基于Φ-OTDR系统的声波信号管道监测方法  

Acoustic signal pipeline monitoring method based on Φ-OTDR system

作　　者：张旭[1] ZHANG Xu(Wuhan Research Institute of Posts and Telecommunication,Wuhan 430074,China)

机构地区：[1]武汉邮电科学研究院,湖北武汉430074

出　　处：《电子设计工程》2022年第2期82-86,共5页Electronic Design Engineering

摘　　要：为了准确地掌握管道线路的运行状态,保障油气管道的安全运行,在基于相位敏感光时域反射(Φ-OTDR)原理的光纤分布式振动系统的基础上,提出了一种泄漏声波信号监测方法。介绍了Φ-OTDR系统的结构和工作原理。针对管道周围环境复杂与噪音强的问题,提出一种新型小波阈值算法对信噪进行降噪处理。选用梅尔倒谱系数(Mel Frequency Cepstral Coefficents,MFCC)作为声波信号的特征向量,建立BP(Back Propagation)神经网络识别模型完成管道泄漏识别。实验结果表明,文中提出的BP神经网络泄漏识别方法有较好的识别率,且经过新型小波阈值函数算法降噪后,其平均识别率比降噪前提高了26.74%,最高识别率达到91.1%,具有一定的应用潜力。In order to accurately grasp the operation status of pipeline and ensure the safe operation of oil and gas pipeline,based on the optical fiber distributed vibration system on the foundation of the phase sensitive Optical time domain refleection(Φ-OTDR) principle,a leakage acoustic signal monitoring method is proposed. The structure and working principle of Φ-OTDR system are introduced. Aiming at the problem of complex environment and strong noise around the pipeline,a new wavelet threshold algorithm is proposed to denoise the signal-to-noise. Using Mel Frequency Cepstral Coefficients(MFCC)as feature vector,BP neural network recognition model is established to complete pipeline leakage recognition. The experimental results show that the BP neural network leakage recognition method proposed in this paper has a good recognition rate,and after the new wavelet threshold function algorithm denoising,the average recognition rate is 26.74% higher than before,and the highest recognition rate reaches 91.1%,which has a certain application potential.

关 键 词：管道泄漏检测 小波降噪 MFCC BP神经网络 信号识别 

分 类 号：TN912.34[电子电信—通信与信息系统]