Open Access
Subscription or Fee Access
Statistical Machine Learning and Sampling for Composite Fabrication and Performance
Abstract
We apply manifold learning and sampling to the tasks of fabrication, manufacturing, and testing of composites. We specifically address the challenge associated with statistical inference on these tasks from a small size sample. Limitations on the sample size could emanate from constraints on computational resources as well as constraints on physical experiments. In either case, the analyst is typically presented with a short table that contains observations of environmental conditions and quantities of interest (QoI). In the case of numerical simulations, the QoIs can be at the discretion of the analyst while in a laboratory setting these are typically limited by access to sensing devices. We augment the statistical knowledge captured by the available dataset with knowledge of physics constraints (eg conservation laws) in order to enhance the predictive value of the dataset. Imposing these constraints typically requires additional experiments (either physical or numerical). We proceed differently as we discover, within the dataset, an intrinsic structure that is consistent with the manner in which the available data is interrelated. To that end, we rely on diffusion maps, a recent data-analytics procedure. This allows us to rapidly characterize feasible domains for complex phenomena involving multiscale and Multiphysics interactions. We augment the diffusion map procedure with a stochastic sampler guaranteed to sample on the manifold, thus allowing us to impute a very large sample that is consistent with the statistics of the original dataset and its learned intrinsic features.
DOI
10.12783/asc33/26008
10.12783/asc33/26008
Full Text:
PDFRefbacks
- There are currently no refbacks.