Research on Fault Identification Method of Power System Communication Network Based on Deep Learning

Abstract

With the communication network scale, the increasing bandwidth and complexity of the constant improvement of the quality of network service, and user requirements, an urgent need to intelligent communication system of the current high speed communication network for effective and reliable management, and fault management is becoming more difficult and important than ever, when the network produces a fault or failure, Many thousands of alarms are generated in a short period of time, so analyzing the signals of these alarms becomes more complicated. Some existing alarm analysis systems have some shortcomings, such as poor scalability, difficulty in dealing with complex situations, and lack of learning ability. This paper proposes a method of fault identification and alarm correlation analysis based on deep learning algorithm. Combined with deep reinforcement learning technology, a sleep scheduling strategy based on multi-level is designed to reduce energy consumption, and its effectiveness is verified by simulation. Experimental results show THAT this method can overcome the limitations of common alarm correlation analysis methods, and create favorable conditions for improving the efficient utilization of spectrum resources in private networks.