面向故障联合定位的告警数据信息填充算法

Research on alarm data information filling algorithm for fault joint location

数据完整性是网络告警信息的基本质量属性,是进一步进行网络告警故障分析的基础。然而在现实中,网管系统可能接收到信息缺失的网元告警信息,从而影响网络故障定位的准确性。分别使用决策树算法、窗关联规则挖掘(WARM)算法和相似度算法对网络告警信息中缺失的属性进行数据填充,并在国家电网信息网络和通信网络共存的场景下研究分析上述算法对联合故障定位性能的提升。实验结果表明,在该场景下决策树算法有更高的联合故障定位准确率。

The data integrity is the basic quality attribute of network alarm information and the foundation for further network alarm fault analysis.However,the network management system in reality can receive the network element alarm information with information loss,which influences the location accuracy of network fault.The decision tree algorithm,WARM algorithm and similarity algorithm are respectively used to perform data filling for the missing attribute in network alarm information.The performance promotion of the fault joint location with above algorithms is studied in the scenario of the co-existence of StateGrid information network and communication network.The experimental results show that the decision tree algorithm has higheraccuracy of fault joint location in the above scenario.