Abstract:

Firefly algorithm is one of the new swarm intelligence optimization algorithms inspired by the flashing behavior of fireflies and the phenomenon of bioluminescent communication.In the algorithm,a firefly is attracted to a brighter firefly and the brightest will move randomly,which causes slower convergence speed of the algorithm,oscillation on the optimum value in the later of iteration and poorer stability.So an improved firefly clustering algorithm is put forward based on optimal classcenter disturbance.The movement and randomization of the firefly are modified,and the values of step factors are compared in experiment.The simulation results on UCI benchmark data sets show that the algorithm is feasible;it can be efficiently used for clustering problems and eliminate the repeated oscillation while the stability and convergence are improved obviously.

Key words: clustering;firefly algorithm;swarm intelligence optimization algorithm