Research On Optimization Of Network Robustness Based On Multi-objective Evolutionary Algorithm

Networks can represent many real-world complex systems.Systems like internet,power grids and fuel distribution networks need to be robust and capable of surviving from failures or intentional attacks.Designing robust networks has attracted increasing attentions in recent years.There are various types of malicious attacks and multiple types of malicious attacks may happen simultaneously in the real-world situations.However,most existing work focuses on improving the robustness of networks against a specific type of attack or ignores other conditions such as the cost of structural adjustment of networks.The multi-objective optimization problem is a common problem in scientific research and engineering practice.According to each objective with contradictory relationships,designing appropriate objective functions can solve the problem with a multi-objective optimization method.As one of many algorithms for solving multi-objective optimization problems,evolutionary algorithms have attracted extensive attention due to their parallelism,global search,and other advantages.Evolutionary algorithms have also become mainstream methods for solving multi-objective optimization problems.In this thesis,multi-objective evolutionary algorithm is used as a tool.An algorithm for simultaneously optimizing the node-robustness and the link-robustness of the network and an algorithm of robustnessoptimization based on minimum-cost are proposed.This thesis proposes two algorithms.One is an algorithm based on multi-objective evolution to simultaneously optimize the nodes-robustness and link-robustness of the network,which is based on a multi-objective evolutionary algorithm with non-dominated sorting.Appropriate genetic operations for the characteristics of the network structure and objective functions are designed based on nodes-robustness and link-robustness.The algorithm can obtain networks which have balanced nodes-robustness and link-robustness under the multiple types of attack.The optimization result is more suitable for the situation that the network is attacked in the real world.The other is a network robustness optimization algorithm based on minimum-cost.The purpose of this algorithm is to optimize the robustness of networks while constraining these cost of structure adjustment.The decomposition-based multi-objective evolutionary algorithm is used as a framework.The objective functions are designed based on the correlation between the minimum-cost and the robustness of the networks and the genetic operations with better effect are redesigned.The algorithm can improve robustness of networks with a lower cost of these structure adjustment.