自适应变异的量子花授粉算法  

Adaptive Mutation Quantum-behaved Flower Pollination Algorithm

作　　者：陆克中[1,2] 章哲庆 刘利斌[3] LU Ke-zhong;ZHANG Zhe-qing;LIU Li-bin(School of Mathematics and Computer Science,Chizhou University,Chizhou 247100,China;College of Computer Science and Technology,Nanjing University of Aeronautics and Astronautics,Nanjing 211106,China;School of Mathematics and Statistics,Nanning Normal University,Nanning 530023,China)

机构地区：[1]池州学院数学与计算机学院,安徽池州247100 [2]南京航空航天大学计算机科学与技术学院,江苏南京211106 [3]南宁师范大学数学与统计学院,广西南宁530023

出　　处：《控制工程》2020年第4期683-691,共9页Control Engineering of China

基　　金：国家自然科学基金(11301044);安徽省自然科学研究重点项目(KJ2016A514)。

摘　　要：花授粉算法是计算智能领域的一个新方法,但该算法也存在收敛精度较低、收敛速度较慢等问题。针对此类问题,提出了一种自适应变异的量子花授粉算法。该算法首先将量子搜索机制引入花授粉算法,利用量子的随机性,提升算法的全局搜索能力;然后给出基于前后两次群体平均位置标准差的群体多样性评判准则,并在此基础之上,对失活个体进行变异操作,改善群体多样性的同时,引导群体向最优解方向搜索;最后在10个标准测试函数上进行实验,结果表明,自适应变异的量子花授粉算法具有较快的收敛速度与良好的收敛精度,优于基本花授粉算法及其他群体智能算法,是解决复杂函数优化问题的有效方法。Flower pollination algorithm(FPA)is a recently-developed method in the field of computational intelligence.There are a few disadvantages in FPA,including low solving precision and slow convergence speed.To overcome these disadvantages,a novel adaptive mutation quantum-behaved flower pollination algorithm(AMQFPA)is proposed.Firstly,a quantum search mechanism is introduced into FPA to enhance the global search capability.Secondly,the rule of diversity is defined by comparing the standard deviation of the average position of the following two generation populations,and a mutation policy is presented to improve the diversity of population and guide the search to the optimal solution.Finally,the simulation experiments on ten benchmark functions show that AMQFPA is the best in the convergence speed and the quality of solutions compared with other algorithms.AMQFPA is a good general algorithm for complex function optimization problem.

关 键 词：花授粉算法 量子花授粉算法 自适应变异 多样性控制 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]