Research On Fault Diagnosis Technology Of Complex Information Network Considering Cascading Failure

The increasing complexity of information networks has brought huge challenges to the operation of network management systems.The core problem of network management is fault management,and faults are inevitable in large-scale information networks.Therefore,real-time diagnosis of faults is essential for the normal operation of the network.However,the complexity of the information network structure,the dynamic nature of the traffic environment,the uncertainty of the observed information,and the diversity of types of faults increase the difficulty of fault diagnosis.In order to reduce the impact of network failures on its communication services and ensure the operating efficiency of the network,two issues need to be resolved.The first is how to obtain accurate status information during the dynamic changes of the network environment.The second is how to formulate efficient and general fault location solutions to quickly find and locate faults.In response to the above questions,the content of this paper are as follows: 1.Based on the topology of the network,the importance of the node is evaluated from different angles,and the protective measures for the node are determined by judging the role of each node in the network.The traditional cascading failure model only considers the topology of the network,and the application environment of the model is not analyzed.In this paper,a cascading failure model based on the network self-recovery function is established based on the information network equipment attributes and the change trend of the network state,and then discusses the degree of impact of different models on network performance.Finally,the relationship between the shortest path length and the information transmission delay is analyzed,and the effect of information delay on the efficiency of network service processing is illustrated through simulation experiments.2.Considering the complexity of the information network and the diversity of equipment failure types,the symptoms corresponding to each failure type are counted,a neutrosophic set fault diagnosis method is introduced to diagnose the failure type,and proposed a new method of calculating the weight vector.The classification algorithm in machine learning technology is used to classify the symptom data of known fault types,so as to locate the fault node corresponding to the symptom,and compare the performance of different classification algorithms to obtain an algorithm suitable for network fault location.The limitations of machine learning technology in fault location are analyzed,and then Bayesian inference method is introduced and combined with machine learning technology to jointly use for network fault location.Finally,the performance of different fault diagnosis schemes is compared through simulation experiments.3.From the analysis results of network failure types,it can be seen that most node failure types are overloaded.This type of failure is usually caused by cascading failures.It is necessary to trace the source of cascading failures based on these overloaded nodes.Based on the cascading failure model,a combination of redistribution load calculation and Bayesian inference is used to locate the cascading failure source node,thus providing a theoretical basis for formulating a fault recovery strategy.Finally,by analyzing the change trend of the network topology and its traffic environment,the fault recovery scheme is introduced from different angles.