分布式光纤传感信号高识别率技术研究  

Research on high recognition rate technology of distributed optical fiber sensing signal

作　　者：张毅 徐方辰 戴丽娟 奚扬 ZHANG Yi;XU Fangchen;DAI Lijuan;XI Yang(West-East Gas Transmission Branch of National Oil and Natural Gas Pipeline Network Group Co.,Ltd.,Shanghai 200126,China)

机构地区：[1]国家石油天然气管网集团有限公司西气东输分公司,上海200126

出　　处：《智能计算机与应用》2022年第9期179-182,188,共5页Intelligent Computer and Applications

摘　　要：为提高基于相敏光时域反射计的分布式光纤声传感系统对振动信号的识别速度与识别准确率度,本文提出使用端点检测对光纤传感振动信号进行预识别的方法。该方式首先使用小波阈值折衷去噪对振动信号进行去噪处理,然后使用基于短时能量与短时过零率的端点检测方式对采集到的振动信号进行检测,如果检测到振动信号中存在有振动,则使用Resnet18网络对振动类型进行进一步的识别;若未检出振动信号,则将振动信号判定为噪音,不再进行进一步的检测。实验证明该方式能有效降低分布式光纤声传感系统对振动信号的识别时间,并且通过使用Resnet18网络对一维振动信号进行识别提高了对振动信号的识别准确率,对噪音信号的识别时间降低为3.5 ms,对振动信号的平均识别准确率达到96.3%。In order to improve the recognition speed and recognition accuracy of the vibration signal of the distributed optical fiber acoustic sensing system based on the phase sensitive optical time domain reflectometer, a method of pre-recognition of the optical fiber sensing vibration signal using endpoint detection is proposed. The method first uses wavelet threshold compromise denoising to denoise the vibration signal, and then detects the collected vibration signal using endpoint detection based on short-time energy and short-time excess zero rate. If vibration is detected in the vibration signal, the vibration type is further identified using the Resnet18 network;if there is no vibration signal detected, the vibration signal is judged as noise and no further detection is performed. The experiment proves that this method can effectively reduce the recognition time of vibration signals by distributed optical fiber acoustic sensing system, and the recognition accuracy of vibration signals is improved by using Resnet18 network to recognize one-dimensional vibration signals. Therefore, the recognition time of noise signals is reduced to 3.5 ms, and the average recognition accuracy of vibration signals reaches 96.3%.

关 键 词：分布式光纤声传感系统 端点检测 小波去噪 神经网络 

分 类 号：TP281.1[自动化与计算机技术—检测技术与自动化装置]