For the complex multi-peaks function with high dimension, the particle swarm optimization and variable eighborhood search algorithm with convergence criterions(VNS-PSO-CC) is proposed on the basis of analyzing the problem f premature. This method combines the particle swarm optimization(PSO) with the global search ability of variable eighborhood search(VNS) algorithm, and adds the convergence criterions. Firstly, the preferable swarm is obtained by sing the fast searching ability of PSO algorithm. Furthermore, the premature swarm, which is estimated by convergence riterions, is optimized by using VNS algorithm. Finally, experimental results show that the performance of VNS-PSO-CC lgorithm is superior to the traditional PSO algorithm.