摘要
蚁群算法作为一种新的智能计算模式,由于其离散性本质而在组合优化问题上取得巨大成功,但这也限制了它在连续问题求解中的应用。为此,提出一种用于连续域寻优的改进蚁群算法。算法的局部搜索基于解决离散问题的经典蚁群优化思想,全局搜索利用AntWalk和AntDiffusion技术,且每代寻优结束后均采用"精英策略"把本代最优个体保留到下一代中。最后在理论上对其进行了收敛性分析,证明可较快地收敛到全局最优解,并用几个基准函数对算法做了仿真测试,均取得良好效果。
As a new model of intelligent computing, ant colony optimization （ACO） is a great success on combinatorial optimization problems, however, it is restricted to settle the problem of continuous domains because of its discrete nature. An improved ant colony optimization was proposed. In the local search, the improved ant colony approach is based on the idea of ACO that is used for discrete domains, but utilizes Ant Walk and Ant Diffusion operation in the global search, and while each generation accomplished, preserves the best individual to next generation by the idea of ＂Elitist Strategy＂. Then its convergence was analyzed theoretically, and was proved to converge to the optimization solution rapidly. This algorithm was tested by several benchmark functions, and could handle these optimization problems very well.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2008年第15期4021-4024,共4页
Journal of System Simulation
关键词
蚁群算法
连续域
遗传算法
收敛
ant colony optimization
continuous domains
genetic algorithm
convergence
