Accurate prediction of the energy produced by photovoltaic modules is a key issue to make a proper integration into the energy grid. Even though crystalline silicon based modules have a bigger presence on photovoltaic installations, thin film technologies are getting more market share. The effect of varying spectrum is not always taken into account but several studies indicates that spectrum effect should not be obviated in thin film technologies, especially those with narrower spectral response, as happens for amorphous silicon technologies. A new method to show the performance ratio of thin film modules having into account the spectrum is developed in this work by clustering all the spectra into a few groups. For this characterization, both, statistical and data mining techniques have been used to cluster all spectra in groups by means of its APE value. Afterwards a depiction using contour graphs of modules PR indexed by clearness index and module temperature has been carried out in every one of the APE clusters. The data used have been registered in the Photovoltaic Laboratory of the University of Malaga for more than one year in order to take into account seasonal effects that could occur.
Buscar en Dimension