Network fault positioning method and device

Abstract

The invention discloses a network fault positioning method and a device. The method comprises steps of: dividing an original network domain into multiple sub network domains according to occurrence probability of different types of faults to a network node; acquiring history fault data in the network and obtaining a network fault positioning model through the neural network training; and acquiring information network fault data in the current network, inputting the information network fault data into the network fault positioning model and calculating to obtain communication network fault data so as to obtain a positioning result of the network faults. According to the invention, by considering dependency, between different layers, of the faults, the network domain with complex structure, high volume and many faults is divided into sub network domains with simplified structures, so more precise fault positioning can be performed; parallel positioning based on multiple sub network domains is performed, so fault positioning efficiency is improved; and by use of the neural network, the correlation between the network nodes is considered in the fault positioning, so accuracy of the network fault positioning is improved.

Description

technical field [0001] The invention relates to the technical field of communication networks, in particular to a network fault location method and device. Background technique [0002] At present, a typical method for joint fault location of the network is to use a bipartite graph model. The technical solution usually adopted by this method is to add false fault factors to the existing bipartite graph fault propagation model to establish an uncertain bipartite graph fault model; on the basis of the bipartite graph model, the fault location problem is transformed into a 0-1 The minimization problem of planning, and then use Lagrangian relaxation and subgradient methods to solve the problem, and finally rely on the Bayesian model for positioning. [0003] The above method is based on the bipartite graph model and relies on the Bayesian model for network fault location. Although it can also achieve network fault location to a certain extent, there are also some defects: [...

AI Technical Summary

Problems solved by technology

[0004] First, the positioning method using the bipartite graph model only starts with a single-level fault analysis, lacks consideration of the inter-layer dependencies of network faults, and performs poorly in accuracy;

[0005] Second, when the entire network is relatively large and complex, not only the convergence of the algorithm is difficult to be guaranteed

And the positioning efficiency needs to be improved

Images