Identification analysis of defect topologies in structures based on inverse analysis using non-destructive hammering test data

(Application of topology optimization theory and machine learning)

Abstract

In this study, we present defect topology identification analysis in a structure using hammering response data based on the inverse analysis methods. As for the inverse analysis method, the topology optimization method and the machine learning are applied to identify defect topologies. In numerical experiments using the topology optimization theory, the artificial response data, i.e., numerical simulation result, are employed for the identification analysis, and the practical hammering response data are used for the identification analysis based on the machine learning. In the presentation, the characteristic of each method is introduced, and some considerations based on numerical experiments are shown.