| 基于自适应配对控制的多目标演化算法 |
| |
| 引用本文： | 张秀杰,李欣,张虎,赵杰.基于自适应配对控制的多目标演化算法[J].控制与决策,2018,33(3):392-402. |
| |
| 作者姓名： | 张秀杰 李欣 张虎 赵杰 |
| |
| 作者单位： | 哈尔滨工业大学基础与交叉科学研究院，哈尔滨150001,哈尔滨工业大学控制理论与指导技术研究中心，哈尔滨150001,北京机电工程研究所，北京100074,哈尔滨工业大学机器人技术与系统国家重点实验室，哈尔滨150001 |
| |
| 基金项目： | 国家自然科学基金项目(61333003)；中国航天科技集团公司航天科技创新基金项目. |
| |
| 摘 要： | 为了平衡搜索过程中的开采和勘探,设计一种聚类辅助的基于繁殖效用的自适应配对控制策略,进而提出一种基于自适应配对控制的多目标演化算法(ACEA).利用K-means聚类算法发掘种群分布结构,以配对控制概率限制从同一类邻居或者整个种群中挑选父个体繁殖新解,以加强局部搜索或者勘探.采用的配对控制概率根据不同繁殖机制在过去一定代数的繁殖效用,在每一代中自适应地更新.选取标准测试题以及5种代表性的多目标演化算法测试ACEA的性能,通过结果验证所提出算法的优越性.
|
| 关 键 词： | 多目标演化算法 K-means聚类算法 配对控制策略 自适应配对控制概率 |
Adaptive mating control based multiobjective evolutionary algorithm |
| |
| ZHANG Xiu-jie,LI Xin,ZHANG Hu and ZHAO Jie\.Adaptive mating control based multiobjective evolutionary algorithm[J].Control and Decision,2018,33(3):392-402. |
| |
| Authors: | ZHANG Xiu-jie LI Xin ZHANG Hu and ZHAO Jie\ |
| |
| Affiliation: | Academy of Fundamental and Interdisciplinary Sciences,Harbin Institute of Technology,Harbin 150001,China,Center for Control Theory and Guidance Technology,Harbin Institute of Technology,Harbin 150001,China,Beijing Electro-mechanical Engineering Institute, Beijing 100074,China and State Key Laboratory of Robotics and System,Harbin Institute of Technology,Harbin 150001,China |
| |
| Abstract: | In order to balance the exploitation and exploration during the search process, this paper designs a reproduction utility based adaptive mating control strategy with the assistance of K-means algorithm, and thus proposes an adaptive mating control based multiobjective evolutionary algorithm(ACEA). K-means algorithm is firstly applied to discover the population distribution structure. Then the mating control probability is used to restrict parents to be selected from the neighbors in the same cluster or from the whole population for reproduction to emphasize on exploitation or exploration respectively. The mating control probability is updated at each generation according to the reproduction utility by different reproduction mechanisms in previous generations. This paper adopts the standard test instances and five representative multiobjective evolutionary algorithms to test the performance of ACEA. The results verify the superiority of the proposed algorithm. |
| |
| Keywords: | |
|
| 点击此处可从《控制与决策》浏览原始摘要信息 |
|
点击此处可从《控制与决策》下载全文 |
|
