A Neural Networks approach to characterize material properties using the spherical indentation test

Abstract

Determination of material characteristics using the instrumented indentation test has gained interests among many researchers. The output of a spherical indentation test is usually the load-penetration (P-h) curve. To achieve this goal, the elastic deformation of sphere must be eliminated from the penetration. To determine three parameters of the LUDWIG's equation which are σy, K and m, choice of a prompt numerical procedure is of essences. The purpose of the present work is to determination three parameters of the LUDWIG's equation using the spherical indentation test and Neural Networks. Therefore, a Neural Networks is trained following the spherical indentation test using two parameters that are obtained from the P-h curve. The output of the networks is the three parameters of the LUDWIG's equation. The results were then compared with the finite element predictions and verified using the experimental data. A good agreement was observed. Finally, the weights of Neural Networks layer were extracted for easy use of the above procedure.