基于拐点和区域划分的高维多目标进化算法  

A Many-objective Evolutionary Algorithm Based on Knee Point and Region Division

作　　者：杨景明[1] 郝佳佳 孙浩[1] 魏之慧 李霞霞 YANG Jing-ming;HAO Jia-jia;SUN Hao;WEI Zhi-hui;LI Xia-xia(School of Electrical Engineering,Yanshan University,Qinhuangdao,Hebei 066004,China)

机构地区：[1]燕山大学电气工程学院,河北秦皇岛066004

出　　处：《计量学报》2021年第8期1068-1075,共8页Acta Metrologica Sinica

基　　金：国家自然科学基金(61803327);河北省青年基金(E2018203162)。

摘　　要：针对高维多目标优化问题中收敛性和分布性难以同时保持的问题,提出一种基于拐点和区域划分的高维多目标进化算法KnSP。算法选取拐点作为第1次区域划分的中心点,自适应生成邻域;然后采用角度分层法进行二次区域划分,将点到超平面的距离作为个体选择的准则;最后通过候选解与其余个体的角度来增加或删除个体以保证种群规模。实验结果表明,算法在多个测试函数中性能表现较其它比较算法更优。A many-objective evolutionary algorithm KnSP is proposed based on knee point and region division to solve the problem that is difficult to maintain convergence and distribution.The algorithm selects the knee points as the center point of the first region division and adaptively generates a corresponding neighborhood.Then the angle is used to divide the second area,and the distance of the point to the hyperplane is used as the criterion for individual selection.Finally,from the perspective of the candidate solutions and the other individuals,individuals are added or deleted to ensure the population size.Experimental result shows that the algorithm performs better in some test functions than compared algorithms.

关 键 词：计量学 多目标优化问题 拐点 区域划分 进化算法 

分 类 号：TB973[一般工业技术—计量学]