Abstract
A new technique titled "Particle Refresh" is introduced to particle swarm optimization (PSO) and a hybridization with conjugate gradient method is also discussed in this paper. The particle refresh charges power to inactive particle to improve the recovery ability of PSO after trapping on a local solution, and as a result, it becomes easy to choose suitable values for control-parameters to keep high performance for diverse objective functions. On the other hand, the point of the hybridization is how to determine the changeover timing between a conjugate gradient method and a PSO algorithm.
