Ultrasonic-guided waves can be used to monitor the health of thin-walled structures. However, the run of experimental damage tests on materials like carbon fiber-reinforced plastics can be impractical and costly. Instead, numerical models can be used to create hybrid datasets to train machine learning algorithms, integrating data from numerical and experimental tests. This paper presents a Bayesian-driven framework to compensate for limited experimental data regarding Lamb wave propagation in composite plates. Using Bayesian inference, the framework updates a numerical finite element model, considering observed uncertainties by sampling posterior probability density functions for input parameters using Markov–Chain Monte Carlo simulations with the Metropolis-Hastings algorithm. A neural network surrogate model speeds-up these simulations, leading to a model that replicates the uncertain experimental setup. This model then generates data to augment true experimental data. Finally, a one-dimensional convolutional neural network is trained on a three different datasets to analyze Lamb wave signals and assess damage. Comparing training strategies shows the hybrid dataset augmented by samples generated by the updated FE model gives the most accurate damage size predictions.
1 January 2025 in Latin American Journal of Solids and Structures
Buckling is a critical failure mode for aircraft composite panels. Therefore, determining the critical buckling load is essential for properly modeling. To optimize composite laminate structures, it is important to know the strength of the laminate, which can be obtained through statistical analyses based on test data. The resultant strength value is known as a design allowable. This paper aims to establish an approach to evaluate two types of statistically determined buckling allowables, A-Basis and B-Basis, for an aircraft composite wing. In this study, a finite element model of a composite semi-wing was used to perform linear buckling simulations. Six input parameters were initially selected as relevant to affect the buckling strength of the semi-wing: four material properties and two geometric parameters. A set of Monte Carlo simulations was conducted varying these parameters of interest, followed by a global sensitivity analysis using Sobol indices to identify the influence of each one, individually (first-order) and in pairs (second-order). A surrogate model based on artificial neural networks was then trained using data from the Monte Carlo simulations. Finally, this surrogate model was used to define the A-Basis and B-Basis allowables for a critical loading case.
26 September 2025 in Journal of Sound and Vibration