煤矿工作面先验约束加奇点模型CT重构方法研究  

Research on CT inversion adopting prior constraints and singularity model for a longwall panel

作　　者：杨真 武祥[1] 姚文理 郭昌放 

机构地区：[1]中国矿业大学矿业工程学院煤炭资源与安全开采国家重点实验室,江苏徐州221116

出　　处：《中国矿业大学学报》2017年第5期988-996,共9页Journal of China University of Mining & Technology

基　　金：国家自然科学基金青年基金项目(51004102)

摘　　要：在利用CT层析成像方法重构回采工作面内地质构造等异常区分布时,受回采工作面客观条件影响,投影角度有限,只能采用一边发射信号对边接收的方法,导致投影数据不完备,系数矩阵高度稀疏,从而无法精确重构出工作面内部的地质构造及其分布.针对该问题,本文提出一种新的基于总变分正则化的先验约束加奇点模型重构算法,通过加入回采工作面两巷揭露的地质信息作为先验约束条件,同时对特定区域引入奇点模型以提高反演精度,锐化断层影响区域的边界.经过数值计算和回采工作面现场试验,证明该算法能显著提高重构图像的精度,锐化断层影响区域的范围,显著改善异常区的识别效果.When reconstructing the distribution of abnormal areas such as geological structure in the working face of coal mine with CT tomography,the projection angle is limited due to the influence of objective conditions of coal mining face,so the signals can only be transmitted at one side and received at the opposite side.The geological structure and its distribution in the working face cannot be accurately reconstructed due to the incomplete projection data and highly sparse coefficient matrix.Based on prior constraints,a reconstruction algorithm of total variation regularization was proposed.The boundary of fault zone is sharpened by adding the geological information revealed in two roadways of coal mining face as the condition of prior constraint while introducing the singularities to specific areas to improve the inversion accuracy.The results of numerical calculation and field test show that this algorithm can significantly enhance the reconstructed precision,sharpen the range affected by faults,and improve the recognition effect of abnormal areas.

关 键 词：不完备投影 先验约束 总变分正则化 奇点模型 

分 类 号：TD163[矿业工程—矿山地质测量]