融合折射学习和改进天牛须搜索的黑猩猩优化算法及其应用  

Chimp Optimization Algorithm Based on Refraction-Learning and Improved Beetle Antennae Search and Its Application

作　　者：罗仕杭 何庆 LUO Shihang;HE Qing(Collage of Big Data and Information Engineering,Guizhou University,Guiyang Guizhou 550025,China)

机构地区：[1]贵州大学大数据与信息工程学院,贵州贵阳550025

出　　处：《传感技术学报》2022年第5期600-612,共13页Chinese Journal of Sensors and Actuators

基　　金：国家自然科学基金项目(62166006);贵州省科技计划项目重大专项项目(黔科合重大专项字[2018]3002);贵州省科学技术厅项目(黔科合基础-ZK[2021]一般335)。

摘　　要：针对黑猩猩优化算法(ChOA)寻优存在全局搜索能力弱、收敛精度低、易陷入局部最优等缺陷,提出一种融合折射学习和改进天牛须搜索的黑猩猩优化算法(BCRChOA)。首先,借鉴天牛须算法搜索能力强和Levy飞行机制搜索方向和步长的不确定性的特点,将Levy飞行改进的天牛须搜索算法对ChOA进行搜索优化,提高ChOA的全局搜索能力;其次,在“攻击者”个体位置更新阶段引入云自适应动态权值,以协调算法全局探索和局部开发能力;最后,采用基于折射定律的反向学习策略提高算法跳出局部最优的能力。实验选取10个基准测试函数、部分CEC2014测试函数以及工程优化案例,将BCRChOA与最新的元启发式算法及其改进算法进行跨文献对比,结果表明BCRChOA在寻优能力和鲁棒性上均显著优于原始算法和对比文献方法。Chimp optimization algorithm(ChOA)based on refraction-learning and improved beetle antennae search(BCRChOA)is proposed to overcome the drawbacks of ChOA,such as weak global search ability,low convergence accuracy,easily trapping into local optimum.Initially,it draws on the characteristics of the strong search ability of beetle antennae search algorithm and the uncertainty of the search direction and step length of the Levy flight mechanism,and optimizes the search algorithm of beetle antennae search algorithm of the Levy flight for ChOA to enhance the global search ability of ChOA.In addition,it introduces cloud adaptive dynamic weights in the“attacker”individual location update stage to coordinate the algorithm’s global exploration and local development capabilities.Finally,it adopts opposition-based learning strategy based on the law of refraction to improve the ability of ChOA to jump out of the local optimum.10 benchmark test functions,some CEC2014 test functions and one engineering optimization case are selected for the experiment to compare BCRChOA with the latest meta-heuristic algorithms and its improved algorithms in literatures.The results show that BCRChOA is significantly superior to the original and various improved algorithms in the comparative literatures in terms of optimization ability and robustness.

关 键 词：机械工程设计 黑猩猩优化算法 天牛须搜索 Levy飞行 云模型 折射学习 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP301[自动化与计算机技术—控制科学与工程]