In a new development, AI may find use to explore possible designs for the microstructure of lithium-ion batteries and fuel cells. This may help prevent researchers run 3-D simulations to make changes for improved performance.
The improvements in fuel cells and lithium-ion batteries using AI could help smartphones charge faster, increase the power of hydrogen fuel cells that run data centers, and increase the time interval between charging electric vehicles.
The findings of the study is published in npj Computational Materials.
AI to help Examine Microstructure of Fuel Cells
Meanwhile, fuel cells use renewable hydrogen fuel, which is generated using solar and wind energy, to generate heat and electricity. And, lithium-ion batteries used in laptops, smartphones, and electric cars are a popular form of energy storage. The performance of fuel cells and lithium-ion batteries is significantly related to their microstructure. This is closely related to how pores inside the electrodes are shaped and the arrangement can affect power that fuel cells generate, and how quickly charging and discharging of battery happens.
Nonetheless, due to extremely small size of micrometer-scale pores, their specific sizes and shapes can be difficult to examine at resolution high enough to relate them to overall cell performance.
To present a solution to this, researchers at Imperial have machine learning techniques. This helps them to explore the pores virtually and perform 3-D simulations to foretell cell performance based on their microstructure.
For this, the researchers employed a novel machine learning technique “deep convolutional generative adversarial networks. The algorithms of this novel machine learning technique gains insights to generate 3-D image data of the microstructure. And, the image generated using algorithms of novel machine learning technique are based on training data of synchrotrons obtained from nano-scale imaging.