Study Of Fault Prediction And Fault Propagation In Coupled Networks

With the continuous development of science and technology,the information systems and urban infrastructures,such as transportation systems,social networks and power grid systems,are closely related to people's life and being huge.These systems can be modeled as complex networks with different structures.The systems have a great impact on people's lives and have attracted many scholars.With more analysis,researchers have found that these networks usually do not operate lonely,and that different networks interact and influence each other.For examples,railroad networks are interdependent with highway networks in transportation systems,and power networks are coupled with communication networks.Traditional studies adapted to a single network often have neglected some key information due to lack of inter-network factor analysis to biased results.We define coupled networks(or interdependent networks)as a pair of complex networks that are closely linked together,interacting and influencing each other.Coupled networks can make the system operate more efficiently as well as conveniently.However,coupled networks also bring some problems.For instance,when such an interdependent coupled network fails,it can cause serious consequences than a single network due to the fault cascade among the coupled networks.Therefore,fault management for coupled networks is of importance in the network safety backgrounds.In this study,fault prediction and fault propagation in coupled networks are investigated.Fault prediction,as the source of fault management,plays a very important role in the normal operation of the network.Meanwhile,fault propagation,as part critical section of fault management,poses the challenge for robustness of coupled networks.Therefore,the research on fault prediction and fault propagation in coupled networks has important theoretical value and practical significance.At present,most of the models or solutions for fault prediction are specific to single network structure.Moreover,these studies do not consider network properties,such as the electrical load,the attacked situation of the water network.More importantly,the impact analysis between the coupled networks is lacking in models and methods so that are not well fit for application.For fault propagation,different scholars have studied the fault propagation in coupled networks from various perspectives,including community structure,graph structure,etc.However,most of the studies have not mentioned the asymmetric dependence property of nodes in coupled networks.The asymmetric dependence property is not only relevant to the coupled network itself,but also has an important impact on fault propagation in the coupled network.Most of the studies ignore the influence of coupling model on fault propagation.The coupled networks has different results with the various coupled method.Consequently,so the study about coupling model on fault propagation is interesting.The main contents of the study are as follows.Based on the asymmetric dependence property,firstly,we investigates the effect of network robustness on fault propagation in coupled networks with different coupling methods.Specifically,we take the smart grid as an example,and build a many-to-many coupling network with contrast to one-to-one coupling method.On basis of the asymmetric dependence characteristics,we study the relationship between robustness and asymmetric dependency.We also explore the role of asymmetric dependency and robustness in fault propagation.The simulation experiments show that the coupled networks has better robustness with appropriate parameters of asymmetric dependence in the face of fault cascade.Based on collaborative filtering,we present online fault prediction algorithm in coupled networks.Specifically,firstly,a definition is proposed to measure whether two faults in a smart grid are correlated or not.In addition,a matrix decomposition is used to predict fault probability in smart grid by collaborative filtering.It is improved to propose a fault probability model that can be adapted to online real-time prediction.The proposed model is simulated by real data and simulated data together,and the experiments prove the effectiveness and accuracy of the proposed model.The fault prediction and fault propagation in the coupled network are explored.An online fault prediction model that can better accomplish the fault prediction in the coupled network is proposed.The experiments verified the effectiveness of the model.A new solution to the fault prediction of the coupled network is given.For the fault propagation,this paper focuses on the influence of the coupling model and the asymmetric dependence of the coupled network on the fault propagation.