Hybridizing Particle Swarm and Brain Storm Optimizations for Applications in Electromagnetics

Abstract

This paper compares the performance of two global optimization algorithms: particle swarm optimization (PSO) and brain storm optimization (BSO). We observe that BSO has clear advantage in the speed of global exploration from a random initialization, while PSO outperformed in the accuracy of local exploitation with a predefined initialization. The possibility of hybridizing PSO and BSO is then investigated with an example of patch antenna circuit model determination. It is shown that the BSO-PSO hybrid algorithm could benefit from the advantages of both PSO and BSO, and therefore outperforms single optimization methods within the same number of iterations.