有向复杂网络的可控鲁棒性优化

Controllable Robustness Optimization of Directed Complex Networks

实际复杂网络总会不可避免的遭受到自然或人为的破坏,且遭受破坏的网络会发生故障连锁效应,使得网络的控制难度增加。基于复杂网络结构可控性理论,在经典有向ER和有向BA网络中,对部分负载边遭受攻击发生级联失效时的可控性进行分析。考虑到级联失效对可控性的影响以及网络构建与控制受成本的约束,提出将负载容量非线性模型引入到有向网络中,采用拥挤距离排序的多目标粒子群优化算法(MOPSO-CD)对模型参数进行优化,提高网络抵御级联失效的能力,随后基于负载容量线性模型和参数优化的负载容量非线性模型对网络可控鲁棒性进行了仿真对比,发现参数优化模型可以有效地抵御级联失效,提升网络可控鲁棒性,最后在实际交通网中验证了参数优化模型对网络可控鲁棒性优化的普适性和有效性。

Real complex networks are always suffering attacks from nature or human inevitably,and the network suffered attacks will be cascading failure,making the control difficulty of networks increased.Based on the theory of structural controllability,the analysis to network controllability when parts of the load edges suffer attacks and occur cascading failure is done in classical directed ER and directed BA networks.Considering the influence of cascading failure on the controllability and the establishment and control of networks bounded by the cost,the nonlinear model of load-capacity is introduced into the directed networks in this paper,and multi-objective particle swarm optimization algorithm based on crowding distance sorting strategy（MOPSO-CD）is adopted to optimize capacity parameters.Compared with the linear model of load-capacity,the simulation finds the nonlinear model of load-capacity that the parameters are optimized can resist the cascading failure effectively,and the network controllable robustness is improved.The universal applicability and effectiveness of the parameter optimization model to improve the network controllable robustness is verified in practical transportation networks at last.