融合聚集因子和正余弦搜索的阿基米德优化算法

Archimedes Optimization Algorithm Fusing Aggregation Factor and Sine and Cosine Search

针对阿基米德优化算法(AOA)收敛精度差、跳出局部最优能力弱的不足,提出一种融合聚合因子与正余弦搜索的改进阿基米德优化算法(YMAOA)。首先,引入Sobol序列初始化种群,增强种群多样性;其次,将密度因子重构为非线性递减趋势,同时设计非线性权值平衡算法在不同时期的探索能力和收敛速度;然后,设计基于聚集因子判断的随机反向学习策略,增强全局探索的寻优性能;同时在算法局部优化阶段融合正余弦搜索机制进行位置更新,协助算法跳离局部最优。将改进算法与标准AOA及其他同类算法在9个基准函数上进行对比实验,结果表明:YMAOA算法在寻优精度和收敛能力上有明显提升,对比同类改进AOA算法,YMAOA兼具收敛速度和跳出局部最优能力的优势,Wilcoxon秩和检验结果也证明YMAOA在搜索性能上具有显著性优势。

Aiming at the shortcomings of Archimedes optimization algorithm,which has poor convergence accuracy and weak ability to jump out of a local optimum,an improved Archimedes optimization algorithm YMAOA was proposed which combines aggregation factor and sine and cosine search.First,Sobol sequence was introduced to construct the initial population to enhance the diversity of population.Second,the density factor was reconstructed as a nonlinear decreasing trend,and the nonlinear weight was designed to balance the exploration ability and convergence speed of algorithm in different periods.Then,a random reverse learning strategy based on clustering factor judgment was designed to enhance the optimizing performance of global exploration.In the local optimization stage,the sine-cosine search was integrated to update the position,which helps the algorithm to jump away from a local optimum.The improved algorithm was compared with standard AOA and other same type of algorithms on nine benchmark functions.The results show that compared with similar improved AOA algorithms,YMAOA has advantages in both convergence speed and local optimal ability.Wilcoxon rank sum test results also prove YMAOA has significant advantages in search performance.