摘要
模拟退火算法在有限计算量条件下的收敛性能对自身参数有很大的依赖性,这使得参数设定问题成了算法应用过程中的一个关键环节。考虑到模拟退火算法本身的随机性,将其参数设定问题描述为随机优化问题,提出一种系统可靠地解决该问题的混合优化算法,即基于序的巢分区算法,该算法继承了序优化算法的快速收敛性及巢分区算法的全局搜索特性,能够有效解决复杂的随机组合优化问题。以典型旅行商问题为算例的仿真结果检验了方法的高效性与可靠性。
With limited computing budgets, convergence rate of Simulated Annealing （SA） algorithm is dependent heavily on selected parameters, which makes parameter setting a pivotal step for application of SA. Based on existing literature on this problem, the parameter setting was proposed as a stochastic combinatorial optimization problem and a new kind of hybrid optimization algorithm, Order based Nested Partitions （NP） method, was developed to solve this class of problems. New method retains quick convergence of order optimization and global search capability of nested partitions method and can effectively solve difficult combinatorial optimization problems. Results of simulation with several typical traveling salesman problems （TSP） as example validate the affectivity and robustness of hybrid algorithm.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2008年第1期245-247,共3页
Journal of System Simulation
基金
国家自然科学基金资助项目(59885005)
关键词
模拟退火
参数设定
随机优化
序优化
巢分区
simulated annealing
parameter setting
stochastic combinatorial optimization
nested partitions
traveling salesman problem
