基于多层次信息反馈的混合蛙跳算法

The Shuffled Frog Leaping Algorithm Bases on the Interation of Multi-Level Information

摘要: 基于多层次信息反馈的混合蛙跳算法(Shuffled Frog Leaping Algoriyhm based on the interation of Multi-level information, MSFLA)，吸收遗传算法的交叉算子及粒子群算法(PSO)的粒子进化方式，将整个优化过程划分为标准混合蛙跳优化层、青蛙进化与学习层、外部档案信息交换层。混合蛙跳优化层保证青蛙进行正常的局部搜索优化(蛙跳算法)；青蛙进化与学习层保证青蛙每次迭代结束时都能得到更好的自身位置(PSO粒子进化方式)；外部档案信息交换层可以保证青蛙种群获得最优解(交叉算子)。通过各层次之间的信息交流，提高算法的性能。从实验结果对比能够得出，改进后的MSFLA算法可以有效地改善早熟收敛问题，具有更好的收敛速度和更高的寻优精度。

Abstract: The Shuffled Frog Leaping Algorithm bases on the interation of multi-level information (MSFLA), which assimilates the Crossover of the Genetic Algorithm and the evolutional pattern of Particle Swarm Optimization (PSO). The whole optimization process includes three aspects—the optimizing layer of standardly blended frog leaping, the layer of flog evolution and learning, and the commutative layer of external file information. The first layer guarantees the frog to normally perform local optimal search (Fog Leaping Algorithm) and the second layer ensures that the frog can get a better self-position at the end of each iteration (the evolutional pattern of PSO). Moreover, the last layer can assure that the fog population gains the most optimal solution (the Crossover).Through the exchange of information between all levels, improve the performance of the algorithm. Experimental results show that the improved MSFLA can avoids premature convergence effectively, and has better convergence results, higher accuracy.