10.1
Improvements in short-term solar energy forecasting

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Thursday, 6 February 2014: 8:30 AM
Room C114 (The Georgia World Congress Center )
Sergio A. Bermudez, IBM T. J. Watson Research Center, Yorktown Heights, NY; and S. Lu, M. A. Schappert, T. G. van Kessel, and H. F. Hamann

The growing renewable energy sector is now challenged to understand and accurately forecast rapid changes in solar irradiance over limited spatial domains. Rapid solar-irradiance variations are primarily due to cloud movement and development (during daytime) and changes in sun angle at sunrise and sunset. Cloud-cover changes significantly impact solar-power generation plants and their interconnection to the electric transmission system, as the electric grid system operators must continuously balance supply and demand to maintain the reliability of the power grid. Towards that end we have developed a new sky imaging system and algorithms, which allow detecting and forecasting the arrival of incoming clouds.

The field of view of the developed imaging system is approximately 2 miles, which yields in up to 10 min prediction horizons at ~ 15 miles per hour cloud movements. After some geometrical corrections of the cloud images optical flow algorithms are applied, which effectively remove the "background" in the image by distinguishing between moving objects (i.e, clouds) and stationary objects (e.g, trees etc.). Cross-correlation between consecutive cloud images yields the direction and speed of the cloud, which in combination with the sun's position and the location of the solar panels is used to predict the solar energy. Although the cloud recognition algorithm is (so far) binary we adjusted the level of solar energy using a modification function F, which depends on time of the day and is based on past data for a "clear" and "cloudy" days (for a given date of the year). Further adjustments (i.e., level of shading) are made continuously using the actual measurements of the solar panels, which are available a couple of minutes later after the prediction was made.

In this paper we demonstrate how this approach was used to make continuously short-term forecasts (<10 min ahead) for a small solar installation. The average forecast error ε was estimated by ε=∫|PM(t)-PF(t)|dt / ∫PM(t)dt with PF(t) and PM(t) as the forecasted and produced/ measured power, respectively yielding 10% for a 6 min forecasts and slightly less for a 3 minutes forecasts, which is one of the highest reported to date.

Acknowledgement: The work is partially funded by the DOE SunShot program.