Analysis And Research For Communication Network Management System's Real-time Faults

Current situations, characteristics and requirements of electric power communication network are introduced in this thesis, and communication network management system's real-time fault alarms are studied. The types, a variety of methods, technical difficulties and characteristics of fault alarms'relativity analysis are discussed. Three kind of classical sequential pattern mining algorithms are introduced and compared. A fault alarms correlation rule analysis model based on sequence clustering algorithm is designed. According to the flow of this model, the less and the more alarm data's correlation rule mining are implemented by analysis tools. Based on the correlation rules, a new alarm filtering method is presented, the practical application analysis is carried out, the validity and rationality of this method is verified, and this method could be used to aid the diagnosis, forecast and location of fault, improve the efficiency of handling the fault.