Improvement And Application Research Of Shuffled Frog Leaping Algorithm

Shuffled frog leaping algorithm, which simulated frog group-foraging behavior, is one of modern heuristic intelligent algorithms. Swarm intelligence optimization algorithm itself is a kind of parallel processing algorithms and certain performance of the SFLA should be by test and verify. After analyzing the optimization mechanism of the SFLA, it is found that SFLA can easily lead to obvious errors and local optimum. In order to overcome this problem, two modified shuffled frog leaping algorithms are proposed and an in-depth study on their performance and electric power engineering applications are presented in this paper. The specific contents are as follows:(1) The basic framework、functional principles and features of the SFLA is introduced and the influence of the parameters on the algorithm are discussed. Shuffled frog leaping algorithm is analyzed in three update operators on the group performance and the average results of optimization. The experimental results showed that the individual random operator was absolutely necessary to maintain higher diversity of population; the algorithm can not only maintain the higher diversity of population but also improve the running speed and the optimization precision without its global extreme learning operator.(2) Having absorbed the "alternative" principle of the genetic algorithm, the SFLA algorithm based on Cauchy mutation operator came up. By means of random perturbations strategy, it enhanced ability of the proposed algorithm for global optimization and extended searching spaces. Furthermore, the local escape capacity of the proposed SFLA is improved remarkably.(3) A modified shuffled frog leaping algorithm based on diversity of population feedback is proposed. A quantitative evaluation indicator of population diversity is presented. The feedback, which changes along with the measure of population diversity, guides the direction of evolution. The modified algorithm defines the two means operation, that is, the attraction and the exclusion. Under the guidance of the diversity controller, the population moves attractively or exclusively, thus reducing blindness in the frog foraging. Compared to the standard SFLA, the modified algorithm can avoid the disadvantage that diversity of the population decreases rapidly in the process of evolution. Therefore, it can effectively prevent local optimum, which due to the over-reliance on the current best individual in SFLA. By the simulation results of three benchmark functions, the results show that the improved algorithm has better global optimization performance.(4) Because the researches applying shuffled frog leaping algorithm to the field of electric power engineering are few. This paper put the shuffled frog leaping algorithm into use to work out economic load dispatch and static load model parameters identification. Comparative analysis and experiment results demonstrate that the proposed algorithm is effective.