基于掘进机随掘震源的地震超前探测，以掘进机切割煤壁和岩石时产生的地震波作为震源，通过在巷道内长时间连续采集地震波并从中寻找反射波实现巷道超前探测。为了使用连续的随掘地震数据实现巷道前方的地质异常体成像，分析了随掘反射波成像的原理，将数据处理和成像的过程分成了5个主要步骤，首先将随掘数据划分为多个一定长度的时间片段并对每个时间片段进行分析，从中挑选出掘进机正常工作时间较长的片段，通过与参考巷道互相关获得脉冲化记录，再依据每个脉冲化记录中直达槽波的信噪比筛选出合格的脉冲化记录，最后，采用基于射线的绕射偏移法对巷道前方和侧帮区域进行成像。并在阳煤榆树坡矿1215工作面回风巷内开展了随掘探测试验，由布置在煤壁内的检波器采集多日的随掘地震数据，对数据进行了处理和成像。结果表明：将掘进机随掘震源地震数据脉冲化后采用绕射偏移法可以实现掘进面侧前方断层的成像；随掘地震数据中包含大量无效数据，评价数据质量并筛选数据是提高成像质量的重要步骤；直接对掘进工作面前方成像缺乏理论依据，由成像结果推断侧前方断层与巷道延伸线的交点可实现巷道正前方断层位置的预测。

The advanced seismic detection based on the seismic source of the roadheader use seismic wave generated by the cutting of coal wall and the falling of rock as the source of seismic and the advanced detection of the roadway is realized by continuously collecting seismic waves in the roadway for a long time and searching for reflected waves from it. In order to use continuous excavation seismic data to realize the imaging of the geological anomaly in front of the roadway, the principle of the excavation reflection wave imaging was analyzed, and the process of data processing and imaging was divided into five steps. Firstly, the excavation data was divided into many segments of a certain time length. Each segment was analyzed and selected according to the ratio of time in which roadheader was working well. The pulsed record was obtained through cross-correlation with the reference roadway, and then based on the direct groove wave in each pulsed record, the signal-to-noise ratio screens out qualified pulsed records. Finally, the ray-based diffraction migration method was used to image the front of the roadway and the side siding area. The excavation detection test was carried out in the return airway of No.1215 working face of Yushupo Mine of Yangmei Coal Group. The geophone arranged in the coal wall collected multi-day excavation seismic data, and the data was processed and imaged. The results show that pulsing the seismic data of the roadway machine source and using the diffraction migration method can realize the imaging of the front fault on the side of the roadway head. Moreover, the tunneling seismic data contains a large amount of invalid data and it is an important step to evaluate data quality and screen the data to improve imaging quality. There is no theoretical basis for directly imaging the front of the tunneling work. Inferring the intersection of the front fault and the extension line of the roadway from the imaging results can realize the prediction of the fault location in front of the roadway.