Abstract: In this paper, two strategies are employed to overcome the shortcomings of the bird swarm algorithm (BSA):(1) in order to improving the search targeting of BSA vigilance behavior, the current optimal individual is adopted instead of random selection; (2) the weighted average step takes the place of the t-step in order to overcoming of sharp jump which originates from the iteration step-size of BSA flight behavior. Experimental results show that the proposed algorithm can improve both the global search ability and the optimization accuracy of BSA, and achieves a better convergence speed.