Digital Twin Machine Learning for Asymmetric Warhead Characterization

Accurate lethality and effectiveness predictions for ordnance systems are a critical objective of the U.S. Air Force. Characterization and prediction of fragment fields from new asymmetric warheads requires innovative testing, diagnostics, and fragment modeling solutions, since the fragment fields are asymmetric about their azimuth due to the weapon cross-section and fusing offsets. As in the past, arena testing diagnostics will only be able to cover some regions of the fragment field, but simple azimuth wedge measurements will no longer be able to represent the overall fragment field. Therefore, a new paradigm for testing and modeling is required for asymmetric weapons. Digital twin data from finite element models will be the cornerstone of a new paradigm for asymmetric weapon characterization. The ability to synthesize digital twin fragment fields with arena test measurements will be a transformative outcome of this effort. Protection Engineering Consultants (PEC) has teamed with Southwest Research Institute (SwRI) and Midland Research (MR) to develop a machine-learning (ML) debris model that will synthesize data from arena tests and digital twin models to characterize asymmetric weapons. The ML framework will consist of a deep learning data fusion technique to synthesize digital twin and arena test data, and a computational Bayesian method for estimating fragment field uncertainty.