Graphene-based metasurface solar absorber design with absorption prediction using machine learning

Abstract

AbstractSolar absorber is required to absorb most of the energy of the solar spectral irradiance. We propose a graphene-based solar absorber design with two different metasurfaces to improve this absorption and increase the efficiency of the solar absorber. The metasurfaces are selected based on their symmetrical/asymmetrical nature (O-shape and L-shape). The O-shape metasurface design is showing better performance over the L-shape metasurface design. The absorption performance is also compared with AM 1.5 solar spectral irradiance to show the effectiveness of the solar absorber. The absorption values are also enhanced by varying the parameters like resonator thickness and substrate thickness. The proposed solar absorber design gives maximum absorption in the ultraviolet and visible range. Furthermore, the design is also showing a high and similar absorption rate over a wide angle of incidence. The absorption of O-shape metasurface design is also predicted using machine learning. 1D-Convolutional Neural Network Regression is used to develop a Machine Learning model to determine absorption values of intermediate wavelength for assorted values of angle of incidence, resonator thickness, and substrate thickness. The results of experiments reveal that absorption values may be predicted with a high degree of accuracy. The proposed absorber with its high absorbing capacity can be applied for green energy applications.