Abstract
Millions of solar panels have been deployed around the world for electricity production. Meanwhile, these solar panels are routinely generating a massive amount of streaming data of current and voltage as they operate at the maximum power point (MPP). The existing characterization methods, however, cannot effectively mine and decode these datasets to provide useful insights of the degradation for solar panels. In this paper, we propose the new Suns-Vmp methodology, which offers a degradation-agnostic approach to monitoring and diagnosing the reliability of solar panel in real time by physically interpreting the MPP data. The physics-based method reconstructs “IV” curves based on MPP characteristics under varying illumination and temperature, from which time-series circuit parameters can be extracted using equivalent circuits. The proposed method has been applied to analyze the data obtained from an NREL test facility. Our Suns-Vmp based analysis indicates that the solar panels have failed within nine years with a rate of ~3 %/year, mostly likely due to yellowing/delamination of EVA and contact corrosion. Integrated with degradation models or machine learning, the method can also serve to predict the lifetime of PV systems.