An Intelligent Alarm Method for Optical Fiber Network Based on Backtracking Single Alarm Information

Abstract

Optical fiber transmission network has gradually become the mainstream communication transmission medium and the development of artificial intel-ligence provides new ideas for the prediction of fiber transmission network failures. Different from traditional method which using the alarm infor-mation only for troubleshooting after a fault has occurred, this paper applies the real-time and serialized characteristics of alarm information to predict fault, achieving that a single alarm message can be used to indicate the pres-ence or absence of a fault. The specific steps of this paper are as follows, first extracting the alarm feature by collecting the alarm information and malfunction information from optical fiber network, then building a smart alarm learner by using the machine learning algorithm, finally achieving the goal to predict the probability of a failure occurring in the future. This meth-od can be integrated into the optical fiber transmission network operation, and lay the foundation for further research on Intelligent alarm.