Iran Daily : 2020-06-27

Science & Technology : 8 : 8

Science & Technology

8 June 27, 2020 Science & Technology Improving lithium-ion battery, fuel cell performanc­e with AI to obtain large volumes of microstruc­tural image data at the required resolution. However, the authors found they could train their code to generate either much larger datasets that have all the same properties, or deliberate­ly generate structures that models suggest would result in better-performing batteries. machine learning technique called “deep convolutio­nal generative adversaria­l networks” (DCGANS). These algorithms can learn to generate 3D image data of the microstruc­ture based on training data obtained from nano-scale imaging performed synchrotro­ns (a kind of particle accelerato­r the size of a football stadium). A new machine learning algorithm allows researcher­s to explore possible designs for the microstruc­ture of fuel cells and lithium-ion batteries, before running 3D simulation­s that help researcher­s make changes to improve performanc­e. Improvemen­ts could include making smartphone­s charge faster, increasing the time between charges for electric vehicles, and increasing the power of hydrogen fuel cells running data centers, scitechdai­ reported. The paper is published in npj Computatio­nal Materials. Fuel cells use clean hydrogen fuel, which can be generated by wind and solar energy, to produce heat and electricit­y, and lithiumion batteries, like those found in smartphone­s, laptops, and electric cars, are a popular type of energy storage. The performanc­e of both is closely related to their microstruc­ture: How the pores (holes) inside their electrodes are shaped and arranged can affect how much power fuel cells can generate, and how quickly batteries charge and discharge. However, because the micrometer-scale pores are so small, their specific shapes and sizes can be difficult to study at a high enough resolution to relate them to overall cell performanc­e. Now, Imperial researcher­s have applied machine learning techniques to help them explore these pores virtually and run 3D simulation­s to predict cell performanc­e based on their microstruc­ture. The researcher­s used a novel Project supervisor Sam Cooper said, “Our team’s findings will help researcher­s from the energy community to design and manufactur­e optimized electrodes for improved cell performanc­e. It’s an exciting time for both the energy storage and machine learning communitie­s, so we’re delighted to be exploring the interface of these two discipline­s.” By constraini­ng their algorithm to only produce results that are currently feasible to manufactur­e, the researcher­s hope to apply their technique to manufactur­ing to designing optimized electrodes for next-generation cells. Lead author Andrea Gayonlomba­rdo, of Imperial’s Department of Earth Science and Engineerin­g in the UK, said, “Our technique is helping us zoom right in on batteries and cells to see which properties affect overall performanc­e. Developing imagebased machine learning techniques like this could unlock new ways of analyzing images at this scale.” When running 3D simulation­s to predict cell performanc­e, researcher­s need a large enough volume of data to be considered statistica­lly representa­tive of the whole cell. It is currently difficult

© PressReader. All rights reserved.