Cascade Failure Control Research Of Dependency Network Based On The Strategy Of Load Redistribution

With the deepening of complex network research,experts have gradually found that the network is not isolated from each other;many of the actual situations are dual-network or multi-network coupling operation,such as power-communication network.In this situation,cascade failure problems become more complicated.For example,the power network node collapse will cause the communication network node energy shortage and collapse,resulting in the power network communication regulation adverse.The network collapse become larger and the actual loss can not be estimated.The paper studies the cascade failure problem of the dependent network.Based on the load-energy model,we analyse the robustness of the dependent network from the aspects of network structure and network dynamic mechanism.Firstly,the influence of the basic topology properties of the network on the robustness of the network is studied by simulating the deliberate attack and the random attack strategy under the single variable principle.Then,on the basis of the above research,we further analyze the changes brought by network heterogeneity,and from the perspective of network dynamic mechanism,we propose a load reallocation strategy based on local real-time processing capability,and prove it to a certain extent can effectively control the size of the network crash.Considering the flow of load in the network is global,the load reallocation can not simply consider the local node capability.Therefore,this paper puts the network node importance into the load reallocation strategy,and proposes a load reallocation index LDPR,based on the PageRank algorithm,taking into account the global characteristics and local characteristics of the node.And in the follow-up experiment,compared with several other classic load reallocation strategies,the LDPR index is verified to be effective and stable.The main contributions of this study include: Firstly,through simulation experiments,it is found that the network structure factors have a greater influence on the robustness of the network than the network dynamic mechanism.Secondly,according to the inherent mechanism of cascade failure,a real-time processing capacity of the indicators is proposed and is verified to a certain extent to improve the dependency network robustness;Third,combined with the local characteristics of network nodes and global characteristics,a redistribution strategy based on LDPR is proposed,which has a very good suppression effect to the cascade failure.