%0 Conference Proceedings %T Loss Analysis of a Reflectarray Cell Using ANNs with Accurate Magnitude Prediction %+ Institut d'Électronique et des Technologies du numéRique (IETR) %+ Thales Alenia Space Italia %+ Centre National d'Etudes Spatiales - Direction Des Lanceurs. (CNES) %A Richard, V. %A Loison, Renaud %A Gillard, R. %A Legay, H. %A Romier, M. %Z CNES %< avec comité de lecture %B 11th European Conference on Antennas and Propagation (EUCAP) %C Paris, France %P 10.23919/EuCAP.2017.7928256 %8 2017-03-19 %D 2017 %K Artificial Neural Networks (ANNs) %K magnitude prediction %K Reflectarrays %Z Computer Science [cs]/Networking and Internet Architecture [cs.NI] %Z Engineering Sciences [physics]/ElectronicsConference papers %X This paper proposes a design methodology to improve the Artificial Neural Networks (ANNs) modeling of reflectarray (RA) cells with regards to the prediction of reflection coefficients magnitude. It is applied to model both types of RA cells (capacitive and inductive) with 5 inputs parameters. The results demonstrate that the final ANNs models are reliable and accurate with an average error on the reflection coefficient magnitude of the scattering matrix (|R11|) of -66dB and -69dB respectively for the capacitive and inductive cells. This accurate prediction of magnitude allows rejecting a priori any cell with loss exceeding a prescribed threshold. Comparison of two canonical RA layouts shows the benefit that could be expected in a synthesis process. %G English %L hal-01622070 %U https://univ-rennes.hal.science/hal-01622070 %~ UNIV-NANTES %~ UNIV-RENNES1 %~ CNRS %~ INSA-RENNES %~ CNES %~ IETR %~ SUP_IETR %~ STATS-UR1 %~ CENTRALESUPELEC %~ UR1-HAL %~ UR1-MATH-STIC %~ UR1-UFR-ISTIC %~ IETR-BEAMS %~ IETR-ADH %~ TEST-UNIV-RENNES %~ TEST-UR-CSS %~ UNIV-RENNES %~ INSA-GROUPE %~ TEST-HALCNRS %~ UR1-MATH-NUM %~ IETR-SURFWAVE %~ NANTES-UNIVERSITE %~ UNIV-NANTES-AV2022