复杂网络结构特征分析与动力学研究

发布时间：2018-02-27 01:30

本文关键词： 复杂网络数据分析网络结构特征分析动态过程免疫控制策略　出处：《浙江大学》2014年博士论文　论文类型：学位论文

【摘要】：随着现实复杂系统数据的可获得和可分析,人们对系统网络化结构特征的理解逐渐深入,尤其是针对结构特征与系统行为机制关系的分析越来越受到研究者的关注。本文在综述了复杂网络理论系统结构特征研究现状的基础上,从网络化建模与分析的角度对对象内在特征进行深入挖掘,并进一步针对网络结构对网络上动态过程机制的影响展开具体研究。在这一主线上,本文的主要研究内容如下： (1)在复杂网络建模与对象结构特性分析方面,针对流程工业炼油系统进行复杂任务—资源网络建模,结合生产方案和装置功能对网络结构进行分析。根据生产规则建立的简化模型可以验证结构特征的可靠性,说明结构特征与生产过程的紧密联系,网络分析方法能够为系统的整体分析提供可用信息。 (2)当数据规模和复杂程度增加时,网络化的分析方法可以进一步结合对象属性和结构特征,深入挖掘数据的内在信息。使用某网站的在线交易数据,构建城市交易关系网络,充分利用网络中节点的全局连接关系,通过骨干分析方法和资金流动试验,深入挖掘对象结构特征中蕴含的信息。从网络的视角对我国各地的在线交易水平进行综合分析,并运用可视化技术将结果进行展示,体现出结构特征分析在数据挖掘方面提供的独特视角。 (3)在网络结构特征与系统动态机制的关系研究方面,首先探讨不同的静态网络结构特征对网络上动态过程带来的影响。对供应链网络上物流过程的分析发现,现实系统中常见的簇结构会明显改善系统资源的有效利用,帮助供应链网络获得更高的物流效率。这一分析明确了结构特征对系统动态机制的影响作用,并在特定对象中进行了量化比较。 (4)针对结构随时间变化的网络,讨论动态网络结构对网络上动态过程机制的影响。在利用活性驱动模型建立的动态网络上,使用种群感染模型模拟消息、疾病等传播过程,结合感染和扩散两个具体传播机制,对传播过程的全局扩散阈值进行了解析分析,仿真试验进一步验证了结构动态特性对传播过程的具体影响。 (5)针对结构随时间变化的网络,探讨利用结构信息的控制方法对网络上动态过程的控制作用。结合活性驱动模型的动态结构信息,提出随机、目标、邻居三种免疫策略。通过理论推导和仿真试验,比较三种策略对传播过程的控制效果。结果显示,三种策略都可以对传播过程起到一定的抑制作用,其中目标免疫策略的控制效果最好,随机策略的效果最弱,邻居免疫策略能够克服全局信息的缺失,提供较好的控制效果,是一种适用于动态网络中传播过程的控制方法。最后在总结全文的基础上,提出了复杂网络理论结构特征分析与应用方面有待深入研究的几个问题。
[Abstract]:With the availability and analysis of real complex system data, people have a deeper understanding of the characteristics of the networked structure of the system. In particular, the analysis of the relationship between structural characteristics and system behavior mechanisms has attracted more and more attention of researchers. In this paper, based on a review of the current research situation of structural characteristics of complex network theory systems, From the angle of network modeling and analysis, the intrinsic characteristics of the object are deeply mined, and the influence of network structure on the dynamic process mechanism on the network is further studied. On this main line, the main contents of this paper are as follows:. 1) in the aspect of complex network modeling and object structure characteristic analysis, complex task-resource network modeling is carried out for refining system of process industry. The network structure is analyzed in combination with the production scheme and the device function. The simplified model established according to the production rules can verify the reliability of the structural features and illustrate the close relationship between the structural features and the production process. The network analysis method can provide the available information for the overall analysis of the system. 2) when the data scale and complexity increase, the network analysis method can further combine the object attribute and the structure characteristic, deeply excavate the internal information of the data, use the online transaction data of a website, construct the city transaction relation network. This paper makes full use of the global connection relationship of nodes in the network, through backbone analysis method and fund flow test, deeply excavates the information contained in the structural characteristics of objects. From the perspective of the network, this paper makes a comprehensive analysis of the level of online transactions in various parts of China. Visualization technology is used to display the results, which reflects the unique perspective of structural feature analysis in data mining. 3) in the research of the relationship between the network structure characteristics and the dynamic mechanism of the system, firstly, the influence of different static network structure characteristics on the dynamic process on the network is discussed, and the analysis of the logistics process in the supply chain network shows that, The common cluster structure in the real system will obviously improve the effective utilization of the system resources and help the supply chain network to obtain higher logistics efficiency. This analysis clarifies the influence of the structural characteristics on the dynamic mechanism of the system. A quantitative comparison is made in a specific object. In this paper, we discuss the influence of dynamic network structure on the dynamic process mechanism of the network. In the dynamic network based on the active driving model, we use the population infection model to simulate the transmission process of message, disease and so on. The global diffusion threshold of the transmission process is analytically analyzed by combining the two specific transmission mechanisms of infection and diffusion. The simulation results further verify the specific influence of the structural dynamic characteristics on the propagation process. In view of the network of structure changing with time, this paper discusses the control effect of the control method of the structure information on the dynamic process on the network. Combining with the dynamic structure information of the active driving model, the random and objective are put forward. Through theoretical derivation and simulation experiments, the control effect of the three strategies on the transmission process is compared. The results show that the three strategies can inhibit the transmission process to a certain extent. The control effect of target immune strategy is the best, and that of random strategy is the weakest. Neighbor immune strategy can overcome the lack of global information and provide better control effect. It is a control method suitable for the propagation process in dynamic network. Finally, on the basis of summing up the full text, several problems that need to be studied deeply in the analysis and application of the structure characteristics of complex network theory are put forward.
【学位授予单位】：浙江大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：O157.5

【参考文献】