Dynamic fracture of a bicontinuously nanostructured copolymer: A deep-learning analysis of big-data-generating experiment

Here, we report measurements of detailed dynamic cohesive properties (DCPs) beyond the fracture toughness a bicontinuously nanostructured copolymer, polyurea, under an extremely loading rate, from deep-learning analyses big-data-generating experiment. We first describe new Dynamic Line-Image Shearing Interferometer (DL-ISI), which uses streak camera to record optical fringes displacement-gradient vs time profile along line on sample's rear surface. This system enables us detect crack initiation and growth processes in plate-impact experiments. Then, present convolutional neural network (CNN) based framework, trained by extensive finite-element simulations, that inversely determines accurate DCPs DL-ISI fringe images. For measurements, experiments were performed set samples with mid-plane crack. A Conditional Generative Adversarial Networks (cGAN) was employed reconstruct missing recorded partial fringes. CNN correlation method applied fully reconstructed get toughness, 12.1kJ/m^2, strength, 302 MPa, maximum separation, 80.5 um, within 0.4%, 2.7%, 2.2% differences, respectively. time, polyurea have been successfully obtained pre-trained cGAN-reconstructed data sets. The strength is found be nearly three times higher than dynamic-failure-initiation strength. high stem both ductility separation.