Research On Fault Location In Optical Network Based On Knowledge Graph

With the exponential expansion of Internet traffic and the continuous development of optical network technology,optical networks have unique advantages and development potential in carrying large-capacity data information and emerging business types.For optical transmission networks with complex structures,large capacity,and high bandwidth,a single source of failure will degrade the quality of service of large-scale transmission networks,and even lead to business interruption.The loss of a large number of business data will lead to economic losses and maintenance costs.Therefore,the survivability of the optical network is an important subject in the research of the optical network.Fault location is a prerequisite for dealing with optical network faults.The main purpose is to accurately and quickly identify the root cause of the alarm from the alarm storm,perform network unit-level fault location,and take corresponding fault recovery strategies.Knowledge graph is an important research topic in the field of knowledge engineering in recent years.It models general knowledge or domain knowledge into the structure of a knowledge network and assists machines to automatically reason about knowledge.Aiming at this topic,this paper models and reasoning about optical network faults by combining knowledge graphs,and constructs fault alarm knowledge graphs from top to bottom using equipment alarm manuals.On this basis,a fusion network topology based on path sorting PRA algorithm is proposed.Structured optical network fault location method,the specific research work is as follows:First,this paper constructs a knowledge graph in the field of optical network faults based on the equipment maintenance manual.First,this paper selects the commercial equipment manuals of Huawei OptiX RTN series equipment and Juniper BTI7800 series equipment as semi-structured data sources;secondly,according to the expert knowledge in the field of optical network faults,the ontology layer structure is constructed from top to bottom,and the fault ontology is predefined.And the alarm ontology structure,the relationship between faults and alarms and the derivative relationship of alarms are proposed.Then,a wrapper is designed to automatically extract semi-structured information.Finally,the Neo4j graph database is used to store the fault alarm knowledge graph and realize the visual output capability.The fault alarm knowledge graph contains 434 alarm entities and 670 fault entities.Second,this paper designs and implements an optical network fault simulation system,which takes into account the functions of resource modeling,simulating single outage of faults,and providing test data with clear transaction division.First,the system can choose NSFNET,CERNET and USNET for resource simulation to provide diversity of test data;secondly,the system can perform routing and resource allocation for business resources to realize resource planning simulation;then,it can simulate a single Fault information,including site receiving end or sending end failure,single break of optical cable,weakened optical cable transmission performance,equipment connection failure and other hard faults,transmit alarm information,and simulate the alarm reporting situation of the network management system;finally,design data acquisition adaptation The layer performs preprocessing and structured storage of alarm data to provide a data foundation for fault location.Third,this paper proposes an optical network fault location method based on the path sorting algorithm.Firstly,it is proposed that the research of fault location based on knowledge graph can be transformed into the path reasoning work of alarm sequence and fault entity;secondly,an instantiated knowledge graph is constructed by combining topology information and simulation training data,and a PRA algorithm optimized method is proposed according to the characteristics of fault simulation.Finally,experiments show that the fault location algorithm based on path sorting has guaranteed accuracy.The actual fault location is close to 70%of the candidate faults,and the proportion of the top 5 candidate faults is close to 95%.