基于聚类算法的钻孔瞬变电磁视电阻率立体成像方法

煤矿井下地球物理水害超前探测要求探测点20 m范围内不得有积水和金属物体,传统电磁法超前探测技术已不能满足要求,钻孔瞬变电磁法通过将收发装置送入掘进工作面前方的钻孔中进行探测,既远离了巷道中的各种干扰,又提高了隐蔽致灾水体的探测精度。为解决该方法对钻孔径向异常体的准确定位解释难题,通过三维正演总结了其水平分量异常响应特征,提出了异常体象限确定准则,研究了根据水平分量幅值和异常象限综合求取异常体工具面角的计算方法。将由垂直分量计算得到的每一个视电阻率视为独立异常体,基于K-means聚类算法对相应的水平分量异常曲线特征值进行二分类,实现了全数据集的视电阻率象限自动划分,结合异常工具面角算法研究得出钻孔瞬变电磁视电阻率立体成像方法。最后计算了三维数值模型的立体成像结果,对钻孔径向的小规模低阻异常体取得了良好效果。结果表明:基于K-means聚类算法的钻孔瞬变电磁视电阻率立体成像方法是地球物理与机器学习的有机结合,该方法能够为井下掘进工作面隐伏水害超前探测精细解释提供技术支撑。

Stereo Imaging Method of Borehole Transient Electromagnetic Apparent Resistivity Based on Clustering Algorithm

The advanced detection of geophysical water hazards in coal mines requires no water or metal objects within 20 m of the detection point.However,the traditional electromagnetic advanced detection technology has not met the requirements.Borehole transient electromagnetic method carries out detection by sending the transceiver into the borehole in front of the tunneling face,which is not only away from all kinds of interference in the roadway,but also improves the detection accuracy of hidden disaster-causing water bodies.In order to solve the problem of accurate positioning and interpretation of the radial anomalous body of the borehole by this method,the horizontal component anomaly response characteristics were summarized through three-dimensional forward modeling,the criterion for determining the quadrant of the anomalous body was proposed,and the method for calculating the tool face angle of the abnormal center based on the horizontal component amplitude and the abnormal quadrant was also studied.Each apparent resistivity calculated from the vertical component was regarded as an independent anomaly.Based on K-means clustering algorithm,the corresponding eigenvalues of the horizontal component anomaly curve were classified into two categories,which realize the automatic division of the apparent resistivity quadrant of the entire data set.Combined with the study on abnormal tool face angle algorithm,the stereo imaging method of borehole transient electromagnetic apparent resistivity was obtained.Finally,the stereo imaging results of three-dimensional numerical model were calculated,and good results were obtained for small-scale low-resistance anomalies in the radial of borehole.The result shows that stereo imaging method of borehole transient electromagnetic apparent resistivity based on K-means clustering algorithm is an organic combination of geophysics and machine learning.The method can provide technical support for fine interpretation of advanced detection of concealed water hazards in underground tunneling face.