研究成果：Hou Y, Xing X, Li M, et al. Physica A: Statistical Mechanics and its Applications, 2017, 481: 160-166. (doi: 10.1016/j.physa.2017.04.039)

简介：

现实的系统，包括互联网、电力系统以及金融系统等，它们大都经历过罕见，但却规模极大的，由某些初始的小扰动所引发的超载级联失效。已有的复杂网络模型正在开展对于类似现象的研究。绝大多数复杂网络模型都是基于负载重新分配的过程，并且假设一旦某个节点失效，它的负载将被平均地或者根据某种偏好规则地转移到近邻的节点上。在实际的电力系统中，某个电站的超额负载往往倾向于分配给高负载容量的近邻电站。受此启发，我们在简易的沙堆模型基础之上，研究了一个异质的负载分配机制。我们发现，弱异质性的分配机制能够有效地缓解极限效应的发生，强异质性的机制可能会使得最终失效的规模增加。我们引入参数来刻画分配机制的异质性程度，并定义代表一个鲁棒的最优值。最后我们发现当初始沙粒分配是同质的情况下，会更大。

Abstract：

Many real systems including the Internet, power-grid and financial networks experience rare but large overload cascading failures triggered by small initial shocks. Many models on complex networks have been developed to investigate this phenomenon. Most of these models are based on the load redistribution process and assume that the load on a failed node shifts to nearby nodes in the networks either evenly or according to the load distribution rule before the cascade. Inspired by the fact that real power-grid tends to place the excess load on the nodes with high remaining capacities, we study a heterogeneous load redistribution mechanism in a simplified sandpile model in this paper. We find that weak heterogeneity in load redistribution can effectively mitigate the cascade while strong heterogeneity in load redistribution may even enlarge the size of the final failure. With a parameter  to control the degree of the redistribution heterogeneity, we identify a rather robust optimal . Finally, we find that  tends to shift to a larger value if the initial sand distribution is homogeneous.

原文链接：http://www.sciencedirect.com/science/article/pii/S0378437117303424