Swiss photovoltaic capacity is located. Within the project, the researchers are also developing more precise methods for short-term forecasts of the PV power generated. This should enable improved risk management and lower operating costs.
PV potential at altitudes above 1,000 m is underestimated
The project, in which BFH, ETH Zurich and the Finnish Meteorological Institute are involved, will run until the end of 2025. The first interim results have been published. They show that estimates of solar potential based on satellite data are subject to major errors and that solar potential is underestimated in areas located above 1,000 meters, especially in winter.