Diffraction imaging is beneficial for characterizing the distribution of small-scale structures. Traditional diffraction imaging methods highlight diffractions by applying various energy suppression techniques within migrated dip-angle gathers to attenuate reflections in the Fresnel zone. However, these methods not only suppress reflected waves but also remove diffracted waves within the Fresnel zone, which may degrade the focusing of discontinuities in the diffraction image. We develop a diffraction imaging technique using morphological component analysis (MCA). We first use high-resolution linear Radon transform (LRT) to focus diffracted waves, thereby better preserving them during separation. In the Radon domain, diffracted waves appear as point-like features, whereas reflected waves exhibit curved events. Accordingly, we employ the stationary wavelet transform (SWT) and seislet transform as sparse representation dictionaries for diffracted and reflected waves, respectively. Based on these two sparse representation dictionaries, MCA can effectively extract the diffraction component. The proposed method is a non-Fresnel-zone-muting diffraction imaging technique that better preserves the diffracted waves within the Fresnel zone, thereby improving the focusing of discontinuities. Tests on the Sigsbee 2B model demonstrated that the proposed method effectively removed reflected waves while preserving diffracted waves within the Fresnel zone. Applications to field data indicate the effectiveness of the technique in highlighting deep-seated high-angle fractures, facilitating the detailed characterization of discontinuous features.

A stable electrode/electrolyte interface is essential for ion transport and durability in solid oxide electrolysis cells (SOECs). However, interfacial delamination remains a major cause of performance degradation. In this study, we integrated electrochemical impedance spectroscopy (EIS) with numerical modeling to investigate air electrode delamination, focusing on the location, area, and spatial distribution of the delamination. Distinct features in the distribution of relaxation times (DRT) were observed, resulting from controlled delamination at the YSZ-GDC and GDC-LSCF interfaces. This enabled further analysis of the in-situ characteristics of the delamination sites. A 6 mm diameter delamination at the GDC-LSCF interface increased polarization resistance (RP) from 0.421 to 1.139 Omega & sdot;cm2 at 700 degrees C. The same defect in an LSCF/GDC composite electrode induced about half of this increase, indicating enhanced resilience. Furthermore, distributing the equal total delamination area into multiple smaller defects significantly mitigated performance loss, reducing RP to 0.734 Omega & sdot;cm2. This critical dependence on spatial distribution, which previous models overlooked, was successfully captured by introducing tortuosity correction factors into our simulations. This study quantitatively clarifies how key characteristics of delamination govern electrochemical performance. It provides mechanistic insights into failure pathways and offers practical guidance for designing robust SOEC electrodes.

Ground fissures induced by coal mining activities pose severe risks such as coal fires, water inrush, and environmental degradation. Efficient identification of these fissures is critical for disaster prevention and mining safety. However, the lack of publicly training data tailored to mining scenarios hinders progress in deep-learning-based fissure identification. This article presents the ground fissures in mining areas (GFM) dataset, the first publicly available high-resolution uncrewed aerial vehicle (UAV) image dataset for ground fissure segmentation in coal mining areas, comprising 7368 annotated images across diverse landscapes and introducing a structured fissure categorization scheme. Furthermore, we propose FS-YOLO, a specialized instance segmentation framework that incorporates the dynamic snake convolutional pyramid pooling and deformable large kernel attention modules to enhance learning multiscale morphological features for ground fissure segmentation. Experiments indicate that FS-YOLO outperforms YOLOv8 by 3.5% and 6.5% in box AP@0.5 and box AP@0.95, and by 4.8% and 3.6% in mask AP@0.5 and mask AP@0.95, respectively, and achieves competitive results against other state-of-the-art methods.

Automated subsurface utility detection systems in construction rely heavily on the quality of ground-penetrating radar (GPR) profiles, which are often degraded by high-amplitude horizontal interference. Existing low-rank decomposition methods lack the intelligence and flexibility required for multi-site data processing and involve labor-intensive parameter tuning, impeding their integration into intelligent construction workflows. To address these challenges, this paper proposes a horizontal interference suppression algorithm based on a diffusion model, termed GPR-HIDiff. The proposed model replaces conventional sequential convolutional operators with ResBlocks throughout the encoder, intermediate layer, and decoder of the UNet architecture, enhancing training stability. Lightweight agent attention modules are embedded between ResBlocks at each level to improve global information modeling capability. A spatial attention mechanism is deployed between the encoder and decoder to achieve adaptive spatial feature optimization. Furthermore, the forward diffusion phase adopts a cos θ schedule-based strategy to ensure a smooth temporal variation of noise variance. A standardized dataset comprising real-world measured samples and finite difference time domain simulation samples of urban road models has also been constructed. The effectiveness of the hybrid dataset, the introduced modules, the robustness analysis, and the cos θ schedule is validated through training with single/mixed datasets, ablation studies, evaluation of metric variations before and after the introduction of different noise levels, and comparative experiments with constant, linear, and cos θ schedules. Experimental results demonstrate that GPR-HIDiff significantly outperforms both traditional methods and state-of-the-art deep learning models on both simulated and real-world test samples. It effectively suppresses horizontal artifacts, preserves target hyperbolic contours, and avoids excessive reduction of target scattering, showcasing its exceptional performance. This method provides a powerful algorithmic foundation for high-resolution GPR imaging and target detection.

During the interpretation of Borehole Radar (BHR) B-scan profiles, the accurate determination of the azimuth of geological targets in three-dimensional space is a critical issue for achieving precise anomaly localization and spatial structure inversion. However, existing directional BHR anomaly localization methods exhibit limited intelligence, insufficient adaptability to multi-site data, and weak generalization capability, rendering them inadequate for engineering applications under complex geological conditions. To address these challenges, a robust deep learning model, termed BSS-Pose-BHR, is developed based on YOLOv11n-pose for keypoint detection in directional BHR profiles. The model incorporates three key optimizations: Bi-Level Routing Attention (BRA) replaces Multi-Head Self-Attention (MHSA) in the backbone to improve computational efficiency; Conv_SAMWS enhances keypoint-related feature weighting in the backbone and neck; and Spatial and Channel Reconstruction Convolution (SCConv) is integrated into the detection head to reduce redundancy and strengthen local feature extraction, thereby improving suitability for keypoint detection tasks. In addition, a three-dimensional electromagnetic model of limestone containing a certain density of clay particles is established to construct a simulation dataset. On the simulated test set, compared with current mainstream deep learning approaches and conventional directional borehole radar anomaly localization algorithms, BSS-Pose-BHR achieves superior performance, with an mAP50(B) of 0.9686, an mAP50–95(B) of 0.7712, an mAP50(P) of 0.9951, and an mAP50–95(P) of 0.9952. Ablation experiments demonstrate that each proposed module contributes significantly to performance improvement. Compared with the baseline, BSS-Pose-BHR improves mAP50(B) by 5.39% and mAP50(P) by 0.86%, while increasing model weight by only 1.05 MB, thereby achieving a reasonable trade-off between detection accuracy and complexity. Furthermore, indoor physical model experiments validate the effectiveness of the method on measured data. Robustness experiments under different Peak Signal-to-Noise Ratio (PSNR) conditions and varying missing-trace rates indicate that BSS-Pose-BHR maintains high detection accuracy under moderate noise and data loss, demonstrating strong engineering applicability and practical value.

Material degradation at the interconnect-oxygen-electrode interface induced by interconnect-side anodic polarization poses a critical challenge to the practical application of solid oxide electrolysis cell (SOEC) stacks. Herein, spinel-coated SUS441 samples were tested at 800 degrees C in air with current flowing from the substrate to electrodes to mimic real-stack conditions. Spinel coating protection and material degradation were systematically studied. Results show MnCo-based spinels remain chemically stable under current loading, while Cucontaining spinels readily react with Cr3+ to form Cr-rich phases. Anodic polarization greatly accelerates Cr3+ migration toward oxygen electrodes, and its rate rises with current density. Fast Cr3+ migration worsens Cr2O3 oxidation resistance and causes severe Cr poisoning of perovskite electrodes. Applied current generally aggravates interfacial degradation. Strategies including optimizing coating thickness and oxygen partial pressure, lowering temperature, and building Cr-perovskite barrier layers are proposed to restrain oxygen permeation and Cr3+ migration for long-term SOEC operation.

Clogging caused by fluid-particle interactions in rock fractures significantly impacts transmissivity and fluid transport efficiency. This study employs a coupled LBM-DEM numerical method to investigate particle clogging mechanisms and their effects on fracture transmissivity. Key factors, including fracture surface roughness, particle characteristics, and flow dynamics, are comprehensively analyzed. Results reveal that increased roughness intensifies fluid-particle interactions, leading predominantly to single-particle clogging (SPC). Conversely, fractures with lower roughness exhibit a combination of agglomerated-particle clogging (APC) and SPC. Furthermore, the occurrence of SPC decreases with increasing particle size and particle number, as well as lower Reynolds numbers. Additionally, the evolution of inter-particle contact forces shows exhibits enhanced anisotropy with increased particle size, particle number, and fluid inertial effects, which in turn govern the stability and morphology of the clogging structures. To bridge the gap between microscopic mechanisms and macroscopic behavior, a modified local cubic law incorporating clogging effects is proposed to quantitatively assess transmissivity reductions caused by clogging, demonstrating strong effectiveness across diverse scenarios. Collectively, these findings elucidate the fundamental dynamics governing fracture clogging and contribute to more accurate predictions and management strategies in geological and engineering practices.

During the identification of shallow coal-rock layers using ground penetrating radar (GPR), inversion of the air and coal thicknesses by GPR response is the key to achieving intelligent control over the orientation of the Shearer drum and GPR. The relationship between the offset GPR response and the real thicknesses of complicated shallow layers is uncertain, and there is random noise present in the environment. A novel method based on an artificial neural network (ANN) is proposed to invert the real thicknesses from the apparent thicknesses picked up from the GPR response. First, a relationship model between the real thicknesses and the air-coupled GPR apparent thicknesses of the complex double layers is established through qualitative analysis. Then, two ANN architecture schemes are proposed for inverting the real thicknesses of air and coal, respectively. Afterward, the ANN models are trained by simulation data and are then improved with regard to network input, network structure, and dataset size. The robustness of this approach is then examined under various trace interval and signal-to-noise ratio (SNR) scenarios. Finally, the proposed method is applied to physical model experiments and contrasted with traditional methods. The results show that the proposed method significantly improves the accuracy of the air and coal thicknesses measured by air-coupled GPR. The maximum and average errors of the air thicknesses are 4.97 and 1.15 mm, respectively, with those of the coal thicknesses of 6.04 and 1.60 mm. This method can achieve high-precision measurement of shallow-layer thicknesses, supporting coal mining automation.

Underground 3D mapping plays a vital role in mineral resource exploration and development. Lidar-based simultaneous localization and mapping (SLAM) has become a key technology in this domain, offering autonomous navigation and real-time mapping capabilities. However, in narrow and geometrically repetitive underground mine corridors-characterized by long passages and a lack of loop closures-conventional lidar-SLAMsystems suffer from significant error accumulation, leading to degraded mapping accuracy. To overcome these limitations, we propose a novel lidar-SLAMmethod incorporating connecting traverse constraints. First, edge and planar features are extracted from lidar scans via a segmentation strategy and refined using least-squares fitting. These features are then integrated into a factor graph framework to optimize pose estimation and construct a consistent point cloud map. Furthermore, we introduce an external traverse connection mechanism to impose additional constraints in the pose graph, effectively correcting trajectory drift and 3D point coordinates. Our approach mitigates the effect of noise and error accumulation commonly encountered in traditional lidar-SLAM techniques. Evaluation on a self-collected data set and two public benchmarks demonstrates that the proposed method consistently outperforms four state-of-the-art systems: LOAM, LeGO-LOAM, S-LOAM, and F-LOAM. In our self-collected data set, the trajectory root mean square error (RMSE) values for LOAM, LeGO-LOAM, S-LOAM, F-LOAM, and the proposed method are 0.693, 0.506, 4.062, 2.542, and 0.467 m, respectively. Compared to these baseline methods, our approach achieves error reductions of 32.6%, 7.7%, 88.5%, and 81.6%, respectively. On the two public data sets, the trajectory RMSE values for the same methods are 1.132, 0.351, 5.848, 8.850, and 0.084 m and 0.612, 0.590, 3.671, 9.279, and 0.558 m. This corresponds to relative error reductions of 88.6%, 67.2%, 97.5%, and 98.5% and of 8.8%, 5.4%, 84.8%, and 94.0%, respectively, confirming its robustness and accuracy in challenging underground environments.

Seismic inversion quantitatively extracts reservoir properties from seismic data, which has gained increasing attention in assisting the exploration and evaluation of deep coalbed methane (DCBM) reservoirs. However, the accuracy of seismic prediction is limited because the DCBM reservoirs exhibit complex pore geometries governed by a dual-porosity system. To address this limitation, the study presents a dual-porosity parameter seismic inversion method based on decoupled equivalent medium theory. A dual-porosity rock physics model is constructed and then decoupled to derive a linear forward operator that links matrix porosity, crack porosity and crack aspect ratio to the corresponding elastic parameters. To account for lithological variability, a Gaussian mixture model is employed to describe the joint prior probability distribution of dual-porosity parameters. Well-log data are applied to invert matrix porosity, crack porosity and crack aspect ratio, which serve as prior constraints in the iterative Bayesian inversion framework, thereby enhancing the stability and accuracy of the forward operator. By explicitly treating dual-porosity parameters as inversion targets, the proposed method effectively captures the spatial heterogeneity of pore geometries in DCBM reservoirs. Borehole-side synthetic seismic gather validation results demonstrate that the proposed approach significantly enhances inversion accuracy compared with conventional equivalent-porosity inversion methods. The application to pre-stack seismic data demonstrates the ability of the method to capture the dual-porosity geometry.

Tight sandstone reservoirs, characterized by low porosity and permeability, present substantial potential for CO2 utilization and sequestration. Understanding the mechanical behavior of tight sandstone under the influence of CO2 is critical for assessing geological CO2 sequestration and the CO2 fracturing capabilities of reservoirs. As the burial depth of the target reservoir increases, the formation temperature gradually rises, considerably altering the mechanical properties of reservoir sandstone, especially the interaction between CO2 and sandstone. However, few studies on the coupled effects are available. In this study, we built a high-temperature CO2 soaking system that allows CO2 injection across various formation temperatures (25 degrees C-160 degrees C). Uniaxial compression tests were conducted to explore the mechanical properties of the tight sandstone subjected to CO2 soaking at various temperatures. Scanning electron microscopy (SEM), X-ray diffraction (XRD), and energy-dispersive spectroscopy were used to quantitatively characterize the evolution of the mineral composition and micromorphology of tight sandstone after CO2 soaking at different temperatures. A stress-strain damage constitutive model was established to describe the behavior of tight sandstone under the coupled effects of temperature and CO2. The quantitative relationships between mineral dissolution, pore-throat evolution, and crack propagation revealed by SEM, XRD, and EDS analyses were integrated into the model to describe the prepeak damage evolution and deformation characteristics. The proposed model not only effectively characterized the macroscopic mechanical behavior of tight sandstone under coupled temperature-CO2 conditions but also provides a mechanistic explanation for the role of microstructural changes in controlling damage accumulation and strength degradation.

Surface wave dispersion imaging plays a critical role in near-surface shear-wave velocity characterization. Conventional methods, such as multiple signal classification (MUSIC), suffer from critical limitations, including covariance matrix rank deficiency under strong noise, high computational cost, and poor resolution in the low-frequency band. To address these issues, we propose a high-resolution dispersion energy imaging method based on the fast iterative shrinkage-thresholding algorithm (FISTA). By formulating dispersion imaging as a sparse inversion problem, the proposed approach achieves enhanced resolution with improved computational efficiency. Synthetic and field data experiments demonstrate that the proposed method is robust to noise and highly computationally efficient. Improved imaging performance facilitates more reliable dispersion curve picking, clearer mode separation, and more stable shear wave velocity inversion, which are essential for practical multichannel analyses of surface waves.

In order to alleviate the problems of restoration success rate and ecosystem resilience under the condition of water shortage in the dump area in western China,the Zhunneng Heidaigou open-pit mine was selected as the study area,and different soil layer reconstruc-tion methods(three-layer structure and mixed structure)were used in the dump site,and different microbial combinations(AM,DSE,AM+DSE,CK)were combined for ecological reconstruction to study the water retention,water culvertability,and spatial distribution of water and salt in the reconstructed soil layer.Effects of microorganisms on plant growth and water use strategies.The results show that the ground penetrating radar verifies the physical characteristics of the three-layer structure and the mixed structure,the soil moisture of the three-layer structure is distributed in layers,the surface loess layer(ecological layer)increases with depth,the middle coarse sandstone lay-er(culvert layer)has a balanced water distribution and the moisture content reaches 12％,the bottom sandy clay layer(aquifer)has the function of blocking water infiltration,and there is no spatial difference in the water distribution of the mixed structure.The water reten-tion capacity of the three-layer structure treatment increased by 43.1％,and significantly increased the water absorption of 50-100 cm soil layer,thereby promoting the water utilization of plant roots,and the utilization efficiency reached 71.1％.Compared with CK,the water use efficiency of alfalfa was increased by 42.7％and the coverage was increased by 1.2 times.Therefore,the combination of three-layer structure and AM+DSE treatment has significant ecological restoration potential in the arid and semi-arid mining areas in western China,and this study provides new ideas and technical support for the ecological restoration of mining areas,which is conducive to the sustain-able development of mining areas in western China.

Low-speed mining significantly expands the range of mining-induced stress. When faults and confined aquifers exist in the floor, extensive stress disturbance increases the risk of overall fault activation and induced water inrush disasters. This research applies surface microseismic technology to a deep coal mine, systematically analyzing the failure responses of the floor controlled by faults and confined water. Based on this, the mechanism of floor water inrush is revealed, and a safety factor for quantitatively characterizing water inrush risk is proposed, along with a framework for assessing water inrush risk. Research findings indicate: (1) Microseismic events in the floor primarily occur within aquifer, with their spatial distribution controlled by fault strike and dip. (2) Energy propagates forward along the fault strike, while plastic deformation develops upward from the floor aquifer along the fault planes. (3) The events during fault activation exhibit similar focal mechanisms, primarily characterized by reverse fault type with negative ISO and CLVD components. The primary failure types are shear-compression failure and pure compression failure. (4) F11 fault exhibits high-frequency, small-scale activation characteristics. F12 and F13 fault display low-frequency, large-scale activation features. The proportions of positive stress drop for the three faults are 58.3%, 51.5%, and 64.9%, respectively. (5) The source radiation patterns of the F12 and F13 faults are similar in morphology but exhibit spatial orientation rotation. This study provides guidance for the assessment and early warning of floor water inrush disasters during low-speed mining.

The transparency of geological condition detection and precise exploration are critical challenges hindering safe and efficient coal mining operations. Coalfield seismic exploration technology plays a vital role as a means to address hidden, potentially hazardous geological issues during the mining process. This technology provides high-precision regional geological structures and coal formation patterns, offering reliable geological basis for coalfield development. The evolution of coalfield seismic exploration technology in China can be divided into three distinct phases: the early stage from the 1950s to the early 1970s marked by technological inception; the digital development era spanning the late 1970s through the 1980s; and the current phase since the 1990s, characterized by extensive promotion and practical application. Over the course of more than seventy years, this technology has made significant advancements in coalfield geological surveys and the precise detection of hidden factors contributing to geological hazards. It stands as an indispensable geological support for ensuring safe and efficient coal development. By integrating in-depth research on coalfield exploration techniques with typical engineering practices, the text systematically outlines its technical features and current status across the stages of data acquisition, processing, and interpretation. In response to the urgent need for transparent and intelligent mine construction, future developments in coalfield exploration technology will focus on advancing dense distributed data acquisition, intelligent data processing and interpretation, and the innovative application and development of multi-attribute integration techniques.

Abstract Ground penetrating radar (GPR) is a powerful and non-destructive sensing technique that has become increasingly important for monitoring urban road infrastructure. By enabling rapid and accurate acquisition of subsurface information, GPR supports the scientific management and safe operation of underground utilities. However, as urban subsurface environments become more complex, conventional GPR interpretation methods—often based on manual analysis or rule-based algorithms—struggle to maintain high detection accuracy, robustness, and automation. Recent advances in deep learning have opened new opportunities for intelligent GPR data analysis, significantly improving both the efficiency and reliability of underground object detection. This paper provides a comprehensive review of deep learning applications in GPR-based detection of underground targets within urban road environments. It first summarizes typical GPR signal characteristics and data representations for common subsurface objects, followed by a detailed discussion of deep learning techniques employed for object recognition, localization, and classification. Related studies on subsurface parameter inversion are also reviewed to enhance understanding of target geometry and physical properties. Finally, key challenges and future research directions are outlined. This review aims to offer systematic insights into the integration of deep learning with GPR for intelligent underground sensing and to promote its development in urban infrastructure monitoring.

Water-rich goaf constitutes a primary hazard factor in coal mines, potentially triggering mine instability and ground subsidence. To detect water-rich goaf efficiently, economically, and non-destructively, the extended spatial autocorrelation (ESPAC) method was applied to perform microtremor surveys in the goaf of the mid-deep sections of the Renjiazhuang coal mine in Ningxia. Microtremor signals were obtained using a linear array, and the 2D distribution of subsurface shear wave velocity was inverted and verified against borehole data. The results showed that in the depth range of -200 to 600 m, the low-velocity zone (500-1500 m/s) is closely related to the fissure development and water-rich area, revealing the spatial distribution of hidden disaster-causing factors in the subsurface. Three low-velocity anomalies within the profile were successfully identified by microtremor probing and combined with borehole drainage validation, confirmed the presence of standing water within these anomalous areas. Enhanced application scope and depth of ESPAC methodology in coal mine water-rich goaf detection have been achieved, establishing comprehensive technical support and theoretical frameworks for subsequent water-rich zone risk assessments and mine safety monitoring systems.

Distributed Acoustic Sensing (DAS) is an emerging seismic acquisition technology that offers high spatial sampling density, continuous recording capability, and flexible deployment. These characteristics make it particularly suitable for shallow subsurface exploration in urban environments. However, DAS exhibits limited sensitivity to weak seismic signals and typically captures only the axial component of ground motion. In contrast, conventional geophones offer high signal fidelity and multi-component recordings, but are limited by lower spatial resolution, greater deployment costs, and reduced adaptability in complex terrain conditions. To enhance surface wavefields reconstructed from DAS ambient noise, we propose a method that integrates the complementary strengths of DAS and conventional geophones. The method employs a linear array configuration, where the vertical component of a geophone deployed at the front of the array serves as a virtual source, and the DAS system is deployed along the subsequent positions of the array to serve as receivers. These recordings are combined to form a hybrid ambient noise dataset, which is processed in the frequency domain through normalization and cross-correlation. The acausal parts of the cross-correlation functions (CCFs) are taken as virtual shot gathers (VSGs). This method not only preserves the high spatial resolution of DAS but also incorporates the high signal fidelity of geophones, thereby significantly enhancing the surface wavefields in passive surface wave imaging. By evaluating the signal-to-noise ratio (SNR) of randomly selected traces from the CCFs obtained under various ambient noise stacking durations, the proposed method achieves average SNR improvements of 1.54 dB and 1.00 dB compared to the case where both the virtual source and receivers are derived from DAS data. Under the optimal stacking duration, the extracted dispersion energy shows clearer and more concentrated patterns, with an extended frequency range.

Ground penetrating radar (GPR), as an efficient non-destructive testing technique, plays a crucial role in the structural condition assessment and defect identification of railway ballast. Typical defects such as mud pumping generally exhibit characteristics in B-scan images including weak reflections, blurred boundaries, and irregular structures, which pose significant challenges for stable detection and precise localization using existing methods that rely primarily on spatial feature modeling. Most current deep learning approaches focus on modeling spatial or temporal information, while lacking effective utilization of frequency-domain features, thereby limiting their discriminative capability under complex electromagnetic environments. To address these issues, this paper proposes a single-stage object detection framework, termed YOLO-DGW, based on time-frequency collaborative modeling. Built upon YOLOv8, the proposed method introduces a structure-aware spatial enhancement module to improve the representation of continuous GPR echo structures. Meanwhile, frequency-domain information is incorporated as a modulation prior to guide spatial feature learning, enhancing the model’s sensitivity to weak reflections and complex-shaped targets. In addition, A-CIoU loss function is designed to improve localization accuracy and stability for defect regions of varying scales. Experimental results demonstrate that YOLO-DGW achieves an F1-score of 63.06% and an AP@0.50 of 62.07%, representing improvements of approximately 7.41% and 2.8%, respectively, over the strongest baseline method. Compared with several mainstream object detection models, the proposed approach exhibits superior performance in both detection accuracy and cross-region generalization capability. These findings indicate that integrating frequency-domain information into spatial feature learning through a modulation mechanism can effectively enhance the model’s ability to discriminate weak-reflection anomalies, providing a novel time-frequency collaborative modeling paradigm for railway GPR defect detection.

Faults and periodic weighting pose a serious threat to the stability of the surrounding rock in deep coal mine roadways. To reveal the evolution of failure in the surrounding rock in the deep cross-fault roadway during excavation support and periodic weighting, this research employs a systematic analysis integrating physical model testing and numerical simulation. First, we built a physical model replicating field conditions, with digital image correlation (DIC), strain, and acoustic emission (AE) systems to monitor displacement, strain, and rock failure. Then, a numerical model of the actual mining area was established to further investigate the response characteristics of the surrounding rock stress field, displacement field, plastic zone, and support structure. The results during the excavation support stage show that the surrounding rock displacement primarily follows a quasi-hyperbolic distribution, with tensile failure predominating. Displacement and compressive stress in the roof increase while the sidewalls' compressive stress decreases. The axial stress of anchor cables increases at the fault plane. The results during the periodic weighting stage show that: Roof z-displacement propagates horizontally downward, with strain exhibiting step-like jumps. The extent of damage to the roof of the hanging wall, the floor, and the sidewalls of the foot wall is intensified. The location of AE events is influenced by the strike and dip of the fault. Shear failure dominates in sandstone at the fault plane, while tensile failure prevails in mudstone. Moreover, the axial stress in both bolts and cables generally increases, but locally decreases in the region where roof cables intersect the fault plane.