关键词: 果蝇优化算法, 早熟收敛, 佳点集, 适应性动态步长, 柯西变异

Abstract: The basic Fruit Fly Optimization Algorithm (FOA) has the shortcomings of being easy to fall into local optimal value, slow convergence and low convergence accuracy. Aiming at the problems, a Mutation Fruit Fly Optimization Algorithm Based on Adaptive Dynamic Step Size (MFOAADS) was proposed. Firstly, the selection of the initial position of the population was improved using the optimal point set method, which reduced the randomness of initial point selection and the probability of getting trapped in local optimal value. Secondly, the adaptive dynamic step size optimization strategy was adopted to improve the convergence rate and the accuracy of the solution. Finally, if the algorithm fell into premature convergence, the Cauchy mutation perturbation would be utilized in a certain probability for the sake of making them jump out of local optimum. The test results of the five classical functions showed that convergence accuracy and convergence speed of MFOAADS were obviously superior to the FOA when the number of iterations was fixed. And in the comparison experiments with FOA, the average number of iterations of MFOAADS decreased significantly with a success rate of more than 97% when target accuracy was fixed. The experimental results show that, compared with the basic FOA, the proposed algorithm significantly improves the accuracy, efficiency and reliability.

Key words: Fruit Fly Optimization Algorithm (FOA), premature convergence, good point set, adaptive dynamic step size, Cauchy mutation