Experimental Studies for Glass Light Transmission Degradation in Solar Cells Due to Dust Accumulation Using Effective Optical Scattering Parameters and Machine Learning Algorithm

Environmental impacts influence solar cell performance significantly. A harsh environment may cause temperature, wind speed, or humidity uncertainties. In the case of photovoltaic systems in desert or semidesert areas, dust accumulation critically impacted solar cell degradation. Herein, we developed a novel machine-learning-based optical model that estimates solar cell efficiency's degradation because of dust accumulation. Glass substrates were directed to dust for 16 weeks with a weekly characterization procedure. Samples were morphologically and optically characterized to be seeded into the scattering model. Morphological characterization measurements were conducted using SEM and EDX-mapping to explore the internal composition of dust particles. Optically, the dusty glass substrate sample's transmission spectra were measured using a UV-Vis spectrometer. Refractive index, thickness, and scattering particle radius have been extracted as effective parameters with fitting accuracy of 95%. The extracted dataset is then inputted into a machine-learning model to predict the transparency degradation in the glass substrate. Finally, predicted data is validated against the actual degraded monocrystalline cell, showing 89.4% of matching.