Skip to main content
. 2026 May 15;13(13):nwag287. doi: 10.1093/nsr/nwag287

Figure 2.

For image description, please refer to the figure legend and surrounding text.

Thermal parameters inference network and the simulation/experiment results. (a) Net2 (upper panel) is used to derive the time derivative for the diffusion equation, while Net1 (bottom panel) is responsible for deriving the diffusivity and spatial partial derivatives for the diffusion equation. The combination of these two networks, along with the diffusion equation, allows for the derivation of the temperature evolution, which is then compared to the actual thermal field evolution. (b) Predicted thermal diffusivity distribution based on the simulation dataset. (c) Actual thermal diffusivity distribution in the simulation. (d) A sample printed using 3D printing technology, featuring an externally enclosed structure with two concentric holes in the center. The top right inset shows a physical image of the sample. (e) Inferred thermal diffusivity distribution of sample shown in (d).