Lamb wave damage severity estimation using ensemble-based machine learning method with separate model network

Lamb wave-based damage estimations have great potential for structural health monitoring. However, designing a generalizable model that predicts accurate and reliable damage quantification result is still a challenge due to complex behavior of waves with different damage severities. In recent years machine learning algorithms have been proven to be an efficient tool to analyze damage-modulated Lamb wave signals. In this study, ensemble-based machine learning algorithms are employed to develop a generalizable crack quantification model for thin metallic plates. For this, the scattering of Lamb wave signals due to different configurations of crack dimension and orientation is extensively studied. Various finite-element simulation signals representing distinct crack severities in terms of crack length, penetration, and orientation are acquired. Realizing that both temporal and spectral information of the signal is extremely important to damage quantification, three time-frequency-based damage-sensitive features, namely energy concentration, time-frequency flux, and coefficient of variance, are proposed. These damage features are extracted by employing smoothed-pseudo Wigner-Ville distribution. After that, data augmentation technique based on the spline-based interpolation is applied to enhance the size of the dataset. Eventually, these fully developed damage dataset is deployed to train ensemble-based machine learning models. Here, we propose a separate model network (SMN), in which different models are trained and then link together to predict new and unseen datasets. The performance of the proposed model is demonstrated by two cases: first, simulated data incorporated with high artificial noises and in the second scenario, experimental data in raw form are employed to test the model. Results indicate that the proposed framework has the potential to develop a general model that yields reliable results for crack estimation.