ABSTRACT
Despite the COVID-19 pandemic, the global Photovoltaic market once more grew significantly in 2020, mainly on Grid-Connected Systems. The exponential increase of these systems raises new challenges for Smart Grid operators trying to predict load demand. That occurs because of the panels' output uncertainty and the leak of regional models for solar energy prediction. In this paper, we propose a distributed data approach to predict solar energy generated by Photovoltaic Systems. As input, we combine data from a community of solar panel owners and a historical weather website to build our dataset. This paper evaluates two scenarios: predicting regional next-day generation by using weather forecasting and the impact of a new system in that region. We sort the results seasonally and achieved 7% of MAE weighted percentage for summer predictions. © 2022 IEEE.