Predicting Location-Based Green Energy Availability in Smart Buildings

Mehdi Hatamian
Today, renewal energies have been gaining more attention across the globe due to their clean energy production. Solar energy is the most abundant renewable resource of energy on earth. Solar power represents a clean and green energy source in the energy transition era. Accordingly, the photovoltaic (PV) solar panel system is the most common way in which solar energy is captured.
Companies that generate energy need to predict the amount of energy sold in the electricity pool day-ahead or intra-day to maintain power production and demand in balance. Additionally, the Return on Investment (ROI) is what everyone wants to know for investing in PV solar panels.
Therefore, one of the concerns about solar panels is their efficiency due to the low reliability of certain renewable sources, including the variation of the weather conditions. Solar panels are highly dependent on how much sunlight they receive, and it's difficult to predict the amount of power generated by solar panels. Thus, reducing uncertainty is a solution by an energy forecasting tool that can predict the output of solar panels throughout the year. In this research, a detailed procedure is proposed to forecast the output power of PV solar panels by different machine learning models. The goal is to achieve a best-fitting model which is more accurate by inspecting the data precisely. Several predictive models will be compared to identify the best and most suitable ones for the described case.
Graduation Thesis language
Graduation Thesis type
Master - Computer Science
Chinmaya Kumar Dehury
Defence year