Research On Fault Diagnosis Method Of Mobile Communication Network Based On Big Data Mining

With the development of mobile communication networks,the future network will not exist alone or only use a single technology,but the coexistence and complementarity of multiple technologies and common development.In this trend of network isomerization and densification,how to efficiently detect and diagnose network faults has become a huge challenge.For the traditional network fault detection and diagnosis algorithm research,in the process of fault detection and diagnosis,we need to rely on massive data sets,but in actual situations,it is too expensive to manually label historical data sets and historical data sets are too high.There are fewer and how to efficiently use the data set for fault detection and diagnosis.In response to these problems,this paper has conducted an in-depth study of the intelligent fault diagnosis algorithm for mobile communication networks.Based on the introduction of future network development trends and existing network fault diagnosis methods,this paper mainly conducts the following three aspects of research:(1)Aiming at how to improve the effect of network fault diagnosis in the case of small samples,a fault detection and diagnosis method for heterogeneous wireless networks based on the idea of Generative Adversarial Network(GAN)is proposed.First,it analyzes the sources of common network failures in the heterogeneous wireless network environment,and obtains a large number of reliable data sets by generating a small amount of data based on network failures based on the idea of adversarial network.Then,based on these data,use the e Xtreme Gradient Boosting(XGBoost)algorithm to select the optimal feature combination of the input parameters in the fault detection stage,and complete the fault diagnosis and prediction.The simulation results show that the algorithm can achieve more accurate and efficient fault detection and diagnosis of the network in the case of small samples.(2)Aiming at how to avoid the errors generated by the generative model and improve the robustness,universality and accuracy of the network fault diagnosis model,based on the real network environment,a network fault detection and diagnosis algorithm based on AWGAN-GP(Average Wasserstein GAN with Gradient Penalty)is proposed.First,the fault data of the original database is preprocessed,that is,the data is standardized first,and then by setting a feature weight threshold,only samples with a sample weight greater than the set threshold are selected to achieve the effect of feature screening.Based on the filtered data,the data is used in the GAN frame to perform data fitting under different network states,so as to obtain a large number of labeled virtual data under different network states.In addition,it is used as the test data in the fault detection and diagnosis model for model testing.In addition,the virtual data generated by GAN is used as the training data for fault detection and diagnosis model training.The results show that this algorithm can make the fault diagnosis model more robust,universal and accurate.(3)This paper innovatively combines deep learning and reinforcement learning in the field of network fault detection and diagnosis,and proposes a network fault detection and diagnosis algorithm based on deep Q learning.This algorithm realizes that in the face of complex network structure leading to complex network failures,it can still use fewer features to classify some obvious network states.Simulation results show that compared with traditional algorithms,this algorithm can achieve effective fault detection and diagnosis.