采掘失调是煤矿安全生产的重大隐患之一，巷道快速掘进是保障采掘平衡的重要手段，矿井地球物理超前探测有效距离短，探测精度受井下复杂环境影响严重，制约巷道掘进速度，为解决探掘接续紧张矛盾，提出一种利用掘进工作面前方500 m长度定向钻孔开展探测工作的钻孔瞬变电磁探测方法，该方法可实现超长距离水害超前探测。在以钻孔钻进方向为Z轴正方向，以孔口所在平面右向为X轴正方向，下向为Y轴正方向的坐标系下，钻孔瞬变电磁水平分量异常响应形态均为“正弦曲线”或“反向正弦曲线”形态，通过形态组合判定出孔旁异常体所在象限，结合由水平分量异常场矢量合成得到的异常体中心偏转角度计算了异常体中心XOY平面旋转角。在此基础上，视每一个由垂直分量反演得到的电阻率为独立异常体，基于深度计算系数与电阻率的关系曲线推导出深度-电阻率与采样时间的映射关系，实现了电阻率与测点测道的一一对应，进一步采用K-means聚类算法智能定位其分布象限，计算了其XOY平面旋转角，再以定向钻孔轨迹为旋转轴，以测点所在孔位的倾角和方位角为旋转角进行坐标空间旋转，实现了非直钻孔径向的电阻率立体成像。计算了三维数值模拟和水槽物理模拟结果，对钻孔径向的异常体均取得了良好的立体成像效果。结合山西某煤矿井下的工程实践，对该方法远距离超前探测能力和精细立体解释掘进工作面前方地质异常体的性质、形态和规模的实用性和有效性进行了检验。研究表明:基于聚类的钻孔瞬变电磁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，此外，在掘进前开展定向钻孔中的长距离瞬变电磁水害立体超前探测，可有效保障巷道高效快速掘进。

Mining misalignment is one of the major hidden risks of coal mine safe production.Rapid roadway excavation is an important means to ensure the balance of mining and excavation.However,the effective distance of mine geophysical advance detection is short,and the detection accuracy is seriously affected by the complex underground environment,which restricts the speed of roadway excavation.In order to solve the contradiction between exploration and excavation,a long distance drilling transient electromagnetic detection method is proposed which uses directional drilling in front of the excavation work to carry out the detection work.This method can realize single 500 m advance detection on water hazard.In a coordinate system where the drilling direction of the borehole is the positive direction of the Z axis,the right direction of the plane where the orifice is located is the positive direction of the X axis,and the downward direction is the positive direction of the Y axis,the abnormal response patterns of the horizontal component of the drilling transient electromagnetic are all in the form of “sine curve” or “reverse sine curve” form.By the combination of shapes,the quadrant of the anomalous body near the hole can be determined,and the XOY plane rotation angle of the anomalous body is calculated by combining the center deflection angle of the anomalous body synthesized by the horizontal component abnormal field vector.On this basis,considering that each resistivity obtained from the vertical component inversion is an independent anomalous body,the mapping relationship between depth resistivity and sampling time is derived based on the relationship curve between depth calculation coefficient and resistivity,which realizes the one to one correspondence between resistivity and the measuring point and track.The K means clustering algorithm is further used to intelligently locate its distribution quadrant and calculates the XOY plane rotation angle.Then the directional drilling trajectory is used as the rotation axis,and the inclination and azimuth angle of the measuring point are used as the rotation angle to rotate the coordinate space,which realizes the resistivity stereo imaging in the radial direction of the non straight drilling.The three dimensional numerical simulation and the physical simulation results of the water tank are calculated,and the good stereo imaging effects are obtained for the anomalous body in the radial direction of the borehole.Combining with the engineering practice of a coal mine in Shaanxi province,the practicability and effectiveness of the nature,shape and scale of the geological anomaly in front of the heading work were tested by this method for long distance advanced detection ability and precise three dimensional interpretation.The results show that the cluster based transient electromagnetic stereo imaging method for boreholes is an organic combination of geophysics and machine learning.This method can provide a technical support for the advance detection and fine interpretation of hidden water hazards in underground heading faces.In addition,the three dimensional advance detection of long distance transient electromagnetic water hazard in the orientation borehole is carried out before heading,which can effectively ensure the efficient and rapid roadway excavation.