Research on alarm correlation based on dependency search tree in electric power communication network

Abstract

Aiming at the framework of electric power communication network and the characteristic of alarm data, this paper investigates the alarm correlation and fault location of electric power communication network. Taking into account dependency search tree which can decrease searching space and nodes of overlay, a method of fault location in electric power communication network based on dependency search tree is proposed. Alarms belonged to one class in alarm set are aggregated, and they are replaced by center node. In this way, large numbers of alarms are reduced and they can express the fault information accurately. Then greedy algorithm is repeatedly used to calculate residual alarms until the exact fault nodes are found out. This method based on the model of fault transmission is easy to build a model, and it can make use of the dependency of nodes directly and it is still applicable when many faults occur. The result of practical example proves that this method can analyze alarm information of electric power communication network and locate faults rapidly and exactly. So the faults can be found out in time, then the electric power communication network can be insured to operate steadily.