A deep learning -based system to predict critical events (cracks, fatigue) in a strained polycrystal from synchrotron-based high-energy x-ray diffraction microscopy (HEDM) data. The deep net is trained on simulated x-ray diffraction from synthetic polycrystalline samples that are evolved with a crystal plasticity finite element (CPFE) model. The CPFE model was developed at Purdue University's Advanced Computational Materials lab. The diffraction is simulated with the MIDAS software suite developed at the Advanced Photon Source.