Optimization of the cross-sectional shape of a ridge waveguide using the ellipsoid and the tabu search algorithms

Improved design procedure for double-ridged waveguides

“…6). For such a functions, the gradient-based techniques such as the QN method are known to exhibit faster convergence than the stochastic schemes such as the MGAs and the TB algorithms [8], which require more than 90 iterations for convergence. However, the MGAs are more advantageous to use for minimizing functions with multiple extrema, because, unlike the gradient-based methods, they seldom get trapped in local minima, which is not the case here.…”

“…Also, the scalar FEM formulation suffers from the existence of spurious solutions. By 1996 the finite difference method (FDM), as well as the ellipsoid (EA) and tabu search (TB) algorithms, were used for the design of ridged waveguides [8]. The presence of sharp 90°waveguide corners inside the waveguide necessitates a very fine discretization of the FD mesh.…”

A universal tabu search algorithm for global optimization of multimodal functions with continuous variables in electromagnetics

“…(3) the transient parameters, X ; , for the optimized multisection arc pole shoe machine is reduced; (4) the differences of the parameters SCR and THF under two different pole shoe geometry are also negligible.…”

“…(3) the strength of the multisection arc pole shoe is stronger than that of a traditional pole shoe; (4) the optimal value of the ratio of the length of the pole arc with an equal air gap length to the total length of the pole arc IS 0.5-0.6 i From what just analyzed above, the following conclusions The numerical results on the benchmark problem and practical engineering applications as well as the analysis demonstrate that (1)the present algorithm is more powerful and efficient than the tabu search algorithm so far used in global optimizations, and the number of iterations of the present method is only about 20% of that of SA algorithm, whilst the optimal results are slightly better than those of SA algorithm;…”

A self-learning simulated annealing algorithm for global optimizations of electromagnetic devices

“…Robust algorithms for local minimization are well developed and are readily available. For global optimization problems, the competitive performance of "general purpose" optimization algorithms concerning both the robustness and the speed is still a challenging topic, although several stochastic algorithms such as simulated annealing, genetic, and tabu search methods, all have been investigated and applied in the literatures [1]- [4]. Nevertheless, the convergence speeds of all these algorithms are still very low.…”