RESEARCH BACKGROUND
In coal mining operations, the presence of geological anomalies such as faults, collapse columns, and thin coal seams not only poses safety hazards to the working face but also reduces the efficiency of coal seam extraction and increases equipment wear and tear. Therefore, it is of great importance to identify the internal structure of the working face, the spatial location of geological anomalies, and the physical characteristics of the strata before coal seam extraction. Currently, the main methods for detecting the internal structure of coal seam working faces include radio electromagnetic wave transillumination methods, channel wave detection methods, transient electromagnetic detection methods, and audio frequency electrical transillumination methods. Channel wave seismic exploration involves exciting and receiving seismic waves in the working face roadway, using seismic wave travel time or attenuation to detect structures or geological anomalies within the coal seam. It is currently a relatively effective and high-resolution underground exploration method.
In coal-bearing strata, the coal seam has lower density and wave velocity compared to the surrounding rock, forming a low-velocity interlayer. When a vibration is excited in the coal seam, some of the generated seismic waves are completely reflected and refracted back into the coal seam due to the difference in wave impedance between the surrounding rock and the coal seam, superimposing to form channel waves. Channel waves are a type of guided wave (confined wave or tube wave) that propagates in the coal seam. They have a long propagation distance, strong energy, easily identifiable waveform characteristics, and exhibit distinct dispersion characteristics. As a special type of surface wave, channel waves can be divided into Love-type channel waves (Love waves, L-waves) and Rayleigh-type channel waves (Rayleigh waves, R-waves). L-waves are generated when the shear wave velocity of the coal seam is less than the shear wave velocity of the surrounding rock above and below, while R-waves are generated when both the shear wave velocity and compressional wave velocity of the coal seam are less than the shear wave velocity of the surrounding rock above and below. Channel wave seismic exploration is generally divided into transmission wave methods and reflection wave methods.
Currently, the main seismic source for channel wave seismic exploration in coal mines is explosive sources. However, with the increasing demands for safety in coal mining, the use of explosive sources is subject to many restrictions in more and more mines. For example, some mines have a risk of coal and gas outbursts, and to avoid the high temperatures and sparks generated during blasting causing gas and coal dust explosions, explosive sources cannot be used. In addition, active-source channel wave exploration is expensive and labor-intensive. Therefore, a new type of channel wave exploration method that can be used in coal mines and does not rely on active explosive sources is needed. Although coal mining machines and CO2 sources have the potential to become seismic sources for channel wave exploration, they are currently still in the experimental research stage.
In 1957, Aki first pointed out that dispersion information of surface waves can be obtained from random continuous seismic recording signals, and proposed the theory and method of spatial autocorrelation (SPAC) imaging. Subsequently, research proved that the autocorrelation of transmitted seismic records received from the bottom at the free surface is equivalent to the simulated record of self-excitation and self-reception. This means that the cross-correlation of noise records between two points can recover the wave field excited at one point and received at the other. Weaver et al. found in the laboratory that the cross-correlation function of thermal noise data recorded at two points of an aluminum block was almost identical to the Green's function between these two points, and the structure of the aluminum block could be obtained using the resulting Green's function. The randomness of long-term noise records in space and time allows for obtaining information about the underground medium through cross-correlation. This method is now increasingly used to detect underground structures at different scales. For example, Shapiro et al. used background noise data recorded by a regional array in California to calculate surface wave dispersion information through cross-correlation and performed tomography. The results showed a significant correlation with the mountains and basins on the surface. Bensen et al. provided a comprehensive summary of the background noise tomography method process. Since then, background noise imaging methods based on seismic interferometry have developed rapidly.
Background noise imaging methods do not rely on active seismic sources, but instead utilize ubiquitous ambient noise to image subsurface structures. This method has been widely applied to imaging regional-scale faults, volcanoes, and groundwater structures, and is gradually developing from exploration to monitoring. At the engineering and mining scales, background noise imaging technology has been applied to the detection of potential locations of goaf areas and collapse columns in coal mines due to its high construction efficiency, strong resistance to vibration interference, lack of electromagnetic interference, non-destructive nature, and high resolution capabilities for layered strata and velocity anomalies. In high-gas coal mines, the shallow shear wave velocity structure of the coal mine is obtained based on the background noise imaging method, and the gas content distribution of the coal seam is estimated based on the empirical relationship between velocity and gas content, providing a new technical approach for coal mine gas content prediction research.
However, currently, the application of background noise imaging technology in coal mines is mainly limited to surface exploration.  Research on background noise in underground coal mines is relatively scarce. Wan Wentao et al., using joint seismic observations in an underground roadway at an elevation of -848 m and on the surface in the Pan-1 East mining area of ​​Huainan, found that in the high-frequency range greater than 1.0 Hz, the ground noise power spectral density (PSD) was 20-40 dB higher than that underground, and showed a diurnal pattern that varied with human activity; however, this time-varying pattern was absent in the underground observations. The reason for this is that the 870 m thick sedimentary layer above the underground roadway effectively attenuates shallow or surface anthropogenic noise. Hu Ze'an et al., through an experimental model of random temporal and spatial distribution of noise sources, realized the numerical simulation of natural source channel waves, and used the phase shift method to calculate the dispersion spectrum.  They also successfully collected three-component data of natural source channel waves in a field test at a working face of a coal mine in Huainan, Anhui, proving that the underground coal seam working face environment has the geophysical basis for the development of natural source channel waves.
However, currently, there are no research cases utilizing background noise to extract surface wave imaging for structural detection in underground coal mine working faces. To this end, the author applied background noise imaging technology to underground structural detection in coal mines for the first time, conducting imaging research on the internal structure of the working face.  Using a long-term monitoring array deployed in the underground roadway of the 22205 working face at the Huaning Coal Mine, dispersion data were obtained by extracting inter-station cross-correlation functions.  Subsequently, travel-time tomography was used to obtain the internal velocity structure of the working face. The low-velocity anomalies in the velocity results showed good agreement with the fault fracture zones and flexural structures revealed during mining.

ABSTRACT

The presence of geological anomalies such as faults, collapse columns, and thin coal seams in coal seams severely restricts the safe production of the working face. Therefore, it is necessary to detect and understand their internal structure in advance. Seismic wave exploration technology is widely used for imaging the internal structure of the working face and has high resolution. However, the application of explosive sources has limitations in specific scenarios.  Therefore, it is necessary to develop internal structure imaging techniques for the working face that do not rely on explosive sources. Compared with active source exploration technology, imaging technology based on ambient noise has significant advantages in structural detection because it does not require active excitation of seismic sources. The application of ambient noise surface wave tomography technology at the Huaning Coal Mine demonstrates that using vibration signals generated by underground machinery and mining activities, and filtering waveform data using diffuse wavefield indicators for cross-correlation function calculation, can reduce interference from non-stationary noise sources and obtain reliable surface wave dispersion data. Based on this, the surface wave one-step imaging technique is used to directly invert the three-dimensional velocity structure from the dispersion data, successfully obtaining the internal velocity structure of the working face and delineating the location of abnormal structures within the working face. The velocity imaging results show a significant correlation with the geological anomalies revealed by actual mining, verifying the feasibility and effectiveness of the ambient noise tomography method in working face structure detection. The diffuse wavefield filtering technology and surface wave one-step imaging process proposed in this study provide a new technical means for geological structure detection in coal mines. This method does not require an active seismic source, has the advantage of green exploration, and can provide technical support for safe and efficient mining of the working face.

Some charts

Figure 1: Schematic diagram of detector deployment locations