Austin, Texas — Rising electricity bills could soon become a thing of the past, thanks to innovative advancements in artificial intelligence. A team of researchers comprised of experts from the University of Texas at Austin, the University of Shanghai Jiao Tong, the National University of Singapore, and Umea University in Sweden has unveiled a cutting-edge machine learning technique designed to produce sophisticated three-dimensional thermal meta-emitters.
The research team has successfully engineered over 1,500 distinct materials, each capable of emitting heat in varying ways and intensities. These properties make the materials particularly beneficial for enhancing energy efficiency by offering targeted heating and cooling solutions.
Co-leader of the study, Yuebing Zheng, explained how this new automated approach enables the creation of materials with enhanced capabilities that were previously inconceivable. The researchers conducted experiments with four unique materials, applying one to a model house. Their findings highlighted the material’s impressive cooling effects; it kept the roof temperature 5 to 20 degrees Celsius (9 to 36 degrees Fahrenheit) lower than traditional white and gray paints after four hours in the sun.
This substantial temperature reduction could lead to significant energy savings, potentially conserving around 15,800 kilowatt-hours per year in a typical apartment building located in a warm climate. For perspective, a standard air conditioning unit consumes approximately 1,500 kilowatt-hours annually.
However, the applications of these thermal emitters extend beyond mere energy efficiency. Using machine learning, the researchers have categorized their findings into seven distinct classes of meta-emitters, each tailored for various uses. In busy urban environments, these emitters can play a crucial role in mitigating the urban heat island effect by reflecting sunlight while emitting heat at strategic wavelengths.
Beyond Earth, the implications of this technology could support temperature regulation for spacecraft, effectively managing heat through sunlight reflection and heat dissipation.
The neurons of everyday life also stand to benefit; these meta-emitters can be integrated into everyday fabrics and textiles, enhancing cooling technologies in clothing and outdoor equipment. They could also be applied to automobiles, helping to reduce heat accumulation when parked under the sun.
Historically, the slow and meticulous traditional processes of material design have limited the wide-scale adoption of such innovations. Previous automated attempts struggled to manage the complexities of three-dimensional meta-emitters, often producing only simple, flat designs that fall short in performance.
“Designing these materials has traditionally been a painstaking, trial-and-error process that often leads to less effective solutions,” Zheng noted. “Our approach allows for a more streamlined and efficient path to discovering optimal designs.”
Looking forward, the research team plans to refine these technologies and explore further applications in the field of nanophotonics, which investigates light-matter interactions at microscopic levels. Kan Yao, a co-author on the study and research fellow in Zheng’s group, articulated the unique promise of machine learning in this domain. “While it may not solve all challenges, its ability to meet the specific spectral demands of thermal management makes it exceptionally suited for developing high-performance thermal emitters.”
The findings from this pioneering research were published in a prestigious scientific journal, marking a significant step toward practical applications of advanced thermal management solutions.