多目标演化优化算法的决策空间多样性维护机制研究

发布时间：2022-09-27 12:37

　　多模态多目标优化问题在学术研究和工业应用中广泛存在,比如航天发射任务设计、汽车发动机设计等等。近年来,多模态多目标优化受到越来越多的关注,许多学者对此进行了研究。在多模态多目标优化问题中,优化目标是找到所有具有相同目标向量但在决策空间中分布不同的帕累托最优解。本论文研究了几种具有代表性的多目标演化优化算法在多模态多目标优化问题上的表现。实验结果表明,随着算法的执行,由于没有解的多样性的保护机制,决策空间中解的多样性会变得越来越差。为了解决该问题,本论文提出了两种决策空间中解的多样性的维护机制。本论文的主要工作包括:（1）提出了一种子种群搜索方法来求解多模态多目标优化问题。首先,将一个种群分为几个子种群;之后,在优化过程中,每个子种群对应一个帕累托最优解集。配对、杂交、变异、环境选择等优化过程在每个子种群中独立进行。为了使子种群互相远离,使用了两个指标来使得解在决策空间中形成小环境。第一个指标是解与其所处的子种群的中心之间的距离。第二个指标是解与另一个最近的子种群的中心之间的距离。将这种方法应用到SPEA2和IBEA两种具有代表性的多目标演化优化算法上。实验结果表明对于大多数多模态多目标... 

【文章页数】：88 页

【学位级别】：硕士

【文章目录】：
摘要
ABSTRACT
Chapter 1 Introduction
    1.1 Background
    1.2 Related Work
        1.2.1 Real-world MMOPs
        1.2.2 Algorithms Proposed Before 2016
        1.2.3 Algorithms Proposed After 2016
    1.3 Main Contributions
    1.4 Thesis Organization
Chapter 2 Preliminaries
    2.1 Multi-objective Optimization Problem
    2.2 Multi-modal Multi-objective Optimization Problem
    2.3 Basic Definitions
    2.4 Representative Evolutionary Multi-objective Optimization Algorithms
        2.4.1 Dominance-based EMOA: SPEA2
        2.4.2 Indicator-based EMOA: IBEA
        2.4.3 Decomposition-based EMOA: MOEA/D
    2.5 Evaluation Methods
        2.5.1 IGD and IGDX
        2.5.2 Visual Examination
    2.6 Test Problems
        2.6.1 Scalable Polygon Test Problem
        2.6.2 SYM-PART 1 Test Problem
        2.6.3 SS-UF1 Test Problem
        2.6.4 Omni Test Problem
        2.6.5 TWO-ON-ONE Test Problem
    2.7 Summary
Chapter 3 Subpopulation Searching
    3.1 The Proposed Method
        3.1.1 The Basic Idea of Subpopulation Searching
        3.1.2 Implementation of Subpopulation Searching in SPEA2
        3.1.3 Implementation of Subpopulation Searching in IBEA
    3.2 Experimental Settings
        3.2.1 Parameters of SPEA2
        3.2.2 Parameters of SPEA2 with Subpopulation Searching
        3.2.3 Parameters of IBEA
        3.2.4 Parameters of IBEA with Subpopulation Searching
    3.3 Experimental Results
        3.3.1 Comparison between SPEA2 and SPEA2 with Subpopulation Searching
        3.3.2 Comparison between IBEA and IBEA with Subpopulation Searching
        3.3.3 The Effect of the Number of Subpopulations
    3.4 Summary
Chapter 4 Neighborhood Anchor
    4.1 The Proposed Method
        4.1.1 The Basic Idea of The Neighborhood Anchor
        4.1.2 Implementation of Neighborhood Anchor in SPEA2
        4.1.3 Implementation of Neighborhood Anchor in IBEA
    4.2 Experimental Settings
        4.2.1 Parameters of SPEA2 and IBEA
        4.2.2 Parameters of SPEA2 with Neighborhood Anchor
        4.2.3 Parameters of IBEA with Neighborhood Anchor
    4.3 Experimental Results for Neighborhood Anchor
        4.3.1 Comparison between SPEA2 and SPEA2 with Neighborhood Anchor
        4.3.2 Comparison between IBEA and IBEA with Neighborhood Anchor
    4.4 Summary
Chapter 5 Comparison between DNEA and Our Best Algorithm
    5.1 Experimental Settings
    5.2 Experimental Results
        5.2.1 Quantitative Analysis
        5.2.2 Visual Examination
    5.3 Discussion
    5.4 Summary
Conclusions
References
Acknowledgements


【参考文献】：
期刊论文
[1]A self-organizing multimodal multi-objective pigeon-inspired optimization algorithm[J]. Yi HU,Jie WANG,Jing LIANG,Kunjie YU,Hui SONG,Qianqian GUO,Caitong YUE,Yanli WANG.  Science China（Information Sciences）. 2019(07)



本文编号：3681075