摘要
提出了一种基于相位敏感光时域反射计Φ-OTDR的分布式光纤传感周界安防系统的扰动信号处理算法,先对信号进行降噪和加窗分帧处理,然后采用三层小波包分解实现信号能量特征提取,最终基于BP神经网络实现扰动信号分类.通过Matlab实验分析,本识别方法能够有效的区分风雨、电钻等机械振动、汽车碾压、人为攀爬不同类型的扰动事件,准确率分别达到85.7%,94.3%,91.4%和90%,系统稳定性高、误报率低,有效节约了人力资源成本,提高了工作效率.
A method of analyzing disturbance signal of distributed optical fiber sensing perimeter security system is proposed,which is based on phase-sensitive optical time domain reflectometry.Firstly,the signal is denoised and divided into frame signals,then characteristics of signal is extracted using three-level wavelet packet energy decomposition,finally,classification is realized based on BP neural network.The matlab experimental results show that the method can effectively identify kinds of disturbance events such as wind and rain,mechanical vibration,car crushing and artificial climbing,with accuracy 85.7%,94.3%,91.4%and 90%.The algorithm has high stability and low false alarm rate,which can effectively save the cost of human resources and improve the work efficiency.
作者
杜海龙
DU Hai-long(Electronic Information Engineering College,Zhengzhou Sias University,Zhengzhou 451150,China)
出处
《枣庄学院学报》
2020年第5期52-59,共8页
Journal of Zaozhuang University
基金
河南省科技厅科技攻关项目(项目编号:202102310630)
郑州西亚斯学院2020年科研项目(项目编号:2020-YB-76).
关键词
分布式光纤传感
特征提取
小波包
模式识别
distributed optical fiber
feature extraction
wavelet packet
pattern recognition
