Abstract: A hybrid global optimization algorithm is proposed by the Population Migration Algorithm（PMA） and Simplex Algorithm（SA），and it solves the PMA’s problems of premature convergence and slow computation precision.Every point which is produced randomly by PMA is optimized by SA at first.Through a typical test function（Shaffer） to verify the performance of the improved algorithm and 10 types of particle swarm optimization algorithm for comparison，results show that：The algorithm can obtain relatively good solutions，the success rate of convergence is up to 100%.