Assessment of mechanical properties of Incoloy800H by means of small punch test and inverse analysis

This paper describes the application of inverse finite element model in analysis of material characterization by small punch test. A novel inverse methodology based on Least squares-support vector machine (LS-SVM) model was presented. Tensile constitutive behavior analysis was performed on Incoloy800H by LS-SVM model, which was trained by simulation results with varied parameters. The predicted results show that material parameters can be related to small punch curves successfully by an iterative analysis of LS-SVM model. In the aspect of single parameter evaluation, the initial yield stress of Incoloy800H was investigated by means of three different inverse models which were established based on gold section, LS-SVM and enumeration, respectively. Compared to enumeration, gold section algorithm can narrow the prediction range quickly as only one parameter is required to be calculated. The predicted error is about 2.55%, however, the calculation time is apparently less than that of enumeration.