电力信息与通信技术2025，Vol.23Issue(7)：66-72,7.DOI:10.16543/j.2095-641x.electric.power.ict.2025.07.09
异构多源告警日志分类算法研究
Research on Heterogeneous Multi Source Alarm Log Classification Algorithm
摘要
Abstract
With the increasing complexity of power data communication networks,the accurate classification and processing of heterogeneous alarm logs from devices of different manufacturers has become particularly important.This paper presents a method for classifying alarm logs using an improved TF-IDF algorithm combined with a CNN algorithm,effectively classifying and standardizing heterogeneous alarm logs.By considering the position of words and their distribution across categories,the model's ability to capture important text features is enhanced.Utilizing a CNN classifier,the method effectively identifies alarm information from diverse sources.Experimental results show that this algorithm can effectively process and classify heterogeneous alarm logs,achieving an average accuracy improvement of 4.1%and an F1 score increase of 3.8%compared to commonly used algorithms.Additionally,the algorithm can generate a structured alarm database,providing strong technical support for the intelligent operation and maintenance of power data communication networks.关键词
告警日志/TF-IDF算法/CNN算法/电力数据通信网Key words
alarm logs/TF-IDF algorithm/CNN algorithm/power data communication network分类
信息技术与安全科学引用本文复制引用
施志伟,苏云霞,周广丽,费周辰,宋长平,杨强强..异构多源告警日志分类算法研究[J].电力信息与通信技术,2025,23(7):66-72,7.基金项目
国网上海市电力公司科技项目"面向电力数据通信网监控工具的研究及应用"(52090F240004). （52090F240004）
