摘要
针对煤矿井下监测方法存在监测不准确、图片处理难等问题,采用卷积神经网络的理论方法,进行了图像清晰化算法优化研究。首先,搭建了一个智能化的精准监测系统;然后,对图像清晰度的算法进行了进一步的优化;最后,通过实验来验证其有效性,并对收集到的实验数据进行详尽的分析和讨论。研究结果表明,智能化的监测系统成功地整合了多种数据并进行管理,全方位地监控和评估了钻孔钻进的施工品质和工人的操作方法;提出了一种能够精准监测的图像清晰化新算法,该算法能够实现精准监测,从而自动识别钻杆数量、位置,监测工人的操作方式,并能对井下模糊图像进行清晰化处理。在算法设计中,使用了具有良好扩展性的生成器与判别器组合结构作为核心框架;在算法构建过程中,生成器和判别器都使用了基于卷积模式的组件。生成器能够在不同的分辨率下保存像素级别的详细信息,而判别器则通过重建技术来处理低频部分,并使用GAN方法来处理高频部分。研究有助于提高煤矿井下作业的安全性和经济效益,同时也能有效地提升瓦斯抽取的效率。
Aiming at the problems of inaccurate monitoring and difficult image processing in underground coal mine monitoring methods,the theoretical method of convolutional neural network was adopted to optimize the image clarity algorithm.Firstly,an intelligent and ac-curate monitoring system was constructed;then,the algorithm for image clarity was further optimized;finally,its effectiveness was veri-fied through experiments,and a detailed analysis and discussion were conducted on the collected experimental data.The results show that the intelligent monitoring system successfully integrates multiple data and managed them,comprehensively supervises and evaluates the construction quality of borehole drilling and the operation mode of workers.A new image clarity algorithm that can accurately monitor and automatically identify the number and position of drill pipes was proposed,as well as the operation mode of workers,and can per-form clarity processing on blurry images underground.In the algorithm design,a combination structure of generators and discriminators with good scalability was used as the core framework.In the process of algorithm construction,both the generator and discriminator use components based on convolutional patterns.The generator is able to preserve detailed pixel level information at different resolutions,while the discriminator processes the low-frequency part through reconstruction techniques and uses GAN methods to process the high-frequency part.This study contributes to improving the safety and economic benefits of underground coal mining operations,while also effectively enhancing the efficiency of gas extraction.
作者
刘德成
张文康
赵伟
曹阳
王涛
夏代林
彭洋
李敏
Liu Decheng;Zhang Wenkang;Zhao Wei;Cao Yang;Wang Tao;Xia Daiin;Peng Yang;Li Min(Chensilou Coal Mine,Henan Longyu Energy Co.,Ltd.,Yongcheng 476600,China;Wuhan Tianchen Weiye Geophysical Technology Co.,Ltd.,Wuhan 430205,China)
出处
《能源与环保》
2025年第8期173-178,187,共7页
CHINA ENERGY AND ENVIRONMENTAL PROTECTION
基金
国家自然科学基金项目(52374193)。
关键词
煤矿
智能化监测
图片清晰化算法
卷积
coal mine
intelligent monitoring
image clarity algorithm
convolution
