Research On Robust Optimization Method Of Complex Network Under Cascading Failure Scenarios

In modern society,various complex networks are existing as power networks,communication networks,social networks and so on,which along with common network security problems,such as network attack,cascading failures,etc.In reality,even a small disturbance occurs in complex networks can bring out cascading failures,leading to serious consequences.Especially in recent years,the robustness of complex networks under cascading failures has become one of the hot topics in system science researches.This dissertation models the cascading failures and the corresponding recovery process in complex networks.On this basis,we then discuss the rational resource allocation and the node protection issue.The specific research contents are as follows:(1)We study the impact of dynamic resource allocation of node capacity on cascading failure process.Specifically,a node importance indicator is defined based on node capacity and load.Aiming at the Motter-Lai cascading failure model,we propose an importance-based dynamic resource allocation strategy of node capacity.Furthermore,we study the effect of the resource allocation range on cascading failures in resource redistribution.Finally,the impact of network topology on the robustness optimization effect is discussed.We perform our strategy on complex network models and real network data.It is found that the proposed dynamic resource allocation strategy can effectively control the scale of cascading failures.Simulation results illustrates our strategy is more effective for high-density scale-free networks and the first-order neighbor of nodes is the best capacity allocation range.(2)Aiming at the Gillespie-based cascading failure model,we study the node recovery problem after failure propagation.Specifically,the interaction between failure and restoration and the superimposed iterative network recovery process with failure and restoration are analyzed.We further propose an importance-based node recovery strategy.This strategy stipulates that the repair order of failure nodes is positively related to the node importance.In addition,the recovery resources allocated to each node are proportional to its importance indicator.Simulation results indicate that the best power parameter can shorten the recovery time to greatest scale and gain the best recovery effect,which tends to decrease and then increase as the node tolerance coefficient increases.(3)We study the node protection pattern against cascading failures,and propose a multiple-attribute decision making-based node protection strategy.Specifically,five typical node importance indicators are given as feature attributes,and an improved multiple-attribute decision making model is used to find key nodes that need to be protected.What's more,the redundant capacity resources allocated to key nodes is proportional to its node importance.Then we verify relevant work in interdependent network,compared with the importance-based node protection strategy,the multiple-attribute decision making-based node protection strategy can more effectively control the scale of cascading failures.Meanwhile,as the total amount of allocable redundant capacity resources increases,the control effect first increases and then decreases.