基于权值分配策略的聚类天牛群优化算法  

K-means clustering Beetle Swarm Optimization algorithm based on weight distribution strategy

作　　者：郭晓语 高鹰[1] 李宁[1] 周灿基 严基杰 GUO Xiaoyu;GAO Ying;LI Ning;ZHOU Canji;YAN Jijie(School of Computer Science and Cyber Engineering,Guangzhou University,Guangzhou 510006,China)

机构地区：[1]广州大学计算机科学与网络工程学院,广州510006

出　　处：《智能计算机与应用》2023年第2期6-14,共9页Intelligent Computer and Applications

基　　金：广东省研究生教育创新计划项目(2020JGXM084);大学生创新训练计划项目(202111078029,S202111078052)。

摘　　要：为改进天牛群优化算法在种群更新阶段存在的社会信息利用不足,及其在多峰函数中易陷入局部极值的情况,提出了一种基于权值分配策略的聚类天牛群优化算法。算法使用k均值聚类算法配合轮廓系数法,将天牛种群分为k个最佳聚类子群;分别选取各子群中适应度值最优的个体,并通过给定策略分配影响权值;最后使用多个最优个体共同决策的方式处理原算法中的社会学习部分,从而降低全局最优个体对种群位置更新的影响。实验选取了15个常用基准测试函数对所提算法进行仿真测试。实验结果表明,所提算法能够适应不同类型的优化问题,相较于天牛群优化算法及3个经典的智能优化算法拥有更好的寻优精度和稳定性。To improve the social information utilization rate of the Beetle Swarm Optimization(BSO)algorithm in the population regeneration stage and enhance the global optimization ability of the BSO algorithm in multimodal functions,a K-means clustering Beetle Swarm Optimization(KMBSO)algorithm based on weight distribution strategy is proposed.Firstly,this algorithm uses the k-means clustering algorithm and the silhouette coefficient method to divide the population into k-optimal clustering subgroups.Then the optimal individual in each subgroup is selected and their influence weight is allocated according to the corresponding fitness value.Finally,the social learning part of the original algorithm is optimized by a joint decision-making method of multi-optimal individuals to reduce the impact of the global optimal individual when the population location is updated.The proposed algorithm is simulated in 15 different benchmark functions and the experimental results show that it has better optimization accuracy and stability than the BSO algorithm and three classical intelligent optimization algorithms.

关 键 词：天牛群优化算法 轮廓系数法 K均值聚类 权值分配 社会学习 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]