Surrogate Model for High-Fidelity Physics-Based Post Intercept Debris Generation

This topic seeks the development of fast-running, computationally efficient, surrogate models to be used in place of slow running, computationally expensive, high-fidelity physics-based models inside of end-to-end missile defense simulation frameworks. Recent advances in physics-informed machine learning (PIML) have made possible the substantial speed improvement of physics-based models. During the Phase I effort, Corvid used their existing PIML-based fast-running penetration model to develop a prototype state-of-the-art fast-running surrogate model for the evaluation of target/interceptor erosion and the resulting generation of post-intercept debris (PID). This prototype capability includes both the VIPER lethality tool and the Advanced VIDEO PID model. This combined toolset has the capability to generate predictions of the resulting interceptor/target erosion along with limited debris states (mass, material, velocity, and temperature) that approximate those produced using first-principles hydro-structural physics solvers while running on a many orders of magnitude smaller time scale. In this proposed Phase II effort Corvid will further enhance the VIPER lethality tool and complete the set of debris outputs generated by the prototype Advanced VIDEO tool to match the discrete and micro-debris outputs currently produced by Corvid for Velodyne simulations. Corvid also proposes to refine the neural network training and debris generation techniques used by the model to improve the accuracy and realism of both the lethality result and debris produced. Corvid will focus development during the Phase II program on the Next Generation Interceptor (NGI) design and application to relevant NGI threats (Reentry Vehicles and Non-Lethal Objects). Finally, Corvid will apply the improved debris modeling tool to an archival MDA flight test of interest to demonstrate the utility and accuracy of the debris model. Approved for Public Release | 25-MDA-12052 (27 Mar 25)