多策略协同改进的麻雀搜索算法  

Multi-strategy collaborative improved sparrow search algorithm

作　　者：李大海 曾能智 王振东 Li Dahai;Zeng Nengzhi;Wang Zhendong(School of Information Engineering,Jiangxi University of Science&Technology,Ganzhou Jiangxi 341000,China)

机构地区：[1]江西理工大学信息工程学院,江西赣州341000

出　　处：《计算机应用研究》2023年第11期3269-3275,共7页Application Research of Computers

基　　金：国家自然科学基金资助项目(61563019,615620237);江西理工大学校级基金资助项目(205200100013)。

摘　　要：针对麻雀搜索算法(SSA)种群多样性差、易陷入局部最优等问题,提出了一种多策略协同改进的麻雀搜索算法(ISSA)。首先,ISSA采用一种融合转移概率的边界学习策略的发现者位置更新方式,扩大发现者搜索范围并丰富其种群多样性;其次,ISSA在麻雀跟随者更新过程中引入混合粒子群机制,扩大目标跟随个体的选择范围;最后在算法寻优过程中,ISSA利用模糊推理系统动态监控种群陷入局部最优的概率,以差分变异操作提高种群跳出局部最优的能力。采用CEC2017测试函数中的12个函数作为性能基准函数,将ISSA与标准SSA及其他四种改进的麻雀搜索算法(ESSA、CSSOA、SSASC、MSSA)进行性能测试,基于实验数据的Friedman检验表明,ISSA能获取更好的性能。Aiming at overcoming drawbacks of sparrow search algorithm(SSA),such as poor population diversity and being easy to fall into local optimum,this paper proposed a multi-strategy collaborative improved sparrow search algorithm(ISSA).At first,ISSA adopted a boundary learning strategy incorporating transfer probability to update the location of the discoverer,expanding the search scope of the discoverer and enriching its population diversity.Secondly,ISSA introduced a hybrid particle swarm optimization mechanism in the sparrow follower update process to expand the selection range of target follower individuals.Finally,during the optimization process of the algorithm,ISSA used a fuzzy inference system to dynamically monitor the probability of the population falling into the local optimum,and used differential mutation operations to improve the ability of the population to jump out of the local optimum.This paper selected 12 test functions from CEC2017 testbed as benchmark to evaluate the performance of ISSA with standard SSA,and other 4 improved sparrow algorithms(ESSA,CSSOA,SSASC,and MSSA).Friedman test based on experimental data shows that ISSA can achieve better performance.

关 键 词：麻雀搜索算法 边界学习 转移概率 混合粒子群机制 模糊推理系统 

分 类 号：TP306.1[自动化与计算机技术—计算机系统结构]