进化计算中的复杂网络动力学研究

发布时间：2018-01-03 13:12

本文关键词：进化计算中的复杂网络动力学研究　出处：《河北工程大学》2017年硕士论文　论文类型：学位论文

【摘要】：基于达尔文进化论的进化算法在求解问题时,可将问题描述成自然界中种群的进化过程,遵循适者生存的机制,通过种群的不断进化,求得最优解或满意解。进化计算的研究较早,发展较成熟,且广泛应用于社会的各个领域。但现有的研究只注重算法的最终结果或预测结果,往往忽视了进化过程中个体之间的关系和这些关系随进化过程的变化,以及该变化关系对进化结果和收敛速度的影响。而本文主要针对这个被忽视的问题展开研究。通过研究优化过程中个体之间的变化关系抽象出进化计算形成的网络结构中蕴含的复杂网络结构,并挖掘出进化计算中的复杂网络动力学现象。随着复杂网络理论体系的不断发展和研究,基于复杂网络研究的应用越来越多。由于算法在迭代过程中,参与进化的个体总数不变,即网络结构中节点的总数不变。网络结构随着边的连接概率动态变化。网络结构的变化影响着网络动力学演化过程,而动力学的演化过程也影响网络边的动态重连,网络结构与动力学演化过程之间的动态作用称为“共同演化”过程,且受到广大科研工作者的关注。这两个看似完全不同的研究领域,复杂网络和进化计算,两者之间是否存在某种隐藏的结构关系,复杂网络动力学能否描述算法的优化过程,将成为未来研究的一大亮点。本文讨论了一个完全不同领域的相互交叉研究:进化计算中是否蕴含着复杂网络动力学现象。首先研究一般进化计算及改进算法的优化过程;然后分析优化过程中各个体之间的变化关系;再用复杂网络模型将个体之间的变化关系进行动力学过程描述,并讨论其蕴含的复杂网络结构;最后分析该变化对进化结果和收敛速度的影响,并提出算法的改进意见,提高种群的多样性。实验结果表明进化计算的优化过程可以用复杂网络动力学描述,利用复杂网络理论可有效控制并改进进化算法。本文的研究对于复杂网络的深入研究以及进化计算的改进、优化和控制等应用方面具有一定的理论意义和应用价值。
[Abstract]:The evolutionary algorithm based on Darwin's theory of evolution can describe the problem as the evolution process of population in nature, follow the mechanism of survival of the fittest, and continue to evolve through population. The research of evolutionary computing is earlier, more mature, and widely used in various fields of society. However, the existing research only focuses on the final results of the algorithm or the prediction results. The relationships between individuals in the evolution process and their changes with the evolution process are often ignored. And the influence of the change relation on the evolution result and convergence rate. This paper mainly focuses on the neglected problem. By studying the variation relation between individuals in the optimization process, the network formed by evolutionary computation is abstracted. The complex network structure contained in the network structure. With the development and research of complex network theory system, more and more applications based on complex network are found. Because the algorithm is in the iterative process. The total number of individuals involved in evolution is the same, that is, the total number of nodes in the network structure is unchanged. The network structure changes dynamically with the connection probability of the edge. The network structure changes affect the evolution process of the network dynamics. The dynamical evolution process also affects the dynamic reconnection of the network edge. The dynamic interaction between the network structure and the dynamic evolution process is called "co-evolution" process. These two seemingly different research fields, complex networks and evolutionary computing, have some hidden structural relationship. Can complex network dynamics describe the optimization process of the algorithm. This paper discusses an entirely different field of intersecting research:. Whether there are complex network dynamics phenomena in evolutionary computing. Firstly, the optimization process of general evolutionary computing and improved algorithm is studied. Then the relationship between the individuals in the process of optimization is analyzed. Then the dynamic process of the relationship between individuals is described by using the complex network model, and the complex network structure is discussed. Finally, the influence of the change on the evolution result and convergence rate is analyzed, and the improvement of the algorithm is proposed to improve the diversity of the population. The experimental results show that the optimization process of evolutionary computation can be described by complex network dynamics. The evolutionary algorithm can be effectively controlled and improved by using the complex network theory. The application of optimization and control has certain theoretical significance and application value.
【学位授予单位】：河北工程大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：O157.5;TP18

【参考文献】