基于平均值的混合蛙跳算法

摘要

针对基本混合蛙跳算法收敛速度慢，容易陷入局部最优的问题，提出了基于平均值的混合蛙跳算法。该算法将基本蛙跳算法中子群的平均值，通过2种不同的更新策略分别引用到混合蛙跳算法的局部搜索中，对算法的更新策略进行了适当改进，以期提高混合蛙跳算法的局部搜索能力。结果表明：更新策略1将子群的平均值与局部更新策略相结合，使算法在搜索过程中加快搜索速度，提高了局部搜索能力；更新策略2则通过采用自适应概率随机将子群的平均值取代子群部分最优个体进行策略更新，使算法在局部搜索时提高了寻优能力，有效的避免算法陷入局部最优。通过对5个测试函数进行优化，并同基本混合蛙跳算法和文献中改进的算法进行比较，结果表明：该算法可以有效的避免局部搜索过早收敛，具有较好的优化性能。%Aiming at slow convergence speed and falling into local optimum problems of shuffled frog leaping algorithm easily,the novel shuffled frog leaping algorithm based on average value is proposed.The algorithm references average value using two kinds of different update ideas to the basic shuffled frog lea-ping algorithm and improves the update policy of algorithm appropriately and the local search ability re-spectively.The former combines average value of subgroup with partial update strategy,speeding up the convergence rate in the iteration and improving the local search ability,the latter uses adaptive probability randomly to replace some best individual of partial subgroups by using the average value of subgroups,and increases the local search optimization ability of algorithm,and effectively avoids the algorithm falling into local optimum.The algorithm is based on five test function optimization and compares with basic SFLA and the improved SFLA in related references;simulation experiments show that the algorithm based on average value can effectively avoid premature convergence and have better optimization performance.