东北大学学报（自然科学版）

东北大学学报:自然科学版 ›› 2014, Vol. 35 ›› Issue (4): 484-487.DOI: 10.12068/j.issn.1005-3026.2014.04.007

• 信息与控制 • 上一篇    下一篇

无参数变异的二进制差分进化算法

孔祥勇1，高立群1，欧阳海滨1，葛延峰1,2   

  1. (1. 东北大学 信息科学与工程学院， 辽宁 沈阳110819； 2. 辽宁省电力有限公司， 辽宁 沈阳110014)
  • 收稿日期:2013-07-17 修回日期:2013-07-17 出版日期:2014-04-15 发布日期:2013-11-22
  • 通讯作者: 孔祥勇
  • 作者简介:孔祥勇(1988-)，男，山东滕州人，东北大学博士研究生；高立群(1949-)，男，辽宁沈阳人，东北大学教授，博士生导师.
  • 基金资助:
    国家自然科学基金资助项目(61273155).

Binary Differential Evolution Algorithm Based on Parameterless Mutation Strategy

KONG Xiangyong1， GAO Liqun1， OUYANG Haibin1， GE Yanfeng1,2   

  1. 1. School of Information Science ＆ Engineering， Northeastern University， Shenyang 110819， China; 2. Liaoning Electric Power Company Limited， Shenyang 110014， China.
  • Received:2013-07-17 Revised:2013-07-17 Online:2014-04-15 Published:2013-11-22
  • Contact: KONG Xiangyong
  • About author:-
  • Supported by:
    -

RichHTML

0

PDF (PC)

921

摘要： 鉴于差分进化算法在解决复杂连续问题上的优良性能，针对0-1变量的特点，提出了一种用于求解0-1规划问题的二进制差分进化算法(BDEPM).与采用离散变换和逻辑运算的改进算法相比，BDEPM算法中的变量采用二进制编码方式，在进化过程中无需变异率，即可根据个体间的差异直接在离散域内进行变异，算法的思路清晰、结构简单、控制参数少、易于理解和实现.将BDEPM用于求解0-1背包问题，针对其约束提出了一种二次贪婪变换的修复策略，两个背包实例的仿真对比实验验证了BDEPM算法的优越性.

关键词: 无参数变异, 二进制编码, 差分进化, 0-1背包问题, 二次贪婪变换

Abstract: Owing to the superior performance on dealing with the complicated continuous problems， a binary differential evolution algorithm named BDEPM was proposed to solve 01 programming problem on account of characteristics of the 01 variables. Compared with discrete transform and logical operations， in the proposed method， the scaling factor was not required in the evolutionary process and binary mutation was directly carried out only on the basis of the differences between two individuals. The binary coded BDEPM algorithm was easy to understand， and it was implemented with a simple structure and less control parameters. A repair strategy based on quadratic greedy transform was put forward to handle the constraints when the BDEPM algorithm was applied to the solve 01 knapsack problems. Experimental results of two instances validated the superiority of the BDEPM algorithm.

Key words: parameterless mutation, binary code, differential evolution, 01 knapsack problem, quadratic greedy transform

中图分类号: