基于SQP局部搜索的蝙蝠优化算法

针对基本蝙蝠算法存在寻优精度不高，后期收敛速度较慢和易陷入局部最优等问题，提出一种基于序贯二次规划（Sequential Quadratic Programming，SQP）的蝙蝠优化算法。该算法应用佳点集理论构造初始种群，增强了初始种群的遍历性;为避免算法陷入早熟收敛，引入柯西变异算子对种群中精英个体进行变异操作，增加种群多样性;在迭代后期，对最优个体进行SQP局部搜索，提高蝙蝠算法的局部深度搜索能力，保证个体在靠近全局最优值时能够寻优到全局最优解，加快种群进化速度。通过仿真实验结果证明，改进后的蝙蝠算法性能优越，具有良好的寻优精度和收敛速度。

Hybrid bat algorithm based on sequential quadratic programming local search

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.