Abstract: Shuffled Frog Leaping Algorithm（SFLA） is a new group evolutionary algorithm prompted by the natrural biological phenomena, and it has fast calculation speed and strong search capability. But its local search ability is weak and it is easily caught in prematrue convergence. Combining with the advantages of Cauchy mutation and Gaussian mutation, a modified SFLA（MSFLA） is proposed to overcome the shortcoming. The MSFLA’s convergence speed is enhanced and the pheonomena that SFLA is trapped in local optimal solution will be avoided to a certain extent, so its ability of problem sloving for complex functions is improved. And experimental results prove the validity of the new SFLA.

Key words: Shuffled Frog Leaping Algorithm, premature convergence, Gaussian mutation, Cauchy mutation, optimization

摘要： 蛙跳算法是一种受自然界生物现象启发产生的群体进化算法，计算速度快，寻优能力强，但局部搜索能力较弱，容易陷入早熟收敛。针对其缺点，结合高斯变异和柯西变异的优点，提出了一种改进的混合蛙跳算法。改进后的算法收敛速度加快，在一定程度上避免陷入局部最优，提高了蛙跳算法解决复杂函数问题的能力。实验验证了其有效性。

关键词: 混合蛙跳算法, 早熟收敛, 高斯变异, 柯西变异, 优化