RCS Predictions Using Geometric Deep Learning

Evaluation of Radar Cross Section (RCS) for different designs of electrically large objects like aircrafts can be very time consuming. We demonstrate how Geometric Deep Learning (GDL) could be used for fast RCS predictions. The approach learns for historical simulation data the relation between mesh geometry and the RCS results. As soon as sufficient simulation data are available for reasonable GDL prediction models, the RCS can be evaluated for new designs in real time.