Improving Short-Term Local Solar Energy Forecasts for Optimizing Power Generation Using Machine Learning

This Praxis introduces the SAFE-T (Solar Advanced Forecast Expectation Tool) as an approach to improve local area solar energy forecasts using machine learning based techniques in MATLAB. With the increasing cost competitiveness of solar versus diesel electricity production, there has been a push by utilities to deploy solar power generating capacity across areas that have not historically been popular for solar power. (Georgitsioti, et al., 2015) While typical solar forecast models have been improving for standard locations, model accuracy has not kept up across all areas where forecasts are now desired. (White, et al., 1999) Recent advancements in large area solar forecasting have come from advanced physics and statistics methods. (Du, 2019) (Verbois, Huva, Rusydi, & Walsh, 2018) These improvements, however, have left a gap for medium term forecasting (48 to 96 hours) for single point areas in parts of the world that lack significant collections of historical solar energy measurements. This research focuses on developing a model to enable electric utilities in these areas to decrease forecast error by using a GlobalSearch based machine learning optimization combined with a multiple linear regression model to improve solar forecasts. With a less error prone forecast, electric utilities will be able to better optimize the amount of backup diesel generator capacity (and fuel) to deploy for a project. In doing so, utilities will be able to decrease the deployment of wasted resources, which can reduce the total cost of electricity generated by up to 35%. (Georgitsioti, et al., 2015)

Author

• Cwagenberg, Jeffrey D

Language

• en

Date created

• 2021

Type of Work

• Dissertation

License

• All rights reserved

Rights statement

• In Copyright

GW Unit

• Engineering Management

Degree

• D.Engr.

Advisor

• Eveleigh, Timothy
• Dimitriou, Aristos

Committee Member(s)

• Eveleigh, Timothy
• Dimitriou, Aristos
• Etemadi, Amir

Persistent URL

•  https://scholarspace.library.gwu.edu/etd/6d56zx37f