04.1 - Aerospace Grade Materials, Structural Analysis, Fatigue and Damage Tolerance
DAMAGE QUANTIFICATION ON COMPOSITE STRUCTURES USING NEURAL NETWORKS AND HYBRID DATA
L.P.S. Ferreira¹, M.O. Yano², S. Silva², C.A. Cimini¹; ¹UFMG, Brazil ;²UNESP, Brazil
This work aims to develop a damage classification model that can be trained with numerical and experimental data. A finite element model is adjusted to experimental data with Bayesian inference and used to simulate damaged conditions. A hybrid approach using experimental and numerical data is proposed to train a neural network and the remaining experimental data is used in the test phase.