AI-Driven Electromagnetic Modeling of Cable Harnesses: From RLGC Matrix Prediction to Crosstalk Estimation
1 : DEMR, ONERA, Université de Toulouse [Toulouse]
-
Site web
ONERA, Communauté d'universités et établissements de Toulouse
31000 Toulouse -
France
2 : DTIS, ONERA, Université de Toulouse [Toulouse]
-
Site web
* : Auteur correspondant
ONERA, Communauté d'universités et établissements de Toulouse
31000 Toulouse -
France
We propose a physics-aware graph neural network to accelerate the electromagnetic modeling of complex cable harnesses in aeronautical systems. By representing wiring bundles as graphs and embedding relevant physical constraints, the framework is intended to efficiently predict coupling-related per-unit-length parameters and to generalize to previously unseen, high-dimensional configurations.

