关键词: 蝙蝠算法, 序贯二次规划（SQP）, 柯西变异, 佳点集, 早熟收敛, 寻优精度

Abstract: In view of the basic bat algorithm has a few problems in low optimization accuracy, slow convergence speed and high possibility of being trapped in local optimum and so on, a hybrid bat algorithm based on Sequential Quadratic Programming（SQP） is proposed. The uniform initial population is constructed by the method of good point set, which enhances the ergodic ability of the initial population. In order to avoid premature convergence, Cauchy mutation operation is used to ensure diversity. In the late iterations, the best individual is used by SQP local search to improve the local bat depth search capabilities, which can ensure that the individual can find the global optimal solution close to the global optimal value, and to accelerate the evolution of population. The experimental results show that the improved bat algorithm has better performance, good optimization accuracy and fast convergence speed.

Key words: bat algorithm, Sequential Quadratic Programming（SQP）, Cauchy mutation, good point set, premature convergence, optimization accuracy