Research On Robustness Of Complex Network For Suppressing Cascading Failure

The reliability of infrastructure networks which are the basic and crucial components ofour living circumstances controls the high-speed development of social economy and normallife of the people. But the abominable disasters and large-scale network paralysis whichcaused by cascading failure are frequently occurred in the infrastructure networks. Along withthe increasing of the scale of infrastructure networks, the networks can facilitate the economicdevelopment and people’s life, meanwhile, the complexity of the networks is also enhanced,which may cause the research and control on infrastructure network to become more difficultwhen the network suffers cascading failure. In order to solve the problem that how to obtainhigh robustness with fixed cost, this paper summarizes the static robustness, the dynamicrobustness, redistribution of loads and the measurements of robustness of complex networkbased on the theory of complex network. And the researches on the robustness of railwaynetwork and interdependent network under cascading failure are carried out in this paper.More details and innovation are as follows:1).The dynamic robustness of the complex network is the connectivity of network underdynamic loads. It can be mitigated, even be suppressed the spread of cascaded failure withreasonable redistribution of the load of edges (nodes) when the network suffers failures. Interms of the actual traffic congestion, a local load redistribution strategy is proposed, it canredistribute the additional load which obtained by the traffic congestion to the neighborsections, and relieve the awkward situation of jams. Taking into account the disadvantage ofthe algebraic connectivity and natural connectivity, an improved natural connectivity is putforward, which not only can rapidly and sensitively respond with the failures, but also canclearly distinguish whether the network connects or not.2).It plays a crucial role for railway transportation systems, especially in Chinese railwaytransportation systems in delivery of cargo and personnel. By abstracting Chinese railwaytransportation systems as three network models, we optimize the networks with an optimalalgorithm based on adding combined edges. Then adopt Directed Chaos-mutation SortedDiscrete Particle Swarm Optimization (DCSD-PSO) algorithm to seek for the optimal addededges with different measurements of robustness. Simulations are carried out on typicalnetworks and actual networks, the results show that DCSD-PSO algorithm is feasible andadvanced than the BPSO algorithm. Finally, the optimal topology structure of the networksobtained by the two algorithms is simulated in the situation that the networks are under cascading failure, and it proves that network optimized by DCSD-PSO algorithm performsbetter robustness.3).As a consequence of the coupling influences of network, interdependent network maysuffer more disastrous effects than signal network when a cascading failure occurs in thenetworks simultaneously. In order to study the impact factors on robustness of interdependentnetworks, symmetrically and asymmetrically coupled interdependent network models areconstructed based on three typical network models. Firstly, we calculate the cost threshold ofsix different interdependent networks, and find that the coupling edges have more influenceon the cost of interdependent network than the edges of sub-networks do. Furthermore, therelationship of the two parameters (、) with the cost of network is obtained by simulationand theoretical analysis, and the cost of network correlated with the two parameters is proved.Finally, by setting the measurement of robustness as a function of the cost of network,simulations on interdependent networks for suppressing cascading failure are provided for thevalue of the parameters corresponding to the strongest robustness and the relationshipbetween the robustness with the coupling strength, and it is proved that the robustness ofnetwork neither positive correlation nor negative correlation with the coupling strength.