Research On Fault Diagnosis And Analysis Method Of 5G Network Based On Knowledge Graph

With the development of network technology,5G has been widely used,making our life more convenient and intelligent.5G is characterized by high speed,low delay,high reliability,high connection density and wide coverage.The complexity and high integration of 5G networks have put forward higher requirements for network reliability and stability,so it is necessary to establish a more complete 5G network fault diagnosis and analysis mechanism to deal with the problems of 5G network faults.Traditional methods based on manual diagnosis and analysis of 5G network faults have many limitations.With the development of machine learning and deep learning,researchers have widely applied machine learning and deep learning algorithms in the field of fault diagnosis,which can make good use of network big data for network fault diagnosis.But models such as machine learning and deep learning are often black box models,which are difficult to explain decisions.Simple fault diagnosis and analysis based on machine learning can effectively solve the problem of traditional manual processing of big data,but the computer is still unable to make full use of expert experience,fault cases,employee logs and other existing knowledge.Therefore,in order to solve the above problems,this paper uses tools such as knowledge graph and machine learning to construct a method of 5G network fault diagnosis and analysis based on knowledge graph.By using the characteristics of strong interpretation of knowledge graph and efficient processing of big data by machine learning,this method is more perfect than previous methods.This thesis firstly proposes a method based on Bert-BiLSTM-CRF to extract structured knowledge and information from unstructured knowledge,such as 5G fault cases,so as to pave the way for the construction of 5G domain knowledge map.Through testing and comparison,the proposed knowledge extraction method is superior to the traditional methods based on machine learning and deep learning in terms of accuracy rate,recall rate and F1 value,and the overall accuracy rate of knowledge extraction can reach 89%.Then,this thesis analyzes the characteristics of the domain knowledge of 5G wireless communication network,defines the construction strategy of the domain knowledge map of 5G wireless network,constructs the industry knowledge map of 5G domain through ontology definition,knowledge extraction,knowledge storage and other steps,and stores it in the Neo4 j map database.Meanwhile,a knowledge query method based on Cypher language of Neo4 j graph database is proposed.The application scenario of knowledge graph in 5G wireless communication field is analyzed,and the intelligent retrieval method based on natural language questions and the intelligent retrieval method based on subgraph matching are proposed.The results show that the proposed method can make full use of the characteristics of knowledge graph and Neo4 j graph database,assist related staff to make efficient decisions,and greatly improve the engineering practicability of knowledge graph.Finally,based on knowledge graph and machine learning,a complete knowledge-and data-driven 5G network fault diagnosis and analysis method is constructed.Firstly,the traditional fault cause diagnosis problem based on machine learning is refined into continuous diagnosis sub-problems,and the algorithm model with the highest accuracy is matched for different sub-problems.It can be seen from the test results that the accuracy of fault cause diagnosis can be improved overall by refining problems.A complete 5G wireless network fault diagnosis and analysis model is built by combining the trained model with the knowledge graph,and the network big data difficult to understand is converted into text and graph data through the system model,so that relevant staff can deal with 5G network faults more accurately and quickly.