Deep Learning for Automated Detection of Rooftop Solar System in Vietnam

The proportion of renewable energy system (RES) in electricity production has increased significantly in recent years, thanks to strict environmental regulations, fossil fuel sources is limited and the global demand for electricity is rising. In Vietnam, rooftop solar power is developing rapidly, however, because of challenges in operating power systems with a high penetration of RES, these systems were not allowed to feed into the grid in 2022. This has created difficulties for managers and operators in forecasting sources and loads, since the capacity of these systems cannot be controlled. To address this issue, the paper proposes an deep learning-based solution, such as the YOLOv8 model, to analyze data and automatically detect solar power systems captured by drones accurately and quickly in some places in Quang Nam Province and Da Nang. From the analysis result, the problem of forecasting distributed sources from rooftop solar systems in Vietnam in the current context will be solved in our next work.