About this Abstract |
Meeting |
2024 ASC Technical Conference, US-Japan Joint Symposium, D30 Meeting
|
Symposium
|
2024 ASC Technical Conference, US-Japan Joint Symposium, D30 Meeting
|
Presentation Title |
Residual strength prediction of composites after fatigue loading |
Author(s) |
ali ebrahimi, Farjad Shadmehri, Suong Van Hoa |
On-Site Speaker (Planned) |
ali ebrahimi |
Abstract Scope |
Under fatigue loading conditions, the static strength of composite materials progressively decreases until failure occurs. Knowledge of the residual strength is critical not only for preventing catastrophic failures but also for enhancing the efficiency of maintenance planning. This research proposes a novel approach to estimate the residual strength of glass epoxy composites subjected to fatigue loading. The technique utilizes a piezo-resistivity-based structural health monitoring (SHM) system to collect signals during fatigue loading. These data are subsequently used to predict residual strength through machine learning algorithms. Despite collecting extensive electrical resistance (ER) data, the direct analysis failed to identify clear patterns linking ER changes to residual strength, highlighting the data's complexity and the necessity of applying machine learning techniques to uncover hidden patterns. The study involved embedding carbon nanotubes into glass/epoxy samples to make them conductive and subjecting these samples to fatigue testing while continuously monitoring their electrical resistance. Key electrical characteristics, such as total and relative change in ER indicative of residual strength, were extracted from the data and used as potential input features for various machine learning algorithms to predict residual strength. The findings suggest that the proposed method can accurately predict the residual strength with an exceptionally low average error of less than five percent during the testing phase. |
Proceedings Inclusion? |
Definite: Post-meeting proceedings |