动态进化与交互学习机制融合的蚁群算法  

Ant Colony Algorithm Based on Dynamic Evolution and Interactive Learning Mechanism

作　　者：张海南 游晓明[1] 刘升[2] ZHANG Hainan;YOU Xiaoming;LIU Sheng(School of Electronic and Electrical Engineering,Shanghai University of Engineering Science,Shanghai 201620,China;School of Management,Shanghai University of Engineering Science,Shanghai 201620,China)

机构地区：[1]上海工程技术大学电子电气工程学院,上海201620 [2]上海工程技术大学管理学院,上海201620

出　　处：《信息与控制》2020年第3期297-305,共9页Information and Control

基　　金：国家自然科学基金资助项目(61673258,61075115,61403249,61603242)。

摘　　要：针对蚁群算法收敛速度慢,易陷入局部最优等问题,提出动态进化与交互学习机制融合的蚁群算法(DEILACO).结合小生境的思想在两个种群内部构建动态进化模型,采取自适应进化机制,依据种群内部子群的寻优状态,对子群进行分级淘汰,合理调整进化方向,避免子群陷入局部最优;同时双种群采用交互学习机制,通过衡量种群内各子群适应度的标准差,自适应地调整交互周期,降低种群间的通信开销,并采取学习对象差异配对交流策略,提高交流效率和求解精度.最后采用多组不同规模的TSP(traveling salesman problem)算例实验分析,并与其它多种群算法进行对比.结果表明,该算法在提高求解精度和寻优速度方面表现更优.To address the ant colony algorithm’s problems of slow convergence speed and tendency to easily fall into the local optimum,we propose an ant colony algorithm based on dynamic evolution and interactive learning mechanism.Based on the concept of niche,we construct a dynamic evolutionary model within the two populations and adopt an adaptive evolutionary mechanism.According to the optimization status of the subpopulations within the population,the algorithm eliminates the subpopulations by stages,and the evolutionary direction is adjusted reasonably to prevent the subpopulations from falling into local optimum.At the same time,the two populations adopt an interactive learning mechanism to measure the population.The standard deviation of the fitness of each subgroup can adjust the interaction period adaptively,reduce the communication overhead among the populations,and adopt the strategy of learning object difference pairing to improve the communication efficiency and solving accuracy.Finally,we use a number of TSP examples with different scales for experimental analysis and compared with other multi-population algorithms.Results show that the algorithm performs better in improving the solution accuracy and optimization speed.

关 键 词：动态进化模型 小生境 分级淘汰策略 交互学习机制 蚁群算法 

分 类 号：TPI8[自动化与计算机技术]