函数优化的小生境蝙蝠算法  

Niche Bat Algorithm for Function Optimization

作　　者：高珊[1] 马良[1] 张惠珍[1] 

机构地区：[1]上海理工大学管理学院,上海200093

出　　处：《数学的实践与认识》2014年第15期253-260,共8页Mathematics in Practice and Theory

基　　金：国家自然科学基金(70871081);上海市一流学科建设项目资助(S1201YLXK);上海市教育委员会科研创新项目(14YZ090);高等学校博士学科点专项科研基金联合资助课题(20123120120005)

摘　　要：针对蝙蝠算法易陷入局部最优解的缺点,利用小生境技术对蝙蝠算法进行了改进,提出一种小生境蝙蝠优化算法.算法基于小生境技术的适应度共享来分隔种群,引入了小生境排挤机制来保持种群多样性,在延续蝙蝠算法原有并行搜索等优势的基础上,提高了算法的金局搜索能力和局部收敛速度,具有可在不同邻域内发现多个解的特点.通过对一系列经典函数测试,并与已有算法进行比较,结果表明该算法在函数优化问题的求解中具有较高的计算效率和精度,以及较好的全局寻优能力.Since the basic bat algorithm（BA） has the disadvantage of trapping in local optima, niche bat Mgorithm（NBA） is presented in this paper. The fitness sharing of the niche techniques was used to separate populations, and the exclusion mechanism of the niche techniques was used to maintaining population diversity. The advantages of the basic bat algorithm such as parallel searching was maintained and global searching ability and local convergence rate of NBA was enhanced. Results of computational experiments on typical test functions of optimization and their comparisons with that of other algorithms show that this algorithm can effectively avoid local optima and has better performance for global optimality

关 键 词：小生境技术 蝙蝠算法 适应度共享 排挤因子 函数优化 局部搜索 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]