一种混合改进的鹰栖息优化算法  

A hybrid improved eagle perching optimization algorithm

作　　者：胡洁[1] 王盛洁 张涛[1] HU Jie;WANG Shengjie;ZHANG Tao(School of Information and Mathematics,Yangtze University,Jingzhou 434023,Hubei)

机构地区：[1]长江大学信息与数学学院,湖北荆州434023

出　　处：《长江大学学报（自然科学版）》2022年第4期111-118,共8页Journal of Yangtze University(Natural Science Edition)

基　　金：国家自然科学基金项目“二层多目标规划问题的粒子群算法及在跨流域水库全联合调度中的应用”(61673006)。

摘　　要：鹰栖息优化(eagle perching optimization,EPO)算法模拟了鹰在大自然中栖息的生物特性,在全局范围内随机采样,利用目标函数找到采样点中的最优解,之后将搜索范围缩小,在这个最优解附近进行二次采样,迭代这一过程,执行全局搜索到局部搜索的转变。该算法原理简单、易于实现,是一种收敛速度较快的新型群智能算法,但在解决高维问题时算法收敛精度低、易陷入局部最优。基于自适应调优和混合算法的思路,提出了一种混合改进的鹰栖息优化(hybrid improved eagle perching optimization,HIEPO)算法:一方面引入成功率作为反馈参数自适应调整算法的收缩变量,改变了原有定值和线性递减设置,更好地实现全局搜索和局部搜索之间的转变;另一方面,结合粒子群(particle swarm optimization,PSO)算法收敛速度快,全局搜索能力强的优点,将引入成功率的EPO算法与PSO算法串行,提高收敛精度且避免了局部最优。单峰函数(f_(1)~f_(4))、多峰函数(f_(5)~f_(8))和定维多峰函数(f_(9)~f_(12))这12个标准测试函数求解得到的平均值、标准差以及拉伸/压缩弹簧设计和压力容器设计2个工程约束优化问题的求解结果表明,改进后的HIEPO算法在收敛精度和避免局部最优方面均有一定优势。The eagle perching optimization(EPO)algorithm simulates the biological characteristics of eagles perching in nature.The algorithm first samples randomly in the global scope and uses the objective function to find the optimal solution among the sampling points,and then narrows the search scope and performs a secondary sampling near this optimal solution.This process is iterated to perform the transition from global search to local search.EPO is a new swarm intelligence algorithm with fast convergence speed,which is simple in principle and easy to implement,but it has low convergence accuracy and is easy to fall into local optimum when solving high-dimensional problems.Based on the idea of adaptive tuning and hybrid algorithm,a hybrid improved eagle perching optimization(HIEPO)algorithm was proposed.On the one hand,the success rate was introduced as a feedback parameter to adjust the shrinkage variable of the algorithm adaptively,which changes the original fixed value and linear decreasing settings to better realize the transformation between global search and local search;on the other hand,combining with the advantages of fast convergence speed and strong global search ability of particle swarm optimization(PSO)algorithm,the eagle perching optimizationalgorithm based on the success rate was serialized with the PSO algorithm to improve the convergence accuracy andavoid local optimal solution.The mean value and standard deviation obtained by solving 12 standard test functions,namely unimodal function f_(1)~f_(4),multimodal function f_(5)~f_(8)and fixed-dimensional multimodal function f_(9)~f_(12),as well as the results of solving two constrained optimization problems in engineering of tension/compression springdesign and pressure vessel design,show that HIEPO algorithm has some advantages in convergence accuracy andavoidance of local optimum.

关 键 词：EPO算法 PSO算法 混合优化 自适应 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]