Graph-based AI derives interpretable constitutive equations from solid-mechanics data

Researchers at the Eastern Institute of Technology in Ningbo published a graph-based method in Science Advances that extracts explicit constitutive equations directly from experimental solid-mechanics data. The framework, GraphED, encodes equations as directed graphs so equation topology and material parameters are optimized together, rather than relying on predefined empirical formulas. The team reported that discovered equations for alloy steels, lithium metal, and filled rubbers matched experimental measurements more closely than mainstream empirical models, including the Johnson-Cook model for alloy steels. The work is a peer-reviewed laboratory result; the source does not establish deployment in industrial design or manufacturing workflows.