Deep learning-based prediction of interfacial conditions in coated plates using guided waves

This paper proposes a framework of using deep learning-assisted methods for the prediction of interfacial conditions in coated plates using guided wave data. The coating-substrate interface is modeled as a linear spring layer of zero thickness, and the mechanical behavior of this spring layer is characterized by the spring compliance. Both tangential and normal spring compliances are introduced to characterize the bond quality. Numerical simulations are conducted for a wide range of spring compliances to generate the corresponding dispersion curves. A long short-term memory (LSTM) network is utilized to predict the interfacial conditions. In addition, we consider the delamination cases where the coating layer is completely separated from the substrate over the delaminated region. Finite element simulations are carried out to model guided wave generation, propagation, interaction with delamination, and reception. The time-space images are formed by measuring the time-domain signals by receivers at several locations downstream from the source transducer, which are then fed into the developed convolutional neural network (CNN). Once trained, this deep-learning (DL) model enables the accurate prediction of delamination location and size. Results of this paper demonstrate that the proposed methodologies have tremendous potential for characterizing interfacial conditions in practical nondestructive evaluation (NDE) and structural health monitoring (SHM) applications.