Network fault diagnosis method, equipment and storage medium based on wavelet neural network

Abstract

The invention discloses a network fault diagnosis method based on wavelet neural network, the steps are as follows: S1, obtain network data under fault and normal state; S2, perform numerical and normalization processing on network fault data, and use PCA to reduce Dimensionality algorithm for data dimensionality reduction; S3, create a wavelet neural network model, select the improved gray wolf optimization algorithm, and use the optimized parameters as the parameters of the wavelet neural network model; then use the network fault data processed in step S2 as input, and reverse Add the momentum factor when adjusting the parameters, and complete the establishment of the network fault diagnosis model through continuous training; S4, input real-time network status data, and judge whether the network is faulty; S5, output the network fault diagnosis result and fault category. The invention introduces the momentum factor to improve the local optimization ability of the diagnosis model; adopts the improved gray wolf algorithm to optimize the initial parameters of the fault diagnosis model and avoid the randomness of initial parameter selection.

Description

technical field [0001] The invention relates to a network fault diagnosis method, equipment and storage medium, in particular to a network fault diagnosis method, equipment and storage medium based on wavelet neural network. Background technique [0002] During the operation of the network system, failure is a very worthy of attention. Even when a small failure occurs, it is very likely to affect the healthy operation of the entire network system, which may further lead to the paralysis of the entire system, causing economic and property damage. Loss. Therefore, it is of great significance to quickly and accurately discover various abnormalities in the network system through effective fault detection and diagnosis technology, locate faults, and restore faults to maintain the healthy operation of the network system. [0003] At present, fault diagnosis techniques are mainly divided into qualitative analysis methods and quantitative analysis methods. Among them, qualitative ...

AI Technical Summary

Problems solved by technology

However, the traditional wavelet neural network method is easy to fall into local optimum, which leads to the network fault diagnosis model not being optimal in terms of diagnosis accuracy.

In addition, the initial parameter selection of the WNN model is random, and the selection of different initial parameters may cause the model training to fall into different local optima, which will lead to fluctuations in the diagnosis results of network faults and poor diagnostic stability.

Images