摘要
传统演化算法在求解复杂多峰函数优化问题时经常会出现早熟、收敛速度慢等问题,特别是对于有多个最优解的函数,往往只能找到个别的最优解。针对这些问题,提出了一种基于隔离机制与排挤机制相结合的小生境演化算法。利用隔离机制增强引导进化能力,利用排挤机制保证种群的多样性,同时,采用反序交叉算子进一步加强局部寻优能力。实验表明,使用该改进小生境演化算法求解函数优化问题能更有效地克服传统演化算法存在的收敛速度慢和容易陷入局部最优解等缺点。
When the traditional evolutionary algorithm was used to solve the optimization problems of multimodal functions,it was not easy to avoid the problems of premature and slow convergent speed,especially when it faced the functions those had many global optimal solutions,fewer of the global optimal solutions could be found.To be aimed at these problems,an improved niche evolutionary algorithm based on the combination of the isolation mechanism and the crowding mechanism was proposed.In this new algorithm,the isolation mechanism was used to guide the evolution direction,and the crowding mechanism was introduced to keep the diversity of population.Besides,a niche inver-over operator was adopted to further enhance the local optimization ability.The results show that the shortcomings of slow convergent speed and easy to fall to local optimization solutions could be overcome by using this improved niche evolutionary algorithm to solve multimodal function optimizations problems,compared with the traditional evolutionary algorithm.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2013年第6期1170-1175,共6页
Journal of System Simulation
基金
国家自然科学基金(70971043
60873114)
广东省科技攻关项目(2012A020602037)
江西省教育厅科学技术研究项目(GJJ12348
GJJ12368)
关键词
演化算法
函数优化
隔离机制
小生境
evolutionary algorithm
function optimization
isolation mechanism
niche.
作者简介
李康顺（通信作者）（1962-），男，江西兴国人，教授，博士后，博导，研究方向为演化计算、图像视觉、演化硬件；
余锡伦（1987-），女，江西九江人，硕士生，研究方向为演化计算；
张文生（1966-），男，河南郑州人，研究员，博导，研究方向为机器学习理论与算法；
董文永（1973-），男，教授，博导，研究方向为演化计算。
