The particle swarm optimization algorithm has a few disadvantages in solving complex functions

including low solving precisions and high possibilities of being trapped in local optimum.According to genetics good parent often produce hybrid good offspring

an improved program is proposed

as the personal best position of the particle contains more useful information than the current position

in each iteration

crossover is operated to the best personal particle

a better position may be got

and when the particle swarm trap premature convergence

without destroying the existing population structure by using new bacterial feeding chemotaxis and all particles gradually get rid of the shackles of local optimum.Numerical examples show the effectiveness of the proposed algorithm.