Abstract
In this work, we present a comparison of Performance Loss Rate calculation methodologies based on originally 8,367 PV system datasets, primarily installed in central Europe. The system data are summarized in the Pearl PV database. Pearl PV is a European COST action aiming at improving the energy performance and reliability of PV systems. The datasets consist of only power measurements. Significant dataset filtering was necessary to remove short time series as well as outliers and resulted in the exclusion of over 90% of the datasets. Three different PLR calculation methodologies have been tested, namely statistical clear-sky fitting, the Year-on-Year approach and seasonal and trend decomposition. Depending on the chosen approach, average system performance losses between -0.75%/a and -1.1%/a have been recorded.