无约束优化问题的多种群混合类电磁机制算法  

Hybrid Multi-population Electromagnetism-like Mechanism Algorithm for Unconstrained Optimization Problem

作　　者：梁存利[1] 董建民[2] 刘泽国[2] 

机构地区：[1]西藏民族学院教育学院,陕西咸阳712082 [2]西藏民族学院信息工程学院,陕西咸阳712082

出　　处：《青岛科技大学学报（自然科学版）》2014年第6期640-644,共5页Journal of Qingdao University of Science and Technology:Natural Science Edition

基　　金：国家民族事务委员会科研基金项目(12XZZ003)

摘　　要：在分析了标准类电磁机制算法不足之处的基础上,提出了一种处理无约束优化问题的新的类电磁机制算法。新算法用两个种群进化,从两个进化种群中选出优势个体互相学习,交换信息;提出了基于粒子电荷量和它们之间距离的自适应调节的新的受力计算公式。仿真试验结果表明,和已有算法相比,该算法能有效克服早熟收敛,具有收敛快、求解性能好的优点.On the basis of analyzing the disadvantages of standard electromagnetism-like mechanism algorithm,a new electromagnetism-like mechanism algorithm for unconstrained optimization problem is presented.In the new algorithm,two populations are used for global search.Excellent individuals selected from the populations learn each other and exchange information.An adaptive formula for forces between the charged particles is proposed.The results of simulation experiment show that the algorithm can avoid premature convergence more effectively than other existing algorithms.This algorithm has fast convergence and good performance.

关 键 词：类电磁机制算法 无约束优化 多种群 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]