A physics-informed machine learning model for go/no-go criteria on reactive metamaterials.

We present a physics-informed machine learning framework for predicting Go/No-Go criteria on reactive metamaterials. The effectiveness of this framework was demonstrated by analyzing shock propagation through a one-dimensional laminate structure. The laminate material is composed of an HMX bed with equally distributed 2mm thick copper pillars. The Wide-Ranging equation of state (EOS) was used to model HMX while the Romenski EOS is used for an elastic regime of copper and perfect plasticity is assumed. A gauge was placed at the entry of the first copper pillar and at the exit of the last pillar and an Aluminum impactor was used to initiate the shock. A modified machine learning model was then developed to predict the criteria for the laminate structure.The proposed model uses only short time measurements for predicting this behavior expecting in large reductions of computational cost in higher dimension analysis. This framework can suggest a data-driven guideline for the design of optimal laminate structures (e.g number of copper pilers, thickness and distribution).