作者: Alexander Phinikarides , Nitsa Kindyni , George Makrides , George E. Georghiou
DOI: 10.1016/J.RSER.2014.07.155
关键词:
摘要: Abstract This paper provides a review of methodologies for measuring the degradation rate, RD, photovoltaic (PV) technologies, as reported in literature. As presented this paper, each method yields different results with varying uncertainty depending on equipment, data qualification and filtering criteria, performance metric statistical estimation trend. imposes risk overestimating or underestimating true rate and, subsequently, effective lifetime PV module/array/system proves need defining standardized methodology. Through literature search, four major analysis methods were recognized calculating rates: (1) Linear Regression (LR), (2) Classical Seasonal Decomposition (CSD), (3) AutoRegressive Integrated Moving Average (ARIMA) (4) LOcally wEighted Scatterplot Smoothing (LOESS), LR being most common. These analyses applied following metrics: electrical parameters from IV curves recorded under outdoor simulated indoor conditions corrected to STC, regression models such Photovoltaics Utility Scale Applications (PVUSA) Sandia models, normalized ratings Performance Ratio, RP, PMPP/GI scaled PMPP/Pmax, PAC/Pmax kWh/kWp. The have shown that produced lowest RD large variation largest uncertainty. ARIMA LOESS methods, albeit less popular, low good agreement between them. Most importantly, showed is not only technology site dependent, but also methodology dependent.