Multimodal AI Reveals Mechanism of Shape-Memory Alloy Degradation

bravo_abad · x · 2026-09-01

NiTi shape-memory alloys lose their shape-recovery ability after repeated loading, but the underlying microscopic processes have been difficult to disentangle.

Yuejie Hu and colleagues developed an interpretable multimodal learning framework that combines crystal-orientation maps, phase and dislocation-density imaging, nanoscale precipitate information, and loading conditions. Instead of simply stacking these distinct measurements, the model processes them separately before aligning and fusing their representations, achieving R²=0.981.

The interpretation reveals that the mechanism of degradation changes with the intensity of the material's strain. At lower strains, the process is irreversible…

Original post →

More from Research

Research channel →