Research On The Invulneriability Of Interdependent Networks And Its Applications In Power System

Individuals in the nature will be connected into a network in a certain way. With the progress and developments of society, these connections will be more intensive. However, some networks have exceeded the description of traditional graph theory due to the increment of scale. We call these networks as complex networks, such as various infrastructure networks. Complex network not only provides interaction and convenience for individuals in the wide area, but also shows the potential risk and vulnerability. This makes failures possible to propagate rapidly in the whole network which makes the network like The Sword of Damocles. In the past decades, most researches on network invulnerability always concentrate on the isolated network. However, recent research indicates that the network (system) will not exist independently, and there will be coupling dependent relations with other networks, such as the power-communication networks, the power-water supply networks, etc. These relations not only promote communications between networks, but also bring in risks of cascading failures within the whole network. Therefore, it is an urgent task to analyze the cascading failures in interdependent networks (infrastructures) and find out the approaches to improve the invulnerability.Network's structure determines its property. To study the dynamic characteristics of interdependent networks, we should start from its vital structures and parameters. As the important component of interdependent networks, the dependent (coupled) edge determines its special properties which are different from the isolated network. However, most researches always adopt the topology cascading model that does not take into account the load and capacity of nodes in the actual networks. Based on a degree load-based model, this paper proposes a new index to evaluate invulnerability, the total load loss, and validates the accuracy of simulation experiments with theory. First, this paper investigates the influence of different coupled patterns, coupling strengths and attacking strategies on the invulnerability of interdependent networks. Second, this paper analyzes how the invulnerability will change with different load and capacity parameters. In the end, this paper compares different network topologies. All the results obtained can help to understand the cascading process of interdependent infrastructures, and can provide some references for the construction of infrastructures.In actual networks, some nodes seeming unordered always present a certain hierarchical structure which can be regarded as the basic mesoscale unit for the network segmentation. Furthermore, the hierarchical structure is an unavoidable structure during the evolution of most networks. However, recent researches are always limited to the analysis and improvement of methods for hierarchical classification, and there is little work to rise to a height of dynamic analysis of interdependent networks.K-core structure is a common kind of mesoscale unit of network which is not only related to network topology, but also can change its some dynamic characteristics. To investigate the influence of the depth of k-core layers on the invulnerability of interdependent networks, this paper proposes an evolving method to generate the scale-free network with successive k-core layers, and validates its veracity comparing with actual networks via theoretical derivation and simulations. Based on the load-based cascading model, this paper studies the vulnerability of both symmetric and asymmetric k-core interdependent networks to find out the better k-core parameters to improve the invulnerability.Cluster structure is another common kind of mesoscale unit of network, including core-periphery structure and community structure. Nodes and edges in network are not random, and some similar or close nodes or edges always tend to present a phenomenon of group. To investigate the influence of different cluster structures and parameters on the invulnerability of interdependent networks, this paper proposes a unified evolving network model which has the adjustable cluster structures, and validates its scale-free and cluster properties via theoretical derivation and simulations. Based on the load-based cascading model, this paper studies the influence of tightness of different cluster structures on the invulnerability of interdependent networks. This paper also compares the pros and cons of the two kinds of cluster structures. Furthermore, this unified framework is extended to the interdependent networks with multiple communities, and the influence of the number of communities and coupling patterns on invulnerability is also investigated.As a kind of critical infrastructure network, the power system receives much concern due to its special power background. Traditional power system is only responsible for the generation, output and distribution of electricity. However, with the rapid developments of information and control technology, information (communication) network has gradually become an integral part of power system. Previous studies pay more attentions to independent power or communication network which cannot understand the overall characteristics of power system.To study the influence of structure and parameters of information network on the cascading failures, this paper analyzes the interactive process on energy and information of the power and information networks.Inspired by the accident process of "Ukraine blackouts" recently, a coupled network model with electricity background is proposed. Based on a DC power flow cascading model, this paper investigates the influence of different coupled patterns (inner self-similarity and random coupled patterns) and coupling strengths on the power cascading failures, and investigates the influence of failure probability between networks on the power cascading failures to validate the differences between various cascading models. This paper investigates the influence of the topology and dispatching center of communication network on the power cascading failures. Furthermore, this paper also proposes a novel coupled network model with both communication and energy edges, and analyzes the relations between the number of the two kinds of edges and power cascading failures. All the results obtained can provide references and guidance for the construction of the new generation power system.