Efficiency Evaluation and Comparisons of Solar Cell Technologies Based on Measurements from the Arabian Peninsula
Abstract
In this work, we model the power output (a.k.a., energy yield output) of three photovoltaic (PV) cell technologies using statistical learning tools and comparing their performance efficiency under real-field conditions. We introduce and train regression models to elucidate the relationship between irradiance and energy yield output. The training is performed on historical records obtained from an adverse-for-solar-panels location and a long period of time. We utilize both standard and robust estimation approaches to compute the model’s parameters. We explored various families of predictive models and found that the best-performing model includes an intraday variability factor. We then applied residual error analysis and relate the models’ coefficient with the efficiency of the solar cells. Our analysis showed that the efficiency decreases over time, and each PV technology has a different rate of deterioration. Moreover, we observe seasonal fluctuations of the efficiency for each PV technology which have been quantified. The decrease in efficiency during the summer months can reach up to 40% relative to the efficiency during the winter months.