摘要
通过模拟仿真实验,研究了不同冷却调度参数对模拟退火算法性能的影响。结果表明:在5种不同的冷却调度下,不同的降温参数对算法的收敛效果存在差异。其中,线性冷却的降温速率呈等速率直线下降,并不能很好地控制收敛速率;指数冷却的冷却速率对降温参数的变化敏感,参数越小,冷却速率越快;逆冷却降温参数过大或过小均不利于算法的收敛效果;对数冷却和逆线性冷却在最初的迭代过程中,冷却速率较快,随后冷却速率变得极为缓慢,算法的收敛性较差。
Through simulation experiments,the influence of different cooling scheduling parameters on the performance of the simulated annealing algorithm was studied.The results showed that:under the five different cooling schedules,different cooling parameters have different convergence effects on the algorithm.Among them,the cooling rate of linear cooling decreases sharply at an equal rate,and the convergence rate cannot be well controlled;The cooling rate of exponential cooling was sensitive to the change of cooling parameters,and the smaller the parameters,the faster the cooling rate;The inverse cooling parameters with too large or too small were not conducive to the convergence effect of the algorithm;Logarithmic cooling and inverse linear cooling had a fast cooling rate during the initial iteration,and then the cooling rate were becomed extremely slow,and the convergence of the algorithm was poor.
作者
甘秋云
李兢思
杨佳翰
GAN Qiu-yun;LI Jing-si;YANG Jia-han(School of Applied Science and Engineering,Fuzhou Institute of Technology,Fuzhou,350001,Fujian)
出处
《蚌埠学院学报》
2023年第5期47-52,共6页
Journal of Bengbu University
基金
福建省教育厅中青年教师教育科研项目(JAT210599)。
关键词
模拟退火算法
冷却调度
收敛性
组合优化
simulated annealing algorithm
cooling scheduling
convergence
combination optimization
作者简介
通讯联系人:甘秋云(1986-),女,福建宁德人,副教授,硕士。E-mail:472493023@qq.com。
