Machine learning plastic deformation of crystals

Abstract

Plastic deformation of micron-scale crystalline solids exhibits stress-strain curves with significant sample-to-sample variations. It is a pertinent question if this variability is purely random or to some extent predictable. Here we show, by employing machine learning techniques such as regression neural networks and support vector machines that deformation predictability evolves with strain and crystal size. Using data from discrete dislocations dynamics simulations, the machine learning models are trained to infer the mapping from features of the pre-existing dislocation configuration to the stress-strain curves. The predictability vs strain relation is non-monotonic and exhibits a system size effect: larger systems are more predictable. Stochastic deformation avalanches give rise to fundamental limits of deformation predictability for intermediate strains. However, the large-strain deformation dynamics of the samples can be predicted surprisingly well.

Machine learning and predicting the time-dependent dynamics of local yielding in dry foams

“…We pose the following question: Given a single frame, are we able to predict the occurrence of a T1 event in the subsequent frame or frames? We start with a convolutional neural network (CNN) [33,34] and modify it to accept grayscale and skeletonized image as an input. Figure 3(a) depicts a schematic illustration of the CNN application leading to a single binary classification: Either T1 event occurs after t seconds or not.…”

Plastic yielding and deformation bursts in the presence of disorder from coherent precipitates

“…The emergence of experimental techniques such as compression of micron-scale samples with nanoindentors [3][4][5][6] and high-resolution acoustic emission (AE) measurements of bulk samples [7] has revealed a novel paradigm: dislocation plasticity is a spatiotemporally fluctuating and intermittent process [8]. On micron scales, discrete strain bursts with a broad size distribution can be seen directly in the stress-strain curve [9][10][11][12]. Macroscopic samples tend to exhibit a smooth stress-strain curve, but AE measurements show acoustic energy bursts spanning several orders of magnitude in energy [7,13].…”

Prediction of creep failure time using machine learning

“…Predictions of irreversible deformation and failure processes were based on data describing local atomic structure in amorphous solids 12,13 , mesoscale microstructures 14 (for an overview see e.g. 15 ), as well as monitoring data obtained in macrosopic tests [16][17][18] . Here we use Random Forest regression 19 for extracting information regarding sample specific failure times from spatio-temporal records of energy release signals prior to failure.…”