煤炭工业是我国社会经济发展和能源战略规划的重要支柱，在煤炭开采前，采用探地雷达技术对煤层中不良地质异常体进行精细探测，是确保煤矿安全高效生产的有力保障。空间步进频雷达是多发多收阵列体制和步进频信号相结合的超宽带探地雷达，在阵列接收端能够获取较大的带宽和孔径，具有良好的测距和测角性能。然而，实际矿井环境中存在严重的多路径效应和非高斯脉冲噪声，导致空间步进频雷达的探测性能急剧下降甚至失效。因此，本文针对多路径和脉冲噪声环境下空间步进频雷达的煤层异质体探测方法展开研究，主要研究内容如下：

1．针对煤层异质体目标，建立空间步进频雷达多路径接收信号模型，并计算出电磁波在煤层介质中的反射系数，此外，采用 稳定分布建立井下脉冲噪声信号模型。

2．针对多路径环境下空间步进频雷达对煤层异质体估计性能严重下降的问题，提出基于时间反演的二维稀疏渐进最小方差（TR-2D-SAMV）估计方法。通过对接收信号进行时反处理，消除多路径效应对回波信号的影响；再采用 2D-SAMV 算法搜索距离和角度进行迭代更新，获取目标距离-角度参数的精确估计值。实验结果表明，在低信噪比和少快拍的多路径环境下，该方法能够准确地分辨出直达信号和多径信号，对煤层异质体具有较高的估计精度。

3．针对井下脉冲噪声和多路径环境下空间步进频雷达对煤层异质体估计性能失效的问题，提出基于协方差矩阵的中值差分相关熵（MCBCM）算法，对强脉冲噪声进行有效抑制，并利用 TR-2D-SAMV 算法修正多路径效应造成的相位畸变，实现目标距离-角度参数的二维联合估计。实验结果表明，在低信噪比和少快拍的脉冲噪声环境下，所提的 MCBCM-TR-2D-SAMV 联合算法能够有效地抑制脉冲噪声，准确分辨出直达信号和多径信号，实现对煤层异质体的精确估计。

Coal industry is an important pillar of our country’s socioeconomic development and energy strategic planning, Before coal mining, the use of ground penetrating radar technology to finely detect unfavorable geological anomalies in coal seams is a powerful guarantee to ensure safe and efficient production of coal mines. The spatial stepped frequency radar is an ultra-wideband ground penetrating radar that combines the multi-transmit and multi-receive array system and the stepped frequency signal. It can obtain a large bandwidth and aperture at the receiving end of the array, and has good ranging and angle measurement performance. However, there are serious multi-path effects and non-Gaussian impulse noise in the actual mine environment, which lead to a sharp drop in the detection performance or even failure of the spatial stepped frequency radar. Therefore, this paper conducts research on the detection method of coal seam heterogeneous bodies under the environment of multi-path and impulse noise. The main research contents are as follows:

1. Aiming at the coal seam heterogeneous target, a multi-path receiving signal model of the spatial stepped frequency radar is established, and the reflection coefficient of the electromagnetic wave in the coal seam medium is calculated. In addition, thestable distribution is used to establish the downhole impulse noise signal model.

2. Aiming at the problem that the estimation performance of spatial stepped frequency radar for coal seam heterogeneity is seriously degraded in multi-path environment, a two-dimensional sparse asymptotic minimum variance (TR-2D-SAMV) estimation method based on time inversion is proposed. Through time-reverse processing of the received signal, the influence of multi-path effect on the echo signal is eliminated; then the 2D-SAMV algorithm is used to search the distance and angle for iterative update, and obtain the accurate estimated value of the target distance-angle parameter. The experimental results show that in the multipath environment with low signal-to-noise ratio and few snapshots, the method can accurately distinguish the direct signal and multipath signal, and has high estimation accuracy for coal seam heterogeneity.

3. Aiming at the problem that the estimation performance of spatial stepped frequency radar for coal seam heterogeneity fails in downhole impulse noise and multi-path environment, a Median Difference Correlation Entropy (MCBCM) algorithm based on covariance matrix is proposed to effectively suppress strong impulse noise, and use the TR-2D-SAMV algorithm to correct the phase distortion caused by the multi-path effect, and realize the two-dimensional joint estimation of the target distance-angle parameters. The experimental results show that the proposed MCBCM-TR-2D-SAMV joint algorithm can effectively suppress impulse noise, accurately distinguish direct signal and multipath signal, and realize accurate estimation of coal seam heterogeneity.