约束单目标与多目标进化算法及其应用
发布时间：2021-04-16 04:12
约束单（多）目标优化问题广泛存在于科学、经济和工程等诸多领域中,具有十分重要的理论意义和应用价值。在过去几十年中,如何有效求解约束单（多）目标优化问题日益受到相关学者的关注,并且提出了一系列的约束处理方法。由于约束本身的难度,约束间的相互作用,目标函数的难度以及目标函数与约束的相互作用,很难提出一种高效的约束处理方法。目前处理这些问题的算法主要可以为两种,第一种是使用传统的数学方法来求解约束问题,但通常需要给定问题的梯度信息,这对于有些实际问题,例如离散或目标函数连续但不可微的约束优化问题,是无法求解的。对于求解约束优化问题,在不考虑梯度信息的情况下开发其他方法是非常重要的。进化算法受自然的启发,结合约束处理方法,在求解约束问题时较于传统数学方法有之独特的优势。它是一个功能强大的优化工具,不需要计算梯度信息,且易于实现。在过去的几十年里,进化算法引起了许多研究者的兴趣,并提出了一些求解约束优化问题的约束优化进化算法。然而,约束单（多）目标优化进化算法的研究还没有得到充分、广泛的研究。本文对约束单（多）目标优化问题的算法设计及其相应的实际应用进行了讨论与研究。首先,针对约束单目标约束优化...
【文章来源】：广东工业大学广东省
【文章页数】：176 页
【学位级别】：博士
【文章目录】：
ACKNOWLEDGEMENTS
ABSTRACT
摘要
INDEX OF ABBREVIATIONS
CHAPTER 1 Introduction
1.1 Background
1.2 Basic Concepts
1.3 Constrained Single-objective Optimization
1.4 Constrained Multi-objective Optimization
1.5 Motivation
1.6 Contributions
1.7 Outline of the Thesis
CHAPTER 2 A Novel Constraint-Handling Technique Based on Dynamic Weights forConstrained Optimization Problems
2.1 Introduction
2.2 The Proposed Algorithm
2.2.1 The Proposed Constraint-handling Technique
2.2.2 The Framework of the Proposed Algorithm
2.3 Experimental Study
2.4 Discussion
2.4.1 The effectiveness of biased dynamic weights
2.4.2 The Sensitivity of the Proposed Algorithm to a
2.4.3 The Sensitivity of the Proposed Algorithm to a
2.5 Conclusion
CHAPTER 3 A Constrained Multi-objective Evolutionary Algorithm Based on BoundarySearch and Archive
3.1 Introduction
3.2 The Proposed Algorithm
3.2.1 Decomposition
3.2.2 Constraint-handling Method Based on Boundary Search and Archive
3.2.3 Crossover and Mutation
3.2.4 The Framework of the CM2M
3.3 Experiments
3.3.1 Parameter Settings
3.3.2 Performance Metrics
3.3.3 Experimental Results
3.4 Conclusion
CHAPTER 4 An Evolutionary Algorithm with Directed Weights for ConstrainedMulti-objective Optimization
4.1 Introduction
4.2 The Proposed Algorithm
4.2.1 The Proposed Constraint-handling Technique
4.2.2 The Proposed Framework
4.3 Experimental Studies and Discussion
4.3.1 Parameter Settings
4.3.2 Experimental results on CFs and CTPs
4.3.3 Experimental results on two engineering design problems
1 and N2"> 4.3.4 Sensitivity of the proposed algorithm to N1 and N2
4.3.5 Sensitivity of the proposed algorithm to λ
4.4 Conclusion
CHAPTER 5 Handling Multi-objective Optimization Problems with Unbalanced Constraintsand their Effects on Evolutionary Algorithm Performance
5.1 Introduction
5.2 The Proposed Test suite
5.2.1 The Proposed Test Problems
5.2.2 Illustrative Problem
5.3 Evolutionary Algorithms for Solving UCMOPs
5.3.1 Two Related Constraint-handling Techniques
5.3.2 Characteristics of the UCMOPs and Their Effects on EvolutionaryAlgorithms 615.4 Experimental Studies
5.4 Experimental Studies
5.4.1 Parameter Settings
5.4.2 Performance Metrics
5.4.3 Numerical Results on UCMOPs
5.4.4 Results on UCMOPs with different degrees of imbalance
5.5 Discussion
5.5.1 Effect of control function k on DW
5.5.2 Sensitivity of M2M-DW to
5.6 Conclusion
CHAPTER 6 A Cooperative Evolutionary Framework Based on an Improved Version ofDirected Weights for Constrained Multi-objective Optimization with Deceptive Constraints
6.1 Introduction
6.2 The Proposed Test suite
6.3 The Proposed Framework
6.3.1 The Pseudo-code of the Proposed Framework
6.3.2 The First Phase
6.3.3 The Second Phase
6.3.4 An Infeasibility Strategy
6.3.5 The Switching Condition
6.4 Experiments and Discussion
6.4.1 Compared Algorithms and Parameter Settings
6.4.2 Experimental Results on the Compared Algorithms
6.4.3 A Comparison between Two Frameworks
6.4.4 A Comparison between M2M-DW and IDW-M2M-CDP
6.4.5 Investigation of the Tolerance Value r
6.4.6 Investigation of the Parameter a
6.4.7 Effectiveness of the Infeasibility Utilization Strategy
6.4.8 Benefit of the Two Switching Phases
6.5 Conclusion
Conclusion and Future Work
REFERENCE
List of Published/Submitted Papers
Published papers
Papers under review
本文编号：3140715
【文章来源】：广东工业大学广东省
【文章页数】：176 页
【学位级别】：博士
【文章目录】：
ACKNOWLEDGEMENTS
ABSTRACT
摘要
INDEX OF ABBREVIATIONS
CHAPTER 1 Introduction
1.1 Background
1.2 Basic Concepts
1.3 Constrained Single-objective Optimization
1.4 Constrained Multi-objective Optimization
1.5 Motivation
1.6 Contributions
1.7 Outline of the Thesis
CHAPTER 2 A Novel Constraint-Handling Technique Based on Dynamic Weights forConstrained Optimization Problems
2.1 Introduction
2.2 The Proposed Algorithm
2.2.1 The Proposed Constraint-handling Technique
2.2.2 The Framework of the Proposed Algorithm
2.3 Experimental Study
2.4 Discussion
2.4.1 The effectiveness of biased dynamic weights
2.4.2 The Sensitivity of the Proposed Algorithm to a
2.4.3 The Sensitivity of the Proposed Algorithm to a
2.5 Conclusion
CHAPTER 3 A Constrained Multi-objective Evolutionary Algorithm Based on BoundarySearch and Archive
3.1 Introduction
3.2 The Proposed Algorithm
3.2.1 Decomposition
3.2.2 Constraint-handling Method Based on Boundary Search and Archive
3.2.3 Crossover and Mutation
3.2.4 The Framework of the CM2M
3.3 Experiments
3.3.1 Parameter Settings
3.3.2 Performance Metrics
3.3.3 Experimental Results
3.4 Conclusion
CHAPTER 4 An Evolutionary Algorithm with Directed Weights for ConstrainedMulti-objective Optimization
4.1 Introduction
4.2 The Proposed Algorithm
4.2.1 The Proposed Constraint-handling Technique
4.2.2 The Proposed Framework
4.3 Experimental Studies and Discussion
4.3.1 Parameter Settings
4.3.2 Experimental results on CFs and CTPs
4.3.3 Experimental results on two engineering design problems
1 and N2"> 4.3.4 Sensitivity of the proposed algorithm to N1 and N2
4.4 Conclusion
CHAPTER 5 Handling Multi-objective Optimization Problems with Unbalanced Constraintsand their Effects on Evolutionary Algorithm Performance
5.1 Introduction
5.2 The Proposed Test suite
5.2.1 The Proposed Test Problems
5.2.2 Illustrative Problem
5.3 Evolutionary Algorithms for Solving UCMOPs
5.3.1 Two Related Constraint-handling Techniques
5.3.2 Characteristics of the UCMOPs and Their Effects on EvolutionaryAlgorithms 615.4 Experimental Studies
5.4 Experimental Studies
5.4.1 Parameter Settings
5.4.2 Performance Metrics
5.4.3 Numerical Results on UCMOPs
5.4.4 Results on UCMOPs with different degrees of imbalance
5.5 Discussion
5.5.1 Effect of control function k on DW
5.5.2 Sensitivity of M2M-DW to
5.6 Conclusion
CHAPTER 6 A Cooperative Evolutionary Framework Based on an Improved Version ofDirected Weights for Constrained Multi-objective Optimization with Deceptive Constraints
6.1 Introduction
6.2 The Proposed Test suite
6.3 The Proposed Framework
6.3.1 The Pseudo-code of the Proposed Framework
6.3.2 The First Phase
6.3.3 The Second Phase
6.3.4 An Infeasibility Strategy
6.3.5 The Switching Condition
6.4 Experiments and Discussion
6.4.1 Compared Algorithms and Parameter Settings
6.4.2 Experimental Results on the Compared Algorithms
6.4.3 A Comparison between Two Frameworks
6.4.4 A Comparison between M2M-DW and IDW-M2M-CDP
6.4.5 Investigation of the Tolerance Value r
6.4.6 Investigation of the Parameter a
6.4.7 Effectiveness of the Infeasibility Utilization Strategy
6.4.8 Benefit of the Two Switching Phases
6.5 Conclusion
Conclusion and Future Work
REFERENCE
List of Published/Submitted Papers
Published papers
Papers under review
本文编号：3140715
本文链接：https://www.wllwen.com/shoufeilunwen/xxkjbs/3140715.html
