基于增量式挖掘算法的电力通信网告警关联分析

Analysis of power communication network alarms correlation based on incremental mining algorithm

电力通信网络管理系统每天都会产生大量的告警信息,给实时网络监控和故障管理带来很大困难。告警关联分析是网络故障诊断中的重要任务,对通信网络的管理和维护具有十分重要的意义。序列模式挖掘作为一种时序数据分析的有效手段,能自动从告警中提取有助于关联分析的情景规则。研究了告警序列模式挖掘算法,并根据网络告警数据的增量式特征,在传统序列模式挖掘算法的基础上,为提高告警数据挖掘效率,提出了一种改变时间窗的增量式挖掘算法。通过实验证明了它的优越性。

Plenty of alarms are generated in power communication network management system every day, so serious problems are brought to network monitoring and fault management.Alarms correlation is an important task in network fault diagnosis and plays an important role in network management and network maintenance.As an effective way to analyze timed data sequential pattern mining can extract episode rules from alarms,which is helpful to analyze correlation rules.Alarms sequential pattern mining algorithms are studied in this paper.According to increment character of network alarms,this paper presents a variational time window incremental mining algorithm based on traditional sequential pattern mining algorithms to improve efficiency of data mining.And the superiority of this algorithm is indicated by the experiment.