A k?means clustering machine learning?based multiscale method for anelastic heterogeneous structures with internal variables

A new machine-learning based multiscale method, called k-means FE 2 , is introduced to solve general nonlinear problems with internal variables and loading history-dependent behaviors, without use of surrogate models. The macro scale problem reduced by constructing clusters Gauss points in a structure which are estimated be the same mechanical state. clustering—machine learning technique employed select on their strain state sets variables. Then, for all cluster, only one micro solved, its response transferred integration cluster terms properties. solution converges respect number clusters, weakly depends mesh elements. Accelerations calculations up factor 50 observed typical applications. Arbitrary behaviors including can considered at level. method applied heterogeneous structures local quasi-brittle elastoplastic and, particular, nuclear waste package subject expansions.