On the basis of simulated annealing

a particle swarm algorithm was put forward to enhances the ability of global searching.The algorithm combines with the roulette wheel selection operator

which provides a mechanism to restrain the predominating of super particles in early stage and can effectively avoid the premature problem.Variety experiments are conducted on several test functions taken from the literature.The simulation results show that the algorithm has better probability of finding global optimum and mean best value and has faster convergence and maintains the population diversity in the process of evolution.