Optimization Process for Bending a Periodic Structure: Start Ahead with Neural Networks

Optimizing high-dimensional designs requires essential efforts; bending of a radiating device is one of such examples. Such appearance appears in various applications like medical, space, radar, etc. Consequently, strong numerical methods are becoming necessary. In the present work, an optimized bent periodic structure, multiple-input and multiple-output (MIMO) antenna designed through a neural network (NN) is presented. The main goal of this optimization is to focus on the curvature of the antenna to find the suitable results for S-parameters both in terms of reflection S 11 , and transmission S 21 coefficients, and also total active reflection coefficient (TARC). Through the proposed method, the antenna configuration loaded by meandered slot and surrounded by printed meander line of different periodicities is designed and optimized in three frequency bands within the 12.94 GHz to 25 GHz range, namely, 12.94 - 14.79 GHz, 16.09 - 17.25 GHz, and 19.17 - 25 GHz.