基于ResNet-UNet模型的DAS矸石漿體充填堵管監(jiān)測技術(shù)

打開文本圖片集

中圖分類號：TD528.1 文獻(xiàn)標(biāo)志碼：A 文章編號：1672-9315（2025）04-0650-13

DOI：10. 13800/j. cnki. xakjdxxb.2025.0402

DAS gangue slurry filling pipe blockage monitoring technology based on ResNet-UNet model

CHAI Jing1，2， WANG Ziming'，MA Chenyang'， ZHANG Dingding1，2，LI Zhi 1 ZHOU Sen'，QIU Fengqi，3， WU Yuyi3，JI Wenli12，ZHAO Pengxiang4 （1. College of Energyand Mining Engineering，Xi’an Universityof Science and Technology， Xi'an 710o54，China; 2. Key Laboratory of Western Mine Exploitation and Hazard Prevention，Ministry of Education， Xi'an University of Science and Technology，Xi'an 710o54， China; 3.China Coal Energy Research Institute Co.， Ltd.，Xi’an 710o54，China;

4. College of Safety Science and Engineering， Xi'an University of Science and Technology，Xi'an 71Oo54， China）

Abstract： The coal gangue slurry transportation pipeline is prone to issues such as blockage and corrosion during the transportation process.At present， precise positioning still faces huge challenges in addressing the blockage problem in slurry pipeline transportation. A method combining image noise reduction with a ResNet-UNet composite network was proposed for monitoring and identifying blockage points ，using Distributed Acoustic Sensing as the monitoring technique; In order to evaluate the proposed solution，a 15.14 meter ring pipe model was constructed，and a grouting blockage simulation test was conducted.The results demonstrate that： Compared with traditional UNet and ResNet models，the ResNet-UNet network can accurately identify blockage point images while effectively mitigating the issue of gradient explosion， the blockage location accuracy reaches 97. 83% ，with a precision of 97.76% ，a recall rate of 94.80% ，and an F1 score of O.958 9.This study successfully addresses the noise processing challenges associated with the high sensitivity of DAS-based monitoring， significantly improving the accuracy of blockage point positioning within the scope of comprehensive coal gangue pipeline monitoring，it provides an intelligent and precise solution for monitoring coal gangue slurry transportation pipelines and identifying blockage points.

Key words： Distributed Acoustic Sensing; gangue slurry pipeline transportation; noise reduction algorithms; ResNet-UNet model; image recognition; blockage localization

0 引言

在煤炭等礦產(chǎn)資源開采過程中，大量矸石廢棄物的高效運(yùn)輸和管理是實(shí)現(xiàn)綠色開采的重要環(huán)節(jié)[1-3]。（剩余17635字）