在图卷积网络(GCN)的帮助下，提出了一个更有效的搜索框架，以在有限的尝试下识别尽可能多的关键级联故障。通过离线训练一个GCN可以很好地捕捉级联故障的复杂机理。借助训练好的GCN模型，可以显著加快对临界级联故障的搜索。同时，通过分层相关传播算法实现了GCN模型的可解释性。结果表明，GCN导引的方法不仅可以加速临界级联故障的搜索，而且可以揭示潜在级联故障的预测原因。
2022-10-10 21:05:39
2.28MB
1
Complex Networks Toolbox for MatLab is designed to analyze large-scale graphs, model them, explore with simulations of dynamic processes and generate appealing and insightful layouts.
example：
function [BetweenneessCentrality, varargout]= GraphBetweennessCentrality(Graph,SourceNodes)
% Computes betweenneess centrality of each node.
%
% Receives:
% Graph - Graph Struct - the graph loaded with GraphLoad
% SourceNodes - array of double - (optional) nodes, from which passes start. Default: [] (all nodes).
%
% Returns:
% BetweenneessCentrality - array of double - Betweenneess Centrality for each node.
% Nodes - array of double - (optional)List of all nodes for which betweennessn centrality is computed
%
% Algorithm:
% http://www.boost.org/libs/graph/doc/betweenness_centrality.html
%
% See Also:
% mexGraphAllNodeShortestPasses
%
warning('Use the more optimized mexGraphBetweennessCentrality.dll');
error(nargchk(1,2,nargin));
error(nargoutchk(0,2,nargout));
if ~exist('SourceNodes') | isempty(SourceNodes)
SourceNodes = unique(Graph.Data(:,1));
end
Nodes = unique(Graph.Data(:,1:2));
%TotalPasses = zeros(GraphCountNumberOfNodes(Graph),GraphCountNumberOfNodes(Graph));
Betweenness = zeros(GraphCountNumberOfNodes(Graph),1);
for Node = Nodes(:).'
[ShortesPasses PassesHistogram]= mexGraphAllNodeShortestPasses(Graph,Node);
%TotalPasses = TotalPasses + sum(PassesHistogram(2:end));
tic
for i = 1 : numel(ShortesPasses)
%T = ShortesPasses(i).Passes(end);
%TotalPasses(Node,ShortesPasses(i).Passes(end)) = size(ShortesPasses(i).Passes,2); % compute total number of shortes passes from Node to some other node.
Passes = ShortesPasses(i).Passes(2:end-1,:);
NodesOnTheWay = unique(Passes);
if numel(NodesOnTheWay)==1
Count = 1; % hist behaves differently in this case.
else
Count = hist(Passes(:),NodesOnTheWay);
2022-04-26 20:46:45
24.47MB
1
为了揭示电网信息物理系统(power grid cyber-physical systems，PGCPS)中由信息攻击引发跨空间连锁故障的演化过程及爆发可能性，从攻击者视角提出了一种基于攻击损益原则的跨空间连锁故障选择排序方法。首先，分析攻击者在此类故障演化过程中的推动作用，构建故障演化过程的有向攻击图，将此类故障抽象为攻击路径；进而，综合考虑攻击代价与攻击收益等因素，提出各攻击路径的攻击损益值计算方法，并探讨不同条件下攻击者对各攻击路径的倾向性选择.排序，以推断不同跨空间连锁故障爆发的可能性；最后，在基于CEPRI-36 节点的PGCPS 仿真环境中，模拟攻击者对各类跨空间连锁故障的倾向性选择并予以仿真验证。
2021-12-29 05:37:44
836KB
1
复杂网络容量负荷模型MATLAB程序，原文用的美国西部电网uspowergrid，但是找不到相关文件，需要自己替换一个电网节点数据文件。文参考文献[Motter A E , Lai Y C . Cascade-based Attacks on Complex Networks[J]. Physical Review E, 2003, 66(6 Pt 2):065102.
2021-10-27 09:51:22
3.56MB
1