ELPVPower: A dataset for large scale PV power prediction using EL images of cellshttps://doi.org/10.26165/JUELICH-DATA/TVWUUPHoffmann, MathisBuerhop-Lutz, ClaudiaReeb, LucaPickel, TobiasWinkler, ThiloDoll, BerndWürfl, TobiasPeters, Ian MariusBrabec, Christoph J.Maier, AndreasChristlein, VincentJülich DATA2020-08-282022-03-16T15:31:06ZMeasurements are provided in the folder `data` as PNG-Images. The original measurements have been rescaled to the range [0, 255] globally, such that intensities between measurements are comparable. The dataset comes with a file `data.csv` that provides additional meta data associated with the measurements. This includes: peak_power: The measured maximum power of the module in [W] nominal_power: The given nominal power of the module in [W] pressure: The pressure that has been applied during mechanical load testing on a linear scale [0, 1] excitation_class: Indicates, whether this measurement has been taken at a `high` or `low` excitation current module_type: Module types as specified in the paper module_instance: Some of the module identities have been measured multiple times under varying conditions. Here, we specify the distinct instances power_group: Discretized `peak_power` used for stratified sampling fold_i_train: Indicates, whether this module is part of the i'th training fold of the crossvalidationComputer and Information ScienceEarth and Environmental SciencesEngineeringPhysicsarXiv, arXiv:2009.14712, https://arxiv.org/abs/2009.14712Denz, Janine2020-08-27CCBY