Addition of atmosphere turbulence in the Particle Swarm Optimization algorithm

In this work is proposed an enhancement forthe Particle Swarm Optimization (PSO)technique, introducing the concept of aturbulent atmosphere. The original algorithmmimics the behavior of a bird flock in flight,where each bird represents a candidatesolution for the problem and updates itsposition in the search space taking inconsideration the previous best find positionobtained by itself and by the flock. What isproposed in this work is the addition of anatmospheric turbulence that directly affects inan independent, random and irregular way, theflying pattern of each bird present in the flock.The turbulence that has been introducedassists the algorithm to escape from localminima, and subsequently increases thechances of finding a better solution and eventhe optimal. Tests was performed usingbenchmark optimization function (Griewank,Schwefel, Rosenbrock and Rastrigin), in orderto compare the original version of PSO andthis new one, therefore designed PSO-t. Theaddition of turbulence is effective, and thealgorithm reveal itself more robust concerningthe choice of tuning parameters.References [1] Boyd, R. and Richarson, P., Culture andthe Evolutionary Process, University odChicago Press, 1985. [2] Clerc, M. The swarm and the queen:Towards a deterministic and adaptiveparticle swarm optimization. Proceedingsof the IEEE Congress on EvolutionaryComputation (CEC 1999), pp. 1951-1957,1999. [3] Eberhart, R. C. and Kenney, J., A newoptimizer using swarm theory.Proceedings of the Sixth InternationalSymposium on Micromachine and HumanScience, Nagoya, Japan, pp. 39-43, 1995. [4] Eberhart, R. C. and Shi, Y., Particleswarm optimization: Developments,applications and resources. Proceedings ofthe IEEE Congress on EvolutionaryComputation (CEC 2001), Seoul, Korea,2001.