Abstract: Nowadays, with the continuous development of science and technology, society becomes more prosperous, but at the same time, systems in society become more complex. The complex network theory plays an important role in many kinds of network research. With the rapid development of the economy, the development of China’s financial network changes each day. The types of financial institutions and financial products are becoming more diversified, and the relations among financial institutions are becoming more complex, and the structure of the financial network has also become more intricate.In order to reduce the cascading failure of complex financial network and prevent the major systemic financial risks and enhance the invulnerability of complex financial networks, it is necessary to clarify the structural relationship between financial networks, analyze the mechanism of cascading failure of financial networks, and study the reliability of financial networks based on cascading failure model, and then explore the methods to avoid cascading failure of financial networks. Based on the dynamic characteristics and cascade failure mechanism of complex network, financial network and its cascade failure model are established, and the initial loads and capacities of nodes and connecting edges are defined.Then, the reliability of cascade failure model of financial network under different conditions is quantitatively analyzed by simulation, and the result shows that the model is effective, which is helpful for the prevention from the failure of complex financial network and further research into it.
Cascading failures have attracted considerable attention due to their widespread occurrence in various real world systems and their highly destructive effect on network robustness. Although the current research on cascading failures in complex networks is relatively in-depth, it still faces many challenges as the structural and dynamic heterogeneity of real financial systems . This paper focuses on the model of complex financial networks with respect to cascading failures. Cascading failures in complex financial networks can lead to large-scale network collapse and a series of disastrous consequences, but cascading failures can be effectively prevented and solved by modeling, simulation, and quantitative analysis.Based on the complex network theory, this paper establishes the financial network model by using the graph theory language, and then applies the cascade failure model to the study of complex financial network for the first time.Finally, based on this model, we analyze the reliability of financial network and draw the following two conclusions:
(1)In the process of cascading failure simulation of financial networks, increasing the number of iterations will lead to the decrease in network efficiency and the increase in node failure efficiency. When the network is deliberately attacked, the network efficiency will decrease rapidly, the node failure rate will increase rapidly, the network will be greatly impacted, and the impact will be much greater than that in the case of random attack.
(2)Under the same attack mode, improving the risk tolerance coefficient will help to improve the network efficiency, reduce the failure rate, improve the network reliability, and even avoid the occurrence of network cascade failure, but it will pay the corresponding cost.
In addition, the relationship between cost and risk tolerance coefficient is discussed, and it is found that the relative change relationship between α and relative increased cost e is as follows: e=1+α. Finally, this paper puts forward some suggestions on how to optimize the reliability of financial network from two aspects: strengthening the protection of key institutions and taking emergency measures after failure.

Key words: complex network, financial network, cascading failure model, simulation, reliability analysis

摘要： 复杂网络中级联失效现象会导致网络大面积崩溃进而导致一系列灾难性后果,通过对其建模、仿真和定量分析可以有效预防和解决级联失效问题。本文在复杂网络理论基础上,基于复杂网络的动力学特征和级联失效机理,运用图论语言建立了金融网络模型;首次将级联失效经典模型运用到复杂金融网络研究中,提出了金融网络级联失效模型。然后通过仿真定量分析了不同条件下金融网络级联失效模型的可靠性,验证了模型的有效性,有助于复杂金融网络失效的预防及进一步研究。研究发现:第一、在金融网络级联失效模型仿真过程中,增加迭代次数,会导致网络效率下降,节点失效率上升。当网络受到蓄意攻击时,网络效率迅速下降,节点失效率快速上升;网络受到极大冲击,且这种冲击远大于随机攻击情况下。第二、同一种攻击方式下,提升风险容忍系数α,有助于提高网络效率、降低失效率,提升网络可靠性,甚至可以避免网络级联失效的发生,但这会付出相应的成本。

关键词: 复杂网络, 金融网络, 级联失效模型, 仿真, 可靠性分析