A Novel ML Approach for Real-Time Prediction of PV Production: Detection and Impact of Shading

With the surging adoption of renewable energy, robust and precise monitoring of photovoltaic (PV) systems is increasingly critical. Traditional diagnostic methods often fall short in large-scale scenarios. This paper introduces a newly patented machine learning method that integrates module temperature, irradiance, and solar angles to capture shading losses and performance deviations. By discretizing data into physically interpretable matrices, the model achieves an R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>≥0.947 on real 1 MW plant data. Its lightweight, platform-agnostic design facilitates seamless adoption, enhancing O&M services through accurate, real-time power predictions.