基于相对距离和历史成功率机制的增强麻雀搜索算法  

Enhanced sparrow search algorithm by adopting mechanismbased on relative distance and historical success rate

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》2024年第6期1640-1648,共9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：针对麻雀搜索算法收敛精度低、易陷入局部最优等问题,提出了一种融合相对距离和历史成功率的增强麻雀搜索算法(RHSSA)。首先,RHSSA引入一种融合适应度值与相对距离的发现者选择方式,使选出的发现者既保持较高质量,又保持在搜索空间的分布广泛;其次,RHSSA在麻雀发现者搜索过程中,采用融合加权重心的反向学习策略,充分挖掘搜索空间的优质位置信息并减弱发现者向原点聚集的趋势;最后,RHSSA引入基于历史成功率的自适应选择算子动态地选择柯西变异与高斯变异对最优解做扰动,提高算法跳出局部最优的能力。选用CEC2017测试函数集中的12个函数作为性能基准函数,将RHSSA与其他五种改进的麻雀搜索算法(AMSSA、SCSSA、SHSSA、ISSA、CSSOA)进行性能评测。基于实验数据的Friedman检验表明,RHSSA能获取最优的结果。为验证提出的改进策略的有效性,还对改进策略进行了消融实验。实验结果表明在综合改进策略的共同作用下,RHSSA的综合优化性能排名为第一名。Aiming to overcome faults of lower convergence accuracy and susceptibility to local optima in sparrow search algorithm(SSA),this paper proposed an enhanced sparrow search algorithm by adopting the mechanism based on relative distance and historical success rate,namely RHSSA.Firstly,RHSSA introduced a discoverer selection method that integrated fitness values and relative distance to make selected discoverers maintaining high quality and wider distribution in search space.Secondly,RHSSA adopted a reverse learning strategy that integrated weighted center of gravity during each search iteration of discovers in order to fully mining the high-quality location information in the search space and weakening discoverers’trend to gather towards the origin.Finally,RHSSA also used an adaptive selection operator based on historical success rate to dynamically select between Cauchy and Gaussian mutations to disturb the optimal solution to improve the algorithm’s ability to jump out of local optimal.12 functions were selected from the CEC2017 test function suit as the benchmark to evaluate RHSSA with five other improved sparrow search algorithms(AMSSA,SCSSA,SHSSA,ISSA,and CSSOA).The result of Friedman test based on experimental data shows that RHSSA can achieve the supreme performance among all evaluated algorithms.To futher verify effectiveness of the proposed improvement strategies,ablation experiments were conducted.The result illustrates that under the combination of all proposed improvement strategies,RHSSA ranks first in comprehensive optimization performance.

关 键 词：麻雀搜索算法 适应度值与相对距离 加权重心 反向学习 自适应选择算子 

分 类 号：TP306.1[自动化与计算机技术—计算机系统结构]