Rigorous analytical solutions to antenna synthesis problems have been obtained for several of their simplest designs, while numerical solutions to specific versions of these problems in a strict electrodynamic formulation for the required practical structures take so much time in most cases that obtaining an optimal solution in an acceptable time is not always possible. For this reason, approximate methods for solving various technical problems, including antenna problems, based on significantly simpler operators than in electrodynamics have become widespread. This allows for a significant gain in time and optimal solutions to synthesis problems in a reasonable time by direct, targeted enumeration of the space of combinations of parameters characterizing the problem under consideration. If it is necessary to increase the accuracy of the solution obtained in this way, approximate methods allow one to select from a possible number of options a significantly smaller number with the best values of the objective function for subsequent analysis of these options by strict methods. A significant place among approximate optimization methods is occupied by evolutionary algorithms (EA), which have been developed since the second half of the last century. They imitate the behavior and evolution of various biological communities, generally obeying the theory of evolution and Darwin's natural selection. The article provides a brief overview of modifications of both well-known algorithms and the control parameters they contain: the genetic algorithm (GA), the method of swarm of biological entities (PSO), the method of differential evolution (DE), with the help of which a number of problems on the synthesis of antennas with discretely and continuously changing parameters have been solved, and a relatively new algorithm – the method of grey wolves (GWO). An example of implementation using the PSO/FDTD optimization algorithm of a dual-frequency and broadband microstrip antenna with an E-shaped plate has been given. A mathematical model and a block diagram of the modified IGWO algorithm have been presented. The superiority of its characteristics in terms of the accuracy of determining the global extremum and the speed of convergence over such well-known algorithms as GWO, SCA, SSA, PSO, FPA, MFO and BA has been shown based on testing these algorithms on twenty-three reference functions.
Read full abstract