基于K-means分段融合模型的采煤沉陷水域水深反演  

Water depth inversion of coal mining subsidence waters based on K-means segmented fusion model

作　　者：付翔 刘潇鹏 刘昊 FU Xiang;LIU Xiaopeng;LIU Hao(School of Spatial Information and Surveying and Mapping Engineering,Anhui University of Technology,Huainan 232001,China;State Key Laboratory for Deep Coal Mining Response and Disaster Prevention and Control,Anhui University of Science and Technology,Huainan 232001,China)

机构地区：[1]安徽理工大学空间信息与测绘工程学院,安徽淮南232001 [2]安徽理工大学深部煤矿采动响应与灾害防控国家重点实验室,安徽淮南232001

出　　处：《黑龙江工程学院学报》2023年第3期14-19,共6页Journal of Heilongjiang Institute of Technology

基　　金：国家自然科学基金项目(41472323);安徽省重点研发项目(2022h11020024);淮南市科技计划项目(2021130)。

摘　　要：为了定量获取采煤沉陷水域水深数据,实现矿区水资源量的监测与合理利用,提出一种K-means分段融合水深反演模型。基于Sentinel-2影像数据与无人船实测水深数据分别建立单波段模型、多波段模型及其对应的水深分段融合反演模型,从整体精度和不同水深段精度等角度进行对比。结果表明:K-means分段融合反演模型的整体反演精度高于单一模型,在单波段和多波段基础上建立的分段融合模型较传统单一模型均方根误差分别降低了0.202 m、0.178 m,拟合优度分别提高了0.169、0.147;其在不同水深段内各项精度指标仍为最优,且随着沉陷水域深度增加反演精度也越高。In order to quantitatively obtain the water depth data of coal mining subsidence waters and realize the monitoring and rational utilization of water resources in mining areas,a K-means water depth segmentation fusion inversion model is proposed.Based on the Sentinel-2 image data and the measured water depth data of the unmanned ship,the water depth segmentation fusion inversion models corresponding to the single-band model and the multi-band model are established respectively,and the overall accuracy and the accuracy of different water depth segments are compared.The results show that the overall inversion accuracy of the water depth segmented fusion inversion model is higher than that of the single model.Compared with the traditional single model,the root mean square error of the segmented fusion model established on the basis of single band and multi band is reduced by 0.202 m and 0.178 m respectively,and the goodness of fit is increased by 0.169 and 0.147 respectively.The accuracy indexes in different water depth sections are still optimal,and the inversion accuracy is higher with the increase of subsidence water depth.

关 键 词：实测水深 水深反演 采煤沉陷水域 K-means分段融合模型 

分 类 号：X87[环境科学与工程—环境工程] TD327[矿业工程—矿井建设]