Design of Microstrip Meander Line SWS with Fully Connected Auto-adaptive Neural Network

We report on the design and demonstration of a practical Fully Connected Auto-adaptive Deep Neural Network (FCADNN) for microstrip meander line (MML) slow-wave structures, which can be used to speed up the design process of MML-SWS with excellent performance. We train FCADNN to design MML-SWS using exact numerical simulation results in a form of supervised learning, which shows that the prediction error is lower than
$0.1\%$
. The simulation results show that the attenuation of the designed structure created by FCADNN is 1.44 dB/mm at 160GHz center frequency with the phase velocity and the pierce impedance is 0.2c and 16
$\Omega$
, respectively.