Abstract: In order to solve the problems of the traditional fruit fly algorithm in the global optimization process, the step size of the traditional fruit fly algorithm is improved. In the process of optimization, the algorithm uses the difference between the best concentration value and the worst concentration value in the current iterative fruit fly group and the previous one. The difference between the best concentration of the Drosophila melanogaster group and the worst concentration value was compared, and the change rate of the concentration difference was obtained. According to the rate of change of concentration difference, the optimal step size of fruit flies should be changed dynamically, which can effectively balance global optimization capability and local exploration ability. The results show that the improved fruit fly algorithm is obviously superior to the other three kinds of fruit fly algorithms in the convergence rate and optimization accuracy.