Solar Panel Identification Via Deep Semi-Supervised Learning and Deep One-Class Classification

As residential photovoltaic (PV) system installations continue to increase rapidly, utilities need identify the locations of these new components manage unconventional two-way power flow and maintain sustainable management distribution grids. But, historical records are unreliable constant re-assessment active PV is resource-intensive. To resolve issues, we propose model solar detection problem in a machine learning setup based on labeled data, e.g., supervised learning. However, challenge for most limited labels or only one type users. Therefore, design semi-supervised one-class classification methods autoencoders, which greatly improve nonlinear data representation human behavior behavior. The proposed have been tested validated not synthetic publicly available set but also real-world from utility partners. numerical results show robust accuracy, laying down foundation managing distributed energy resources