Automated Wire Antennas Design using Dynamic Dominance Evolutionary Algorithm

Abstract

Antenna design problems are complex multi-objective optimization problems. It is difficult to solve such problems using traditional methods. This paper creates a new model to solve antenna problems. In this new model, antennas' requirements are reflected not only in problems' constraints, but also in their objectives. It means feasible solutions are solutions that satisfy antennas' requirements. So if we find out a feasible solution, antenna design problems would have been solved. After that, with the evolution going on, feasible solution would be optimized, and global optimal solution would be found out. This paper uses a new algorithm, dynamic dominances evolutionary algorithm (DDEA), to solve antenna problems. DDEA has been tested by 22 benchmark problems. DDEA can consistently find the global optimum of all test problems with success rate 100%, while any other algorithm can't achieve such high success rate. This paper takes lower earth orbit ST5 antenna designed problem as a real world test problem. The experimental result shows: using this new model and DDEA, ST5 antenna design problem can be solved effectively. In addition the simulation result shows our evolved antennas are quite competitive with NASA's.