Research And Improvement Of Shuffled Frog Leaping Algorithm

With the scale expanding and the complexity increasing of practical engineering problems, lots of optimization problems are intractable. In the past few years, swarm intelligence algorithms with unique mechanism and excellent global search ability have attracted attention of scholars in various fields. Ant colony algorithm, artificial fish swarm algorithm, particle swarm optimization and shuffled frog leaping algorithm are common swarm intelligence algorithms. Shuffled frog leaping algorithm is a newly swarm intelligence algorithm. The algorithm has advantages of few parameters to set, simple and easy to implement, strong robustness, balancing global and local search and so on and is widely used in various areas of production.Firstly, The optimization principle of shuffled frog leaping algorithm is studied deeply and the theoretical basis, mathematical model, the advantages and disadvantages are also analyzed in this thesis. Secondly, In order to eliminate the shortcomings of original shuffled frog leaping algorithm that non-uniform initial population, randomness of step moving and low search efficiency, an improved algorithm is proposed. Chaos theory and opposition strategies are applied to construct the initial population, which not only make the initial population more evenly distributed in the solution space, but also improve the quality of solutions. An adaptive learning factor which can guide global and local search for solutions is designed to improve the convergence precision and avoid the premature convergence. Double center strategy that special center frog and general center frog is introduced. Take the advantages of the center frog to strengthen the information exchange among frogs and improve the efficiency of local search. Finally, several classic test functions are choosed and simulation experimental results show that the improved algorithm has better performance and convergence result.