蛙跳算法（SFLA）是一种全新的启发式群体进化算法，具有高效的计算性能和优良的全局搜索能力。对混合蛙跳算法的基本原理进行了阐述，针对算法局部更新策略引起的更新操作前后个体空间位置变化较大，降低收敛速度这一问题，提出了一种基于阈值选择策略的改进蛙跳算法。通过不满足阈值条件的个体分量不予更新的策略，减小了个体空间差异，从而改善了算法的性能。数值实验证明了该改进算法的有效性，并对改进算法的阈值参数进行了率定。(Leapfrog algorithm (SFLA) is a new heuristic population evolutionary algorithm, has high computing performance and excellent global search ability. The basic principle of SFLA is discussed, and the update operation algorithm based on local updating strategy caused by the change of individual space greatly, reduce the speed of convergence of this problem, proposed an improved shuffled frog leaping algorithm based on the threshold selection strategy. By reducing the individual variance of the individual components without satisfying the threshold condition, the performance of the algorithm is improved. Numerical experiments demonstrate the effectiveness of the improved algorithm and determine the threshold parameters of the improved algorithm.)