Design and optimization of multilayered ultra-broadband solar thermal absorber using machine learning for industrial thermal applications

For the new renewable energy system, the proposed structure for the thermal absorber can promote carbon dioxide reduction and environmental damage prevention. Machine learning optimization is applied to the design, which leads to higher thermal absorption of the solar absorber. The materials and layer sections are crucial in achieving a highly efficient absorber. The developed design consists of three layers made from different materials: Titanium (Ti) for the resonator (square shape), Tin (IV) selenide (SnSe 2 ), and Iron (III) oxide (Fe 3 O 4 ) for the middle ground layers. With a bandwidth range of 1.3 – 1.9 μm (600 nm bandwidth), the output reaches 98.26%, and the overall absorption at 2800 nm is 92.85%. Results for different atmospheric layers (UV-MIR) are also presented. With an R 2 value above 0.99, the parametric variation of the current layers can be analyzed with the Linear Regression Method as part of machine learning processing. The currently investigated broadband-type absorber, with multiple potential applications, can be optimized for solar energy harvesting, thermal imaging, photovoltaics, and optoelectronic devices.