随掘地震技术是巷道智能化快速掘进发展急需的地质保障技术之一，由于随掘震源与炸药等主动震源差别较大，常规的认知与数据处理技术已经无法取得效果。针对目前随掘震源认知不足，缺乏有效随掘地震数据正演模拟方法的问题，开展了随掘震源产生机制分析与震源时间函数构建研究。研究发现：随掘地震数据是具有一定延续时间的复杂、变频、连续信号，其相对低频信号主要与综掘机的自身机械动作、随机落煤以及转载机、带式输送机运作等有关，相对高频强能量信号由截割破煤决定，总体表现为一种多源叠加的伪随机信号特征；其次，在线性排列数据采集模式下，随掘震源可近似为不同时延下连续激发的多源叠加而成的综合虚拟点震源，该震源由综掘机械截割煤层时的震动以及煤体破裂、随机落煤、转载、打钻、输送带运输等联合作用而成，其中截割煤层起主要作用。基于震源产生机制分析，构建了随掘地震震源时间函数，并以该时间函数作为震源加载项，开展了随掘地震数值模型试验，得到的随掘地震模拟数据与实测数据具有相同的时间域、频率域以及时频域特征，表明了构建出的随掘地震震源时间函数可有效进行随掘地震数据正演模拟，提升了随掘震源的认知程度。另外，为验证基于脉冲反褶积与互相关的脉冲化算法在随掘地震数据处理中的有效性，以2种常用地震干涉技术为对比，利用随掘地震模拟数据进行了脉冲化处理实验。试验结果表明：基于脉冲反褶积与互相关的脉冲化处理结果与相同模拟条件下采用常规Ricker子波模拟得到的地震记录波场特征更为相近，验证了脉冲化算法的有效性，为随掘地震技术的深入研究与应用提供了基础。

The seismic while driving technology is one of the most urgently needed geological guarantee technologies for the intelligent rapid driving development of roadways. Due to the large difference between the seismic while driving source and active sources such as explosives, conventional cognitive and data processing technologies have failed to achieve results. In view of the lack of cognitive of the source of seismic while driving and the lack of effective forward simulation methods for seismic while driving data, the research on the generation mechanism analysis and source time function construction of seismic while driving source were carried out. It is found that: the seismic while driving signal is a complex, variable frequency, continuous signal with a certain duration. Its relative low-frequency signal is mainly related to the mechanical rotation of the cutting head, the action of the gangue raking machine, the random coal falling, and the operation of the loader and belt conveyor, its high frequency and strong energy signals are determined by cutting coal, generally characterized by a pseudo random signal superimposed by multiple sources. Secondly, under the linear array data acquisition mode, the seismic while driving source can be approximated to a comprehensive virtual point source formed by the superposition of multiple sources that are continuously fired at different time delays. The seismic source is formed by the combined action of the vibration when the fully mechanized driving machine is cutting the coal seam and the coal body fracture, random coal falling, refuse raking, drilling, belt transportation, etc., in which the coal seam cutting plays a major role. Then, based on the analysis of source generation mechanism, the time function of the seismic while driving source is constructed, and the numerical model experiment of the seismic while driving is carried out with the time function as the source loading item. The simulated seismic while driving data obtained has the same time domain, frequency domain and time-frequency domain characteristics as the measured seismic while driving data, which show that the source time function constructed can effectively carry out forward modeling of seismic data while driving, and improve the cognition of the source while driving. In addition, in order to verify the effectiveness of the impulse algorithm based on spike deconvolution and cross-correlation in seismic while driving data processing, taking two common seismic interference technologies as comparison methods, effective signal extraction experiments were conducted using the simulation seismic while driving data. The experimental results show that: the impulse processing results based on spike deconvolution and cross-correlation are more similar to the wave field characteristics of seismic records obtained by using conventional Ricker wavelet simulation under the same simulation conditions, which verifies the effectiveness of the impulse algorithm. The research in this paper provides a basis for the further research and application of seismic while driving technology.