A surrogate model for finite element simulation of elastic wave propagation in composite materials

Deep learning surrogate models can be employed in solid mechanics to forecast the behavior of structures subjected to various loading conditions, substantially decreasing the computational costs associated with simulations. In this letter, we have utilized convolutional neural networks and Fourier transform to predict the elastic wave output from composite bars. The microstructures of the bar are utilized as inputs to the deep learning model, while the output is the elastic wave response. The convolutional neural network learns to identify crucial input composite features and utilizes this information to predict the output elastic waves. Finally, the mean squared error of the predicted output signals is compared to the actual output signals, which was used to evaluate the model. The outcomes of this study demonstrate that the deep learning model can precisely and swiftly predict the output elastic waves of the composites, thus serving as a surrogate model for time-consuming finite element simulations.