| 一种基于群体分布特征的自适应多目标粒子群优化算法 |
| |
| 引用本文： | 耿焕同,陈哲,陈正鹏,薛羽.一种基于群体分布特征的自适应多目标粒子群优化算法[J].控制与决策,2017,32(8):1386-1394. |
| |
| 作者姓名： | 耿焕同 陈哲 陈正鹏 薛羽 |
| |
| 作者单位： | 南京信息工程大学计算机与软件学院,南京210044,南京信息工程大学计算机与软件学院,南京210044,南京信息工程大学计算机与软件学院,南京210044,南京信息工程大学计算机与软件学院,南京210044 |
| |
| 基金项目： | 国家自然科学基金项目(61403206);江苏省自然科学基金项目(BK20151458);"青蓝工程"资助项目(2016). |
| |
| 摘 要： | 针对多目标优化问题求解,提出基于群体分布特征的多目标自适应粒子群优化算法(pdMOPSO).首先借助统计方法分析归档集在决策空间的分布特征,以此划分进化状态,指导全局引导粒子的选择;然后设计粒子重排策略,动态调控种群的分布;最后依据进化状态设计不同的归档集维护策略,实现归档集中分布性和收敛性的均衡.以ZDT、DTLZ和CEC09为测试集,与7种多目标优化算法对比,指标IGD、Spread和ER结果表明,所提出的算法在收敛性和分布性上均有显著优势.
|
| 关 键 词： | 粒子群 多目标优化 群体分布特征 多样性保持 自适应 |
A self-adaptive multi-objective particle swarm optimization algorithm based on swarm distribution characteristic |
| |
| CENG Huan-tong,CHEN Zhe,CHEN Zheng-peng and XUE Yu.A self-adaptive multi-objective particle swarm optimization algorithm based on swarm distribution characteristic[J].Control and Decision,2017,32(8):1386-1394. |
| |
| Authors: | CENG Huan-tong CHEN Zhe CHEN Zheng-peng and XUE Yu |
| |
| Affiliation: | School of Computer and Software,Nanjing University of Information Science & Technology, Nanjing 210044,China,School of Computer and Software,Nanjing University of Information Science & Technology, Nanjing 210044,China,School of Computer and Software,Nanjing University of Information Science & Technology, Nanjing 210044,China and School of Computer and Software,Nanjing University of Information Science & Technology, Nanjing 210044,China |
| |
| Abstract: | A self-adaptive multi-objective particle swarm optimization algorithm based on the swarm distribution characteristic is proposed. Firstly, the interquartile range(IQR) is applied to analyze the distribution features of archive in the decision space and classify the state of evolution, for guiding global leader selecting. Then, a swarm variation operator is proposed to rearrange particles dynamically. Finally, according to the evolution state, an archive maintaining strategy is designed to achieve a balance between distribution and convergence. The proposed algorithm is compared with 7 state-of-the-art multi-objective optimization algorithms on ZDT, DTLZ and CEC09 Benchmarks. The indicators of IGD, Spread and ER show that the proposed algorithm has certain advantages over other algorithms in terms of convergence and distribution. |
| |
| 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 |
|
