求解约束优化问题的融合粒子群的教与学算法  

Hybrid teaching-learning-based fused with particle swarm optimization of constrained optimization problem

作　　者：赵乃刚[1] 李勇[1] 王振荣 Zhao Naigang;Li Yong;Wang Zhenrong(College of Mathematics&Computer Science,Shanxi Datong University,Datong Shanxi 037009,China;Information Center of Datong People’s Government of Shanxi Province,Datong Shanxi 037009,China)

机构地区：[1]山西大同大学数学与计算机科学学院,山西大同037009 [2]山西省大同市人民政府信息化中心,山西大同037009

出　　处：《计算机应用研究》2018年第5期1307-1309,共3页Application Research of Computers

基　　金：国家自然科学基金资助项目(61672331);山西省高等学校教学改革创新项目(J2017093);山西大同大学科学研究项目(2016K1)

摘　　要：针对约束优化问题,提出了一种融合粒子群的教与学算法。算法采用了一种自适应的教学因子,使得算法的搜索性能可以自适应地调整。引入了自我学习和相互学习的学习模式,使得信息交流更加多样化,增强了算法的全局搜索能力;最后根据适应度值将整个种群分为两个子种群,对适应度值差的子种群采用粒子群算法以提升收敛性能,对适应度值优的子种群采用教与学优化算法以增强种群的多样性,通过两种算法的优势互补,提升了算法的整体优化性能。通过在22个标准测试函数的实验和与其他三种算法的比较表明,融合粒子群的教与学算法求解精度高,收敛速度快,它是一种可行、高效的优化算法。Aiming at the constrained optimization problem,this paper developed a hybrid teaching-learning-based fuse with particle swarm optimization.It adopted an adaptively teaching factors,and it could adjust the search performance of algorithm adaptively.The introduction of the learning mode of self-study and learn from each other made information communication was more diverse and it enhanced the global search ability of the algorithm.Finally,the algorithm divided the whole population into two subpopulations according to the fitness value.The poor particles adopted particle swarm algorithm to improved the convergence of the algorithm,and the better particles adopted teaching-learning-based algorithm to increase the diversity of population.The complementary advantages of both algorithms improved the performance of algorithm.This paper compared the new algorithm with the other three kinds of algorithms on twenty-two test functions,the simulation experiments show that the new algorithm has better performance.It is a feasible and efficient optimization algorithm.

关 键 词：教与学算法 粒子群算法 约束优化问题 自适应 约束处理 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]