基于均值漂移的关联关键设施网络抗毁性评估方法

A Mean-Shift Algorithm Based Evaluation Method for the Interdependent Infrastructure Networks

随着不同基础设施网络之间的关联性日渐增强,也使得其在遭受区域毁伤时更为脆弱.区别于传统的节点破坏方式,考虑伤害衰减的区域破坏方式会导致毁伤区域内的节点或边部分失效,继而对网络系统功能造成严重的影响.文章提出了基于区域毁伤方式下关联基础设施网络抗毁性评估模型,用于识别关联网络的脆弱区域并评估网络抗毁性,并设计一种基于均值漂移法的算法搜索关键区域,采用归一化的网络流指标来评估不同规模之间的网络抗毁性水平.最后,通过数值实验验证了算法的有效性,相比于遗传算法和现有的算法,文章所提出算法求解速度更快,也具有更好的可拓展性.

With the rapid development of the social economy,the connections between interdependent infrastructure networks increase,which makes the system vulnerable to regional damages.The regional damage under the damage decline rule is different from the node damage;the former may lead to failure of nodes and edges in the damaged area,and thus may cause a more serious impact on the system.This paper proposes an optimization model to analyze the invulnerability of an interdependent infrastructure network system under regional damages,which is used to identify the vulnerable areas and evaluate the network invulnerability.Then,a mean-shift algorithm is proposed to identify the vulnerable area,and a normalized network flow index is used to evaluate the network invulnerability level between different scales.Finally,the numerical studies show that the proposed method is faster than the genetic algorithm,and the performance of the solutions is also slightly better and more scalable.Therefore,the algorithm is not only suitable for complete damage rule but also damage decline rule,and can provide accurate visualization results for network managers,providing effective support for the invulnerability of large-scale interdependent infrastructure networks.