Abstract: Group Search Optimizer(GSO) is a new swarm intelligence algorithm, which has a superior performance on high dimensional and multi model problems. However, due to the lower diversity of the population, GSO algorithm easily falls into the local optimum in the late optimization. An improved GSO based on the framework of cultural algorithm is presented in this paper. Moreover, the colony fitness variance is introduced to decide whether to undergo the operation of influence function such that the convergence efficiency can be arisen. The comparison experimentations with GA, PSO and GSO are made, and the improved GSO algorithm is also applied to the optimization problem of the profit in the butane alkylation process. These results show the effectiveness of the improved algorithm.