电子学报 ›› 2021, Vol. 49 ›› Issue (4): 647-660.DOI: 10.12263/DZXB.20200171

• 学术论文 • 上一篇    下一篇

非线性多目标概率优化问题的自适应采样免疫优化算法

张仁崇1, 潘春燕2, 武星3, 杨坤1   

  1. 1. 贵州商学院计算机与信息工程学院, 贵州贵阳 550014;
    2. 黔南民族师范学院数学与统计学院, 贵州都匀 558000;
    3. 贵州商学院管理学院, 贵州贵阳 550014
  • 收稿日期:2020-02-14 修回日期:2020-07-19 出版日期:2021-04-25 发布日期:2021-04-25
  • 通讯作者: 张仁崇
  • 作者简介:潘春燕 女,1990年10月出生,贵州三都人.2017年在贵州民族大学获理学硕士学位.现为黔南民族师范学院数学与统计学院讲师,主要研究方向为数据挖掘.E-mail:chunyanpan666@163.com;武星 男,1988年7月出生,安徽来安人.2014年在贵州大学获工程硕士学位.现为贵州商学院管理学院讲师,主要研究方向为物流与供应链管理.E-mail:wuxingjobmail@163.com;杨坤 男,1987年11月出生,贵州织金人.2017年在贵州民族大学获理学硕士学位.现为贵州商学院计算机与信息工程学院讲师,主要研究方向为模式识别与图像处理.E-mail:kunflv@126.com
  • 基金资助:
    贵州省科技计划(No.QKHJC[2020]1Y423);贵州省教育厅青年科技人才成长(No.QJH KY Zi[2018]276,No.QJH KY Zi[2018]429,No.QJH KY Zi[2018]275,No.QJH KY Zi[2018]269);贵州省教育厅人文社科(No.2018qn42)

Adaptive Sampling Immune Optimization Algorithm for Nonlinear Multi-Objective Probabilistic Optimization Problems

ZHANG Ren-chong1, PAN Chun-yan2, WU Xing3, YANG Kun1   

  1. 1. Computer and Information Engineering College, Guizhou University of Commerce, Guiyang, Guizhou 550014, China;
    2. School of Mathematics and Statistics, Qiannan Normal University for Nationalities, Duyun, Guizhou 558000, China;
    3. College of Management, Guizhou University of Commerce, Guiyang, Guizhou 550014, China
  • Received:2020-02-14 Revised:2020-07-19 Online:2021-04-25 Published:2021-04-25

摘要: 针对噪声环境下非线性多目标概率优化问题,提出问题求解的自适应采样免疫优化算法.在算法设计中,基于克隆选择原理,设计小种群的算法进化框架;提出目标值估计法自适应确定个体的样本大小并估算其目标值;借助传统快速非支配排序法,种群被分割成多级非支配子群协同进化;设计动态交叉分布指数的模拟二进制交叉加强各子群之间的信息交流;设计动态变异分布指数的多项式变异、均匀变异,以及自适应变异概率平衡全局、局部探索.最后,借助3个理论测试问题、海铁联运能耗优化问题以及9个代表性的比较算法,数值实验结果表明,此算法寻优效率优势显著、搜索效果优越、稳定性好.

关键词: 多目标概率优化, 免疫优化, 随机模拟, 自适应采样, 海铁联运

Abstract: In this paper,an adaptive sampling immune optimization algorithm is proposed to solve the nonlinear multi-objective probabilistic optimization problem in noisy environments.In the whole design of the algorithm,we develop an evolutionary framework with small population inspired by the clonal selection principle.The approach for estimating objective value is designed by adaptively determining the sample size of an individual.The population is divided into multilevel non-dominated sub-populations for co-evolution by the traditional fast non-dominated sorting approach.The simulation binary crossover with dynamic crossover distribution index is designed to enhance the information exchanges among all sub-populations.Polynomial mutation and uniform mutation with dynamic mutation distribution index,and adaptive mutation probability are designed to enhance the capability of global and local exploration.Finally,based on three theoretical test questions,energy consumption optimization of sea-rail intermodal transportation and nine representative comparison algorithms,the numerical experiment results show that the algorithm has significant search efficiency,superior search effect and good stability.

Key words: multi-objective probabilistic optimization, immune optimization, stochastic simulation, adaptive sampling, sea-rail intermodal transportation

中图分类号: