Digital Image Correlation-Based Non-Destructive Defect Inspection of Additive Manufactured Metamaterials

Abstract

Metamaterials are artificially engineered structures constructed by unit cells and are designed to have unique mechanical, thermal, and electromagnetic properties. Therefore, they have a broad application in the mechanical, aerospace, and medical industries. However, due to their complex structures, their quality inspection can be challenging as internal defects, such as a missing beam due to design error or cyber-attack can only be observed by x-ray imaging, which can be costly and time-consuming. This study investigates a machine-learning-assisted digital image correlation approach as a non-destructive evaluation technique for identifying defects in metamaterials. Digital image correlation is an image-based strain field measurement technique. By tracking the speckle painted on the part, this method can detect deformation or movement. When under controlled load, If the part being examined has a defect, it will have different deformation compared to the standard part. A machine learning model has been trained to take the observed strain field as input and predict the location and the size of the defects. Preliminary results demonstrate that the proposed machine learning-assisted method can effectively identify various types of defects, such as in-complete beam connections and missing cells, and their locations, providing a cost-effective and rapid non-destructive method for 3D printed metamaterials.