Recently, with the higher complexity of numerous projects optimization and solving problems, traditional optimization algorithm cannot meet the requirements anymore. Thus, the meta-heuristic algorithm surpasses itself in its high efficiency and intelligence than others, which made it develop faster. The shuffled frog leaping algorithm(SFLA) is a kind of algorithm that imitates the frogâ€™s foraging behavior. The SFLA has the advantages of simplified concept, advanced optimization locating ability, accurate solving capacity, and rapid convergent rate. However, for the solutions of some complicated problems, the SFLA has the disadvantages of local optimization lead to low solving accuracy, and later algorithm has low convergent rate.This article is based on the SFLA, aiming at improving the algorithm solving accuracy, convergent rate, and avoiding local optimization. The research presents an improved SFLA with the amelioration on population initialization, subpopulation division, and partial navigation. Population initialization is improved by opposite strategy based on disturbance factor, making the initial candidate solution fitness function value generally superior, accelerating the convergent rate. Meanwhile, importing the disturbance factor enriches the diversity of population, enhances the study ability of the candidate solution in the available zone. Subpopulation division is improved by fore and aft synchronization, narrowing the differences among subpopulations, deepening the communication among subpopulations, and enriching diversity of subpopulations. Partial navigation is improved by the design of combination of adaptive inertia factors and gradient information, accelerating the convergent rate and solving the disadvantage of local optimization efficiently.Through the multiple simulation experiments on typical testing functions, the improved SFLA in this article can obtain a higher globally optimal solution accuracy by a higher convergent rate. Meanwhile, the developed SFLA avoids the premature convergent phenomenon.In the following researches into the SFLA, researchers should further their step into its own researches and analysis first. Then, consider the importance of algorithm parameters to their results and start from the configuration of basic parameters, therefore improve the performance. Lastly, consider the combination of other algorithm and the SFLA, to future improve the performance of the algorithm and to solve the larger and more complicated real problems. |