|Machine Learning-based Classification of Environmental Conditions for High-Throughput Indoor Testing of Photovoltaic Modules|
|Erin E. Looney1, Ian Marius Peters1, Haohui Liu2, Zekun Ren3, Tonio Buonassisi1
1Massachusetts Institute of Technology, Cambridge, MA, United States
/2Solar Energy Institute of Singapore, Singapore, Singapore
/3Singapore-MIT Alliance for Research and Technology, Singapore, Singapore
High-throughput testing of solar modules to accurately predict energy yield (EY) is increasingly important as more of the power grid runs on photovoltaics (PV). Modules are sold based on power ratings measured in standard testing conditions, not fully considering environmental conditions of the real world. In this work, we work towards developing a method to extract the best representative conditions of the environment that minimizes error in EY. We apply this method to define a representative set of solar spectra. To accomplish this, spectral data taken from Boulder, CO are clustered using the k-means algorithm. We show that by clustering we can reduce the size of the sample set by around 1000 times without reducing the uncertainty in EY by more than 5% relative. These results demonstrate the capacity for high throughput, accurate EY prediction.