Document Type


Source Publication Title

Proceedings of The American Society for Composites; Thirty-seventh Technical Conference


Fiber reinforced epoxy based composite materials are widely used in aircraft, marine, and automotive structures to minimize weight and increase performance. Typically, the mechanical performance of a composite part has so far been inspected using destructive testing, which is both expensive and time consuming. The mechanical properties of a composite are influenced by the curing process and its parameters, particularly the mechanical tensile modulus of the polymer, which rises as crosslink density rises and is influenced by curing temperature and holding time. It is, however, critical to manufacture epoxy-based composites with a higher crosslinking density by using the proper and uniform curing temperature. After fabricating the composite part, the most important step is to inspect the mechanical strength. In this study, we have presented a non-destructive quality inspection technique by merging Broadband Dielectric Spectroscopy data with supervised learning algorithms to predict the tensile strength of the cured composite sample based on the sample’s dielectric state variables. Different samples were made using different curing temperature to ensure variability in tensile strength data. A total of 15 features based on the Dielectric characteristics of the sample has been used in the algorithm. In this case, a classifier algorithm has been utilized, which predicts tensile strength based on which subset it belongs to. Overall, an estimation of the mechanical strength can be obtained by measuring the dielectric characteristic using an AC current frequency sweep which is a time-saving and nondestructive procedure and fitting the data in a supervised learning algorithm. []


Engineering | Materials Science and Engineering

Publication Date





Contact if you are the author.

Available for download on Wednesday, January 01, 3000