基于二次繁殖的自适应差分演化算法  

A Self-adaptive Differential Evolution Algorithm Based on Quadratic Propagation

作　　者：黄小城[1] 王希武[1] 许宗贤[2] 靳波[2] 米二强[1] 

机构地区：[1]军械工程学院计算机工程系,河北石家庄050003 [2]92510部队,广东湛江524096

出　　处：《军械工程学院学报》2010年第2期68-73,共6页Journal of Ordnance Engineering College

摘　　要：分析了差分演化算法的基本原理，针对基本差分演化算法对控制参数选择敏感性强、算法后期收敛速度较慢等问题，提出了基于二次繁殖的自适应差分演化算法，并通过实验对改进算法的性能进行测试。实验结果表明，改进算法的性能优于基本差分演化算法和自适应差分演化算法。The basic principle of differential evolution algorithm is analyzed. A self-adaptive differential evolution algorithm based on quadratic propagation is proposed, aiming at the sensitivity of the original differential evolution algorithm at the selection of control parameters and the slow constringency speed in the later evolution period of the algorithm. The capability of the modified algorithm is tested via experiment. The results of the experiment demonstrate that the modified algorithm is superior to the original differential evolution algorithm and the self-adaptive differential evolution algorithm.

关 键 词：差分演化 二次繁殖 自适应 动态更新 

分 类 号：TP311.5[自动化与计算机技术—计算机软件与理论]