相依网络在不同相依方式下级联故障的鲁棒性研究

发布时间：2018-05-12 15:36

本文选题：相依网络 + 相依方式　；参考：《华东交通大学》2017年硕士论文

【摘要】：随着现代化设备的日益更新,先进的科学技术使得生活中各种类型的基础设施网络之间相互依赖的特性越来越紧密。所以,如果像以前一样只是单单研究孤立网络的鲁棒性,那么生活中这些网络与网络之间存在的特殊性质将无法探索,整个网络体系的鲁棒性也无法得到深入的研究。至此,本文以复杂网络理论知识为基础,对“相依网络在不同相依方式下级联故障的鲁棒性”这个主题进行仿真分析,针对相依网络模型的级联故障问题进行仿真比对。本文主要的研究成果如下:首先,本文对复杂网络的理论基础作了详细的介绍,并阐述了基于复杂网络理论的相依网络的结构特性。从相依网络的概念、相依网络级联故障模型以及相依网络鲁棒性三个方面进行了归纳总结,从而为后续的研究工作提供参考。其次,提出了由BA无标度网络模型和WS小世界网络模型两两相互连接构建成的相依网络模型。在随机攻击和蓄意攻击这两种攻击方式下,对比了七种相依方式对于相依网络抵御级联故障的能力,通过仿真图横向分析出哪种网络模型以及哪种相依方式的鲁棒性更好。这七种相依方式分别是:HD-HD、HD-LD、HB-HB、HB-LB、HD-HB、HD-LB以及R-R。而仿真结果表明,含有BA网络的相依网络模型在面对蓄意攻击时,相依方式为HD-LD和HB-LB的鲁棒性较好;但在面对随机攻击时,HD-HD和HB-HB的鲁棒性较好。而不论是蓄意攻击还是随机攻击,WS-WS构成的相依网络中,HD-HD和HB-HB的鲁棒性表现较好。所以能够说明相依方式为HD-HD和HB-HB对于网络在抵御随机攻击时有着较强的鲁棒性,这些结论为构建较好鲁棒性的相依网络具有借鉴意义。最后,将节点负载重分配的概念引入了整个网络的级联故障过程中。定义了参数?作为节点负载中度数和介数对负载容量的贡献比,同时定义了容忍系数?作为构建网络的成本。将三种相依网络模型BA-BA、BA-WS以及WS-WS加入负载重分配的相依网络鲁棒性的研究当中,对比了五种相依方式(HD-HD、HD-LD、HB-HB、HB-LB、R-R)对相依网络鲁棒性造成的影响。通过研究发现,当?(28)0.2时,由度数和介数按比例构成的节点负载容量的相依网络鲁棒性最佳。而对于?(28)0.2时,五种相依方式的相依网络的鲁棒性表现较好是HB-HB和HD-HD相依方式。这些结论为我们以后研究相依网络的鲁棒性以及如何在最低成本下建立一个鲁棒性较好的网络提供了一定的参考依据。
[Abstract]:With the development of modern equipment, advanced science and technology make the interdependent characteristics of various types of infrastructure networks become more and more closely. Therefore, if we only study the robustness of isolated networks as before, the special properties between these networks and networks in life will not be explored, and the robustness of the entire network system will not be studied in depth. So far, based on the theoretical knowledge of complex networks, this paper simulates and analyzes the subject of "robustness of cascading faults in dependent networks under different dependent modes", and makes a simulation comparison of cascading faults in dependent network models. The main research results of this paper are as follows: firstly, the theoretical basis of complex network is introduced in detail, and the structural characteristics of dependent network based on complex network theory are expounded. In this paper, the concept of dependent network, cascading fault model of dependent network and robustness of dependent network are summarized, so as to provide reference for further research. Secondly, a dependent network model is proposed, which is composed of BA scale-free network model and WS small-world network model. Under the two attack modes of random attack and deliberate attack, the ability of seven dependent modes to resist cascading faults is compared, and the robustness of which network model and which dependent mode is better is analyzed horizontally by simulation diagram. The seven dependent modes are: 1 / HD-HDN HD-LDN HB-HB-HB-LBNHD-HBHD-LB and R-R. The simulation results show that the dependent network model with BA network is robust to HD-LD and HB-LB when faced with intentional attack, but the robustness of HD-HD and HB-HB is better in the face of random attack. However, the robustness of HD-HD and HB-HB is better in the dependent networks composed of intentional attack or random attack. So it can be explained that HD-HD and HB-HB have strong robustness to resist random attacks. These conclusions have reference significance for constructing dependant networks with better robustness. Finally, the concept of node load redistribution is introduced into the cascade fault process of the whole network. Defined parameters? As the contribution ratio of the degree and medium to the load capacity in the node load, the tolerance coefficient is also defined. As a cost of building a network. In this paper, the robustness of three dependent network models, BA-BAN BA-WS and WS-WS, which is added to load redistribution, is studied. The effects of five dependence modes, HD-HDN, HD-LDHD-HB-HB-LBCR-R), on the robustness of dependent networks are compared. It is found that the dependant network of node load capacity composed of degrees and mediums is the most robust when it is 0.2. However, the robust performance of the five dependent networks is that of HB-HB and HD-HD. These conclusions provide a reference for us to study the robustness of dependent networks and how to build a robust network at the lowest cost.
【学位授予单位】：华东交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：O157.5;TM732

【参考文献】