电荷耦合频响多光谱反演网络的煤矸石检测方法  

Gangue detection methods by charge-coupled frequency response multispectral inversion network

作　　者：赵栓峰[1] 王力 李小雨 邢志中 ZHAO Shuanfeng;WANG Li;LI Xiaoyu;XING Zhizhong(College of Mechanical Engineering,Xi’an University of Science and Technology,Xi’an 710054,China;School of Rehabilitation,Kunming Medical University,Kunming 650500,China)

机构地区：[1]西安科技大学机械工程学院,陕西西安710054 [2]昆明医科大学,康复学院,云南昆明650500

出　　处：《西安科技大学学报》2025年第1期117-128,共12页Journal of Xi’an University of Science and Technology

基　　金：国家重点研发计划项目(2017YFC0804310);陕西省重点研发计划项目(2020ZDLGY04-06);秦创原“科学家+工程师”队伍建设项目(2023KXJ-249)。

摘　　要：煤矸石作为煤炭开采的副产品,其处理与监测对环境保护和资源利用具有重要意义。传统的煤矸石检测方法存在效率低下、资源消耗大等问题。为解决这些问题,提出一种基于电荷耦合频响多光谱反演网络的煤矸石检测方法。该方法利用无人机搭载高分辨率相机拍摄煤矸堆图像,首先,通过构建光谱重建网络模型MST++,实现对光谱数据的高效重建,弥补了RGB图像在光谱信息上的不足;然后,利用Faster R-CNN模型处理重建后的光谱数据;最后,对矸石堆中的煤炭含量进行预测。结果表明:与原始RGB数据集相比,多光谱重建后的煤矸石与煤炭图像的检测准确率显著提升至91.2%,将真实含碳量与预测结果进行比对,两者误差控制在5%以内,为煤炭资源的回收利用提供了科学依据。该方法充分利用了无人机遥感技术与深度学习算法的优势,通过光谱重建与目标检测的有机结合,实现了煤矸石的精准识别与定量分析,为煤矸石的高效处理与资源化利用提供了新的技术路径,具有良好的应用前景。As a by-product of coal mining,the treatment and monitoring of coal gangue is of great significance to environmental protection and resource utilisation.The traditional coal gangue detection methods have problems such as low efficiency and high resource consumption.For these,a coal gangue detection method based on charge-coupled frequency response multispectral inversion network is proposed.A UAV carrying a high-resolution camera was used to take images of the coal gangue.Firstly,by constructing a spectral reconstruction network model MST++,an efficient reconstruction of the spectral data was achieved,which makes up for the lack of spectral information in the RGB image.Secondly,the reconstructed spectral data were processed using the Faster R-CNN model,and finally the coal content in the gangue pile was also predicted.The results show that compared with the original RGB dataset,the detection accuracy of the multispectral reconstructed gangue and coal images is significantly improved by 91.2%,and the error between the real carbon content and the prediction results is controlled by less than 5%,which provides a scientific basis for the recycling of coal resources.The method makes full use of the advantages of UAV remote sensing technology and deep learning algorithms,with the accurate identification and quantitative analysis of coal gangue achieved through the organic combination of spectral reconstruction and target detection.The research provides a new technical path for efficient treatment and resourceful use of coal gangue,and has good application prospects.

关 键 词：光谱重建 多光谱图像 卷积神经网络 目标检测 煤矸石 

分 类 号：TD849[矿业工程—煤矿开采]