复杂网络中级联失效行为分析与结构可控性研究

发布时间：2018-05-25 06:55

  本文选题：复杂网络 + 级联失效　； 参考：《天津理工大学》2017年硕士论文

【摘要】：信息时代的脚步声越来越近,复杂网络学科的引入为更加深入地研究复杂系统的结构特征与其上的动力学行为提供了详细的理论基础和方法。本论文涉及的研究工作主要与复杂网络系统的模型实现与特性分析相关:基于耦合网络攻击问题的研究,有助于增强对现实生活中复杂相依系统的理解和认知;采用数值仿真的方法分析了复杂网络的结构可控能力,研究结果有助于探寻提升网络系统可控能力的关键因素;基于Fitness模型对幂指数可变的无标度网络进行建模实现,并进一步深入分析其重要的结构特征及动力学特性,研究结果有力证明了幂指数γ对无标度网络性能的重要影响。论文的主要研究内容和创新性成果概括如下:1.攻击引发的相依网络系统中级联失效行为分析。针对蓄意攻击引发的相依系统级联失效行为,进一步验证耦合关系对相依系统抗攻击能力的影响,引入相依网络系统模型,通过控制该网络模型的两个调优参数,深入分析度分布,尤其是度异质程度的变化规律。以数值模拟手段,讨论和验证度分布异质程度对单网络和相依网络结构脆弱性的影响,着重分析网络耦合强度对蓄意攻击情况下的相依网络系统的作用,研究结果有助于进一步深刻理解现实世界中多网络系统的结构脆弱性特征。2.Local-World网络的结构可控性研究。Local-World网络模型已经成为一个成熟的网络模型框架,对其进行网络可控性研究有着重要的意义。以控制信号的可达性为基础,借助仿真手段,研究了Local-World网络中影响结构可控性的关键因素。结果表明,对于密集网络,可控性取决于网络规模N,Local-World效应暂时可忽略不计;对于稀疏网络,与网络规模N相比,局部网络规模M更能显著地影响结构可控能力。3.幂指数可变的无标度网络的建模与分析。度分布服从幂律的网络被称为无标度网络,Fitness模型作为典型无标度模型之一,能生成度分布幂指数可变的无标度网络,其幂指数γ的取值范围为γ∈(2,∞)。在简介Fitness网络建模机理的基础上,给出了模型的具体实现方法,并针对所生成的网络,进一步对其重要的结构特征及动力学特性进行了深入分析,包括度分布、抗攻击能力、同步性能及结构可控能力,重点研究了幂指数γ对网络性能的影响。
[Abstract]:The step sound of the information age is getting closer and closer. The introduction of the complex network provides a detailed theoretical basis and method for the further study of the structural characteristics and the dynamic behavior of the complex system. The research work in this thesis is mainly related to the model realization of complex network system and characteristic analysis. The research based on coupling network attack can help to enhance the understanding and cognition of complex dependent system in real life. The structural controllability of complex networks is analyzed by numerical simulation, and the results are helpful to explore the key factors to improve the controllability of network systems. Based on the Fitness model, the scale-free networks with variable power exponents are modeled and implemented. Furthermore, the important structural and dynamic characteristics are analyzed. The results show that the power exponent 纬 has an important effect on the performance of scale-free networks. The main research contents and innovative results are summarized as follows: 1. Analysis of cascading failure behavior in dependent Network system caused by attack. Aiming at the cascade failure behavior of dependent system caused by intentional attack, this paper further verifies the effect of coupling relation on the anti-attack ability of dependent system, introduces the dependent network system model, and controls the two tuning parameters of the network model. The variation of degree distribution, especially degree heterogeneity, is deeply analyzed. By means of numerical simulation, the influence of degree distribution heterogeneity on the vulnerability of single network and dependent network structure is discussed and verified, and the effect of network coupling strength on dependent network system under intentional attack is analyzed. The results of the study are helpful to further understand the structural vulnerability characteristics of multi-network systems in the real world. 2. The structural controllability of Local-World networks; the Local-World network model has become a mature network model framework. It is of great significance to study the network controllability. Based on the reachability of control signal, the key factors influencing the controllability of Local-World network are studied by means of simulation. The results show that, for dense networks, controllability depends on the network size, which is negligible for the time being, and for sparse networks, the local network size M can significantly affect the controllability of the structure compared with the network size N. Modeling and analysis of scale-free networks with variable power exponents. The scale-free network Fitness model is one of the typical scale-free models. It can generate scale-free networks with variable power exponent. The range of power exponent 纬 is 纬 鈭，

本文编号：1932514