In view of the high false alarm rate of network alarm data and the diffi culty of predicting new network attack behaviors

this paper proposes a network attack prediction method based on the characteristics of alarm events. First

FP

t

ree is used to mine association rules of alarm events and eliminate false alarm. Then

the sequence pattern is used to extract the characteristics of network attacks

and the combination rules of network attacks are formed to realize the effective early warning of network attacks. Through the analysis of the related experimental process and results

it is shown that the proposed method can effectively and real-time warn the attack behavior of the network

and has certain applicability and expansibility.