A MACHINE LEARNING BASED TRANSFER TO PREDICT WARHEAD IN-FLIGHT BEHAVIOR FROM STATIC ARENA TEST DATA
Abstract
There are no standardized or widely accepted methodologies to predict a warhead’s fragment behavior in flight, when only static arena test data is available. That is, the proposed transfer function from static tests to dynamic predictions does not exist. Dynamic tests are also extremely costly and rare, if not non-existent. Most researchers add the terminal velocity to the fragments’ velocities of the static case. This may work for limited cases, but we expect it to be far away from reality when the dynamics inherent in natural fragmentation occurs (e.g. transient changes in aerodynamic drag, lift and moment are non-negligible, with the last two due to nonsymmetrical shapes of the fragments). This challenge is even more acute when hypersonic weapons are used, i.e. the terminal speed may exceed the fragments’ speeds in the static case. In the case of hypersonic warheads, we expect substantial differences compared to subsonic or supersonic, as the weapon is traveling faster than the products of detonation. We envision an entire spectrum of fragments’ flight conditions, ranging from quasi-static to high hypersonic, with impacts so far unknown in terms of lethality and collateral damage. For the above reasons we believe that a completely new approach must be investigated to predict in-flight fragment fly out behavior; machine learning based transfer function to predict warhead inflight behavior from static arena test data.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 12, 2021
- Source ID
- FA95502010200
Entities
People
- Riccardo Bevilacqua
Organizations
- Air Force Office of Scientific Research
- United States Air Force
- University of Florida