光纤监测系统入侵信号的检测和识别  

Detection and Identification of Intrusion Signals in Fiber Optic Monitoring System

作　　者：蒋雯雯[1] JIANG Wenwen(Qinghai College of Architectural Technology,Xining 810000,China)

机构地区：[1]青海建筑职业技术学院,青海西宁810000

出　　处：《云南师范大学学报（自然科学版）》2023年第4期41-45,共5页Journal of Yunnan Normal University:Natural Sciences Edition

基　　金：中国管理科学研究院教育科学研究所科研基金资助项目(JKSC2952).

摘　　要：为改善光纤监测系统监测信号特征复杂及入侵信号检测效率和识别精度低等问题,提出了一种改进的谱减法进行去噪处理,从而有效减少噪声干扰并获得具有高信噪比的监测信号;其次,由于光纤监测信号大部分为静默信号,因此对其进行端点检测从而有效提取入侵信号;最后,基于提取的入侵信号的时频域特征,提出了一种多特征融合深度学习识别模型,从而准确识别入侵类型.In order to improve the complexity of monitor signals′characteristics and low detection efficiency and recognition accuracy of intrusion signals in fiber optic monitoring system,an improved spectral subtraction method was proposed for denoising to effectively reduce noise interference and obtain monitor signals with high signal-to-noise ratio.Secondly,since most of the fiber optic monitor signals are silent signals,endpoint detection was performed to effectively extract intrusion signals.Finally,based on the time-frequency domain features of the intrusion signals extracted,a multi-feature fusion recognition deep learning model was proposed to accurately identify the intrusion signals.

关 键 词：光纤监测系统 谱减法 端点检测 入侵类型识别 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]