融合聚类线性组合与优化状态自适应的差分进化算法OA

To address the issues of the DE algorithm,such as high parameter sensitivity,insufficient global exploration capa-bility,and imbalance between exploration and exploitation processes in high-dimensional complex function optimization,this paper proposed an improved algorithm named clustering linear combination and optimization state adaptive differential evolution(CLOSADE),which integrated a clustering linear combination approach with an optimization state adaptive mechanism.The research aimed to enhance the algorithm's robustness and convergence performance when handling complex optimization prob-lems.This method firstly designed a clustering strategy based on dual factors of fitness and distance to generate multiple clus-ters of linear combination vectors and introduced a dynamic distance threshold to enhance population diversity.Secondly,it constructed an indicator of optimization state(IOS)to quantify population distribution characteristics,driving the adaptive ad-justment of mutation strategies and control parameters.Experimental results demonstrate that,on the CEC2017 and CEC2022 benchmark test functions,CLOSADE significantly outperforms advanced algorithms such as JSO,NL-SHADE-DP,and S-SHADE-DP in terms of both convergence accuracy and speed.Particularly on high-dimensional hybrid and composite func-tions,CLOSADE exhibits remarkable advantages,with an average improvement of 22％in convergence accuracy and approxi-mately 40％in convergence speed.Further population diversity analysis reveals that the multi-subgroup structure formed through clustering effectively maintains parallel search capabilities in the solution space,while the optimization state indicator ensures a dynamic balance between exploration and exploitation behaviors at different evolutionary stages of the algorithm.