A Hybrid Genetic Algorithm for Robot Automatic Tuning of Microwave Filters

Abstract

When using the extracted coupling matrix for tuning, the template matrix needs to be carefully generated to satisfy all the specifications, to accommodate all the parasitic couplings, and to be within the tuning range of each tuning screw. To obtain this coupling matrix with optimization, lots of local minima are encountered due to the complexity of the problem. A hybrid genetic algorithm is introduced to pull the optimization process out of the local minimum and finally converge to the global minimum. A good result is observed for a filter designated for 5G application.