137. SOLAR RESOURCE FORECASTING: FROM INSTRUMENTATION TO REAL TIME FORECASTING

Department: Mechanical & Aerospace Engineering
Research Institute Affiliation: Center for Energy Research (CER)
Faculty Advisor(s): Carlos Coimbra

Primary Student
Name: Lukas Nonnenmacher
Email: lnonnenm@ucsd.edu
Phone: 858-822-1367
Grad Year: 2014

Abstract
The variability and uncertainty in solar power production pose a major challenge in integrating this clean and abundant renewable energy with the power grid. In our research, we are exploring various forecasting techniques to reliably and cost- efficiently integrate solar power to the grid. For better understanding of solar variability, a network of high quality solar instrumentation has been set up at various locations in California and Hawaii. The highly granular solar irradiance (DNI, GHI and diffuse irradiance) data from these instruments is then analyzed to develop forecasting models. Comparisons of several regression and stochastic learning methodologies indicate that hybrid GA-ANN based forecasting models perform better for a number of different time horizons.

« Back to Posters or Search Results