Repeated freeze–thaw (F–T) action, together with dynamic disturbances, can progressively weaken rock masses in cold regions. However, how freezing temperature affects the relationship between microstructural evolution and dynamic strength decay remains insufficiently quantified. This study investigated yellow sandstone subjected to F–T cycles at freezing temperatures of 0, −3, −5, and −20 °C. CT-based 3D reconstruction and Split Hopkinson pressure bar (SHPB) tests were combined with grey relational analysis (GRA) to characterize pore-structure evolution, dynamic strength decay, and their relationship. The results indicated that lower freezing temperatures promoted increases in pore connectivity and structural complexity. After 60 F–T cycles at −20 °C, connected porosity increased from 11.15% to 18.67%, while the ratio of connected porosity to total porosity increased from 51.1% to 85.7%. At an impact pressure of 0.3 MPa, the dynamic strength after 60 cycles decreased by 9.51%, 20.9%, 38.1%, and 61.5% at 0, −3, −5, and −20 °C, respectively. Among the examined microstructural parameters, average throat length had the highest overall grey relational grade (0.821), suggesting that throat development is closely associated with dynamic strength decay. Lower freezing temperatures enhanced pore-ice expansion and unfrozen-water migration, promoting pore enlargement, throat extension, and crack connection. These results quantitatively link pore-network evolution to dynamic strength decay under different freezing temperatures, providing a microstructural basis for assessing the dynamic deterioration of sandstone in cold regions.

Geological reservoirs typically contain numerous natural fractures, whose filled conditions significantly influence their mechanical behavior during reservoir stimulation. This work investigated the fracture and acoustic emission (AE) behavior of tight sandstone containing a partially filled defect using asymmetric semi-circular bending (ASCB) testing. The effects of filled ratio and defect width on mechanical and fracture response were examined, and the underlying fracture mechanisms were elucidated. The results show that peak strength of ASCB testing first increases, then stabilizes as the filled ratio increases, whereas the influence of defect width on peak strength gradually decreases. In terms of fracture mode, a transition from deflection to penetration is observed as the asymmetric loading ratio increases, with a critical threshold value of 0.5. This process is accompanied by higher-amplitude AE events within the partially filled defect, exhibiting a b-value of approximately 5.0. When the loading ratio exceeds 0.5, the b-value demonstrates a U-shaped variation trend with increasing filled ratio or defect width. These findings enhance our understanding of fracture behavior in reservoirs containing filled defects during stimulation processes.

Thermal conduction through building envelopes is a major factor contributing to heating and cooling energy losses, thereby increasing overall building energy consumption. Consequently, the development of novel building materials that combine excellent thermal insulation with mechanical properties has become an urgent research priority. This study reported a segmental cemented honeycomb material that collaboratively optimized mechanical and thermal insulation properties, and the mechanical compression with acoustic emission monitoring and infrared thermal imaging tests were conducted. A thermal insulation theory comprehensively considering the thermal conductivity of the material and the heat transfer limit height of the specific structure was constructed to solve the difficult problem of characterizing the thermal insulation performance of inhomogeneous materials and special-shaped structures. The results show that the Segmental cemented honeycomb material exhibits significant strain hardening characteristics through layer-by-layer deformation and wave load evolution, and its energy absorption increases by 9.73 times with unit densification. Its elastic modulus and initial peak load are negatively correlated with the segment length and number. The gradient honeycomb cavities effectively inhibit crack penetration, and the acoustic emission event increases with the decreases in segment length and number, indicating that densification units disperse stress through local fracture. The cemented wall dominates the heat conduction path, the isolated honeycomb cavities block heat convection, the increase in segment length and number leads to an increase in the size of the honeycomb chamber, an increase in the frequency of temperature fluctuations on the monitoring line, and an increase in the peak-to-valley difference in the fluctuating temperature. The temperature rise rate, heating rate, and heat transfer limit height of the sample are significantly reduced, and the thermal insulation performance is improved by 62.8% by regulating the segmental length and number.

To improve the utilisation efficiency of coal gangue, coal gangue concrete (CGC) confined by fiber reinforced polymer (FRP) has been explored as a novel supporting member for underground mines. However, underground mine environments generally contain high levels of sulfate ions, and meanwhile, supporting structures endure persistent high stress, leading to the deterioration of mechanical performance. Thus, acoustic emission (AE) response and damage mechanism of FRP-confined CGC (FCGC) after sulfate and high stress conditions were investigated during the axial compression test. The variables are different conditions, inner concrete, FRP confinement, stress-to-strength ratio, sulfate concentration and duration time. The results indicate that the AE ring count evolution of FCGC after sulfate and high stress conditions mainly characterizes four stages: elastic deformation, stable propagation of internal cracks, unstable propagation of internal cracks, and failure stages. RA-AF analysis reveals that tensile cracks dominate the entire compressive loading process, with the proportion ranging from 87.46% to 94.83% at the ultimate state. Furthermore, the stress-to-strength ratio is a critical factor affecting the AE ring count and RA-AF values of the specimens. With the process of compression, the crack of AE signals propagated gradually from two ends towards the centre, and eventually distribute in the whole space of the specimen, which is consistent with the macro failure pattern. More importantly, based on the AE parameters and Weibull distribution assumption, the damage evolution model was established. This will provide a theoretical guidance for the application of FCGC in the underground mines.

Using gangue as fine aggregate in cemented rockfill represents a key pathway to achieving sustainable mining development and ensuring the safety of underground structures. However, the low reactivity of gangue and its significant performance degradation at high blending ratios (>60%) severely limit its engineering applications. This study addresses the issue by tailoring the aggregate distribution of cemented rockfill via a new gradation equation, thereby enhancing its key performance metrics. The study systematically investigated the macroscopic mechanical properties (uniaxial compression, acoustic emission), microstructure (XRD, TGA, MIP), and mesoscopic mechanical mechanisms (discrete element simulation) of cemented rockfill under different fractal dimensions. Results show that, compared to the traditional Talbot gradation, the New gradation not only demonstrates a monotonic increase in mechanical properties with rising fractal dimension, effectively overcoming the issue of significant discreteness of performance, but also achieves a 14-day compressive strength of 47.5MPa at a fractal dimension of D=2.65, representing an improvement of 13.12%. It is mainly attributed to the optimization of gradation to make the coarse aggregate form an effective bearing skeleton, and the fine aggregate fills the pores to achieve the densification of the matrix. The transition of the crack failure mode from conjugate shear to single oblique shear is concurrently documented by acoustic emission monitoring. Discrete element simulations further demonstrate the evolution process of force chain from concentrated fracture to dispersed energy dissipation, indicating the toughness is enhanced. This study provides a low-cost and practical New gradation method to significantly improve the engineering reliability of cemented rockfill.

Fluid flow through rough-walled rock fractures is governed by surface roughness, shear deformation, and scale, yet a unified description of permeability evolution remains elusive. Self-affine fracture surfaces with varying roughness levels (JRC = 2-18) were generated using a modified successive random addition algorithm, and progressive shear displacement (3-18 mm) was numerically simulated to resolve aperture field evolution. Fluid flow was subsequently modeled using the Reynolds equation to evaluate equivalent permeability and directional transport behavior across scales (20-200 mm). Three principal physical mechanisms are revealed: (i) a competitive interplay between shear-driven enhancement and roughness-induced resistance governs permeability evolution; (ii) shear deformation fundamentally reorganizes flow topology through channelization aligned with the shear direction and contact-structure anisotropy that restricts transverse transport; and (iii) scale dependence arises from statistical averaging of local geometric heterogeneity, with flow stabilizing as observation windows expand. These mechanisms are synthetically captured in a coupled empirical correlation unifying roughness attenuation, shear-induced enhancement, and scale effects. Validation against numerical simulations demonstrates reliable first-order permeability prediction across the investigated parameter space.

Moisture-rich underground environments demand cementitious lining and grouting materials that are both durable and mechanically reliable. Acrylic redispersible polymer (AP) can render cementitious matrices hydrophobic, but its coupled evolution with load-induced damage development remains poorly understood. Therefore, this study develops a hydrophobic-mechanical synergy framework and establishes cross-scale evidence linking AP dosage to hydrophobic transition, mechanical response and damage evolution, interpreted alongside pore structure and hydration signatures. Water contact angle (WCA) measurements, uniaxial compression with acoustic emission (AE) monitoring, mercury intrusion porosimetry (MIP) and thermogravimetric analysis (TG-DTG) were performed. A clear hydrophobic transition is identified at similar to 1.71-1.96% AP, corresponding to an apparent polymer coverage threshold of similar to 82.6% at the outer surface. Gaussian fits show that stiffness maximises at similar to 0.84-1% AP, compressive strength peaks at 2% AP (38.88 MPa; +12.73%), and deformability maximises near 3% AP (peak strain 1.14%; +29.5%). AE-based damage metrics reveal a crack-minimisation window at similar to 1-2% AP with delayed shear engagement, whereas dosages >= 3% shift damage toward more tensile-dominated, distributed cracking. By combining these results, a robust hydrophobic-mechanical dosage window centred at similar to 1.8-3.6% AP is delineated. Within this window, MIP and TG-DTG results indicate suppressed capillary pores (10 & sup2;-10 & sup3; nm) and a dense, well-hydrated skeleton, in contrast to macro-defects and weakened hydration at excessive dosages. The identified synergy window offers a mechanistic and quantitative basis for designing hydrophobic and mechanically robust cementitious systems for underground lining and grouting.

Achieving ultrahigh-flow grouting materials with sustained mechanical performance under deep geothermal conditions remains critical in underground engineering. This study explores a cellulose nanofibers -tannic acid synergistic modification method to enhance grout performance, overcoming the 'fast setting and low strength' dilemma in geothermal environments. Through multi-scale characterization (fluidity, setting time, compression, hydration heat, XRD, FTIR, TG/DTG, MIP, SEM-EDS, Zeta potential, particle size distribution), the effects of cellulose nanofibers dosage, temperature, and curing age on the working performance, mechanical properties, composition and microstructure were investigated. The results show that the cellulose nanofibers synergizing tannic acid significantly inhibit the performance deterioration caused by high temperature. The interaction between tannic acid and hydroxyl groups on cellulose nanofibers constructs hydrogen-bond network, enhancing initial fluidity by 50 % (up to 252 mm) and 7-day compressive strength by 19.24 % (up to 31.5 MPa). And this network structure reduces the water loss rate caused by high temperature, the hydration products are oriented in 1 mu m pores through heterogeneous nucleation site regulation and bridging effect to form gradient densification structure. The optimal dosage of cellulose nanofibers to tannic acid (0.1 % CNFs + 0.18 % TA) was determined by multi-objective co-optimization. Based on the validation of ASTM C1437 standard, the dosage resulted in a 33.4 % increase in 60-min fluidity and a 26.41 % enhancement in 7-day mechanical properties of the grouting material. The research results provide a theoretical paradigm and industrial benchmark for the design of grouting materials in deep engineering, which can support the demand of engineering practice under high temperature and high pressure environments.

To address the high carbon footprint of traditional backfills and the lack of long-term stability data for highwall backfill mining, a low-carbon loess-slag-based cemented backfill (LSCB) was developed. This study investigates the time-dependent mechanical behavior of LSCB through triaxial creep tests under varying curing times (t c ) and confining pressures (6 3 ), supplemented by microstructural characterization. Results indicate that instantaneous strain increases linearly with stress, with the slope decreasing as t c extends but remaining insensitive to 6 3 . Creep strain exhibits nonlinear growth above a critical stress threshold that rises with both t c and 6 3 . Both long-term and creep failure strengths increase nonlinearly with t c and 6 3 , while their ratios to peak strength remain stable. Extending t c promotes geopolymer gel formation to densify the internal structure, and increasing 6 3 constrains crack initiation and propagation. To predict the creep behavior of LSCB, a nonlinear visco-elasto-plastic creep model was established and extended to three dimensions using tensor decomposition and the DruckerPrager criterion. The proposed model outperforms classical models in capturing all creep stages, particularly the accelerated failure phase. This work establishes a theoretical basis for the stability design of green backfill, advancing solid waste utilization in sustainable mining.

Accurate hydraulic parameterization of low-permeability rocks is often hindered by multiscale imaging practice, where "multiscale" datasets are frequently generated by resampling rather than repeated imaging, potentially confounding resolution effects with reconstruction artifacts. Using a dense sandstone sampled from the Central Yunnan Water Diversion Project, we compare true multiscale SEM imaging (repeated acquisition of the same fixed region of interest) with virtual multiscale reconstructions produced by downsampling (D2), upsampling (U2), and a logical-consistency fusion strategy. All datasets are evaluated on matched target grids to isolate representation effects. Porosity, pore-size distribution and its cumulative form, boundary fractal dimension, and an internally consistent apparent permeability proxy derived from the radius spectrum are quantified. Results show a robust cross-target trend: the apparent permeability increases as the target grid becomes coarser, consistent with systematic target-dependent shifts in pore-size representation toward larger equivalent radii, which disproportionately control permeability-like metrics. Superimposed on this dominant trend, virtual reconstructions introduce strong, target-dependent biases relative to the true multiscale reference: D2 yields conservative underestimation at coarse targets, whereas U2 can produce substantial overestimation at intermediate targets; fusion generally exhibits moderate positive bias. Notably, permeability-like indicators are far more sensitive to these representation effects than porosity. These findings demonstrate that matching nominal grids alone does not guarantee flow-equivalent multiscale representations, and hydrologic applications should explicitly validate pore size distribution tail behavior and throat-/connectivity-sensitive metrics when resampling-based multiscale images are used.

Theoretical analysis of true triaxial rock fracture mechanisms elucidates that the maximum ( σ _1 ) and intermediate ( σ _2 ) principal stresses predominantly govern macroscopic fracture orientation. To validate this principle, a computational framework simulating fracture evolution in randomly damaged thick-walled cylinders was established. Numerical experiments under three distinct principal stress regimes demonstrate that rotation of σ _1 and σ _2 directly dictates fracture plane orientation. Extending this concept to underground engineering, we propose a methodology for controlling fracture geometry in arched roadways: strategic adjustment of principal stress trajectories during excavation, combined with presplitting techniques to modify the local stress field around boreholes, effectively redirects fracture propagation in shoulder and corner zones. Results confirm the pivotal role of σ _1 - σ _2 plane alignment in rock fracture initiation and the directional dependence of surrounding rock fissures. This mechanistic insight provides critical guidance for structural stability optimization and disaster mitigation in deep underground projects.

Dry hot rock reservoirs are characterized by low porosity and permeability, necessitating hydraulic fracturing to enhance reservoir permeability for the construction of enhanced geothermal systems (EGS). However, the presence of filled defects can significantly affect fracture propagation behavior. In this work, granite samples containing filled defects of varying strengths were prepared and then subjected to hydraulic fracturing tests under varying confining pressures. The fracture networks were analyzed using μCT scanning to obtain the three-dimensional pattern. Propagation criteria were established to elaborate on the fracturing behavior in granite containing filled defects. Experimental results indicate that the confining pressure predominantly contributes to the breakdown pressure of hydraulic fracturing, whereas the strength of filled defects exhibits a minimal effect. With respect to fracture pattern, an increase in confining pressure facilitates the activation of gypsum-filled defects in a deflection mechanism. Additionally, as the filled defect strength increases, the interaction between induced cracks and filled defects transitions from deflection to offsetting. The extended R&P criterion, based on classical R&P theory, outperforms both the Blanton and the W&T criteria, achieving an average relative error of 6.3%. It demonstrates good agreement with experimental results and exhibits superior reliability and applicability. This study results contributes to a further understanding of stimulation mechanisms in dry hot rock reservoirs.

Effective stability management of surrounding rock is critical to ensuring the safety and efficiency of underground coal gasification operations. This study investigates the strain rate sensitivity and damage evolution of coal-measure sandstone under coupled high-temperature and high-strain-rate conditions. Using a custom-built near-uniform high-temperature dynamic loading system, the dynamic mechanical properties, energy dissipation behavior, and fractal features of the rock were examined. Experimental results reveal that at elevated temperatures, coal-measure sandstone exhibits a clear strain rate effect. As the strain rate increases, the dynamic elastic modulus, compressive strength, and energy dissipation density increase by 50.31

Understanding the coupled influence of material brittleness and fracture geometry on the mechanical degradation and failure behavior of fractured rock is critical for assessing instability in deep underground engineering. In this study, sandstone specimens with two distinct levels of brittleness were prepared with prefabricated fractures characterized by varying inclination angles and rock-bridge angles. A multiscale experimental framework integrating AE-DIC monitoring was employed to systematically investigate damage evolution and fracture mechanisms. To quantitatively characterize mechanical weakening, strength-degradation and elastic-modulus-degradation indices were proposed. The results show that prefabricated fractures significantly reduce the peak strength of sandstone. Increasing the fracture inclination angle leads to higher peak strength, whereas increasing the rock-bridge angle results in a non-monotonic decrease-increase trend. High-brittleness sandstone exhibits larger fluctuations in strength and stiffness under geometric disturbance, indicating stronger sensitivity to fracture-induced heterogeneity. AE-DIC results demonstrate that fracture inclination primarily controls the orientation and intensity of strain localization, while the rock-bridge angle governs the complexity of crack coalescence. High-brittleness specimens develop more concentrated and rapidly evolving strain localization bands, ultimately leading to abrupt mixed tensile-shear failure. Higher brittleness accelerates damage accumulation and promotes shear-dominated microcrack activity, particularly under high fracture inclination and large rock-bridge angles. In contrast, low-brittleness sandstone is characterized by a higher proportion of tensile microcracks and smoother b-value evolution, reflecting more progressive damage development. The degradation indices further reveal that strength deterioration is more pronounced in high-brittleness sandstone, whereas elastic modulus degradation shows greater sensitivity to brittleness reduction in low-brittleness sandstone. Overall, the findings highlight a coupled degradation mechanism in which fracture geometry controls the spatial evolution of damage, while material brittleness governs the rate and severity of mechanical degradation.

Fractured rock masses reinforced by cement filling are commonly encountered in underground and geotechnical engineering, yet their failure evolution and reinforcement mechanisms remain insufficiently understood, particularly under the coupled influence of fracture geometry and interfacial conditions. In this study, sandstone specimens containing a single pre-existing fracture were prepared with varying fracture dip angles and rock bridge angles under both unfilled and cement-filled conditions. An integrated acoustic emission (AE) and digital image correlation (DIC) multi-field monitoring system was employed to investigate the mechanical response, deformation localization, microcrack evolution, and failure characteristics during uniaxial compression. The results indicate that fracture dip angle and rock bridge angle play dominant roles in governing strength, deformation behavior, and failure modes, while cement filling markedly enhances mechanical performance and stabilizes the failure process. Compared with unfilled specimens, cement-filled specimens exhibit delayed crack initiation, suppressed strain localization, and a systematic reduction in tensile-dominated microcracking, accompanied by a transition from abrupt unstable failure to more progressive and stable damage evolution. AE-RA-AF analysis further reveals that the proportions of tensile and shear microcracks vary systematically with fracture geometry, and that cement filling effectively moderates microcrack activity and alters crack propagation paths. To quantitatively characterize fracture-induced degradation and cement-induced reinforcement, an interface weakening coefficient and a strength recovery coefficient are introduced, providing a unified framework for evaluating the mechanical effects of fracture geometry and interfacial filling. The findings offer new insights into the multi-field failure mechanisms of cement-filled fractured rocks and provide practical guidance for reinforcement design and stability control in fractured rock engineering.

Reliable image-based permeability estimation in tight porous media depends strongly on how pore and throat geometries are resolved across scales. This study investigates the influence of image resolution on pore characterization and permeability estimation in tight sandstone using true multiscale scanning electron microscopy (SEM). A fixed sandstone region was imaged at three resolutions—S1 (0.1 μm/pixel), S4 (0.05 μm/pixel), and S16 (0.025 μm/pixel)—and spatially registered to ensure the same field of view across scales. Porosity, pore roundness, fractal dimension, pore size distribution, and permeability were extracted and compared. With increasing resolution, more fine pores are identified, porosity rises from 4.6% (S1) to 5.55% (S4) and 6.31% (S16), pore roundness and fractal dimension increase, indicating greater complexity and fine-scale heterogeneity. Meanwhile, the pore size distribution narrows and shifts towards smaller pores as large merged pores at low resolution are decomposed into multiple micropores. Permeability derived from individual images becomes more spatially variable at higher resolutions, but the overall permeability decreases, with only a small additional change from S4 to S16. The values at the S4 and S16 scales (1.77 × 10−17 m2 and 1.72 × 10−17 m2) agree well with the measured gas permeability of 1.85 × 10−17 m2. These results indicate that image resolution exerts systematic control on transport-relevant pore descriptors and image-based permeability. Within the investigated resolution range, further refinement from S4 to S16 reveals additional fine-scale heterogeneity but produces only a limited change in the overall permeability estimate. The findings, therefore, highlight the importance of balancing image resolution and field-of-view representativeness in digital-rock workflows aimed at pore-scale transport analysis and permeability upscaling.

The roughness of rock fractures has complex features that affect how fluids move through them. This research looks at how gas and water flows change in rough fractures when they are moved using a model based on fractal geometry. Rough surfaces are created using a method called fractional Brownian motion. When the surfaces are moved, the space in the fractures becomes uneven. By using a level-set method together with a fluid flow model, the study explores how the speed the fluid enters, the roughness of the surface, and the movement of the surfaces affect the change between bubble, slug, and ring-like flow. The results indicate that more roughness and movement make the flow less stable, which causes a reverse change from ring-like flow to slug and bubble flow. A framework based on pressure is built, showing that the outlet pressure decreases quickly with fluid speed, rises steadily with roughness, and changes in a square relation with movement. A single prediction formula is made with R2 = 0.98, allowing precise identification of the flow types using pressure change limits. This research gives insights into flow changes in fractured reservoirs and offers a way to predict flow in real-time.

The long-term stability of coal pillars in underground stopes and open-pit slopes depends critically on the creep behavior of coal–rock combinations, wherein the rock-to-coal height ratio ( K rc ) exerts a governing yet insufficiently characterized influence. In this study, stepwise creep tests, scanning electron microscopy observations, and acoustic emission tests were conducted on five types of specimens: pure coal (0:1), three composite specimens with height ratios of 1:3, 1:1, and 3:1, and pure rock (1:0). The influence of the K rc on creep deformation, long‑term strength, and Acoustic emission crack evolution was systematically investigated. Three principal findings emerged. First, irrespective of K rc , the creep failure stress consistently approached 90% of the peak uniaxial compressive strength across all specimen types, establishing a straightforward yet robust predictive relationship for long-term strength estimation. Second, the 1:1 composite configuration demonstrated the greatest creep longevity, along with the smallest instantaneous and time-dependent strains and the lowest steady-state creep rate among all composite specimens—collectively identifying this ratio as an optimal configuration for creep resistance. Third, as the proportion of rock increased, the relative contribution of shear cracking declined while tensile cracking became progressively dominant, reflecting a transition in failure mechanism from shear-controlled to tension-controlled behavior. Correspondingly, both the crack initiation stress and the crack damage stress thresholds followed a non-monotonic trend, reaching peak values at K rc = 1:1 before declining at higher rock fractions. Taken together, these results provide quantitative guidance for the rational design of coal pillar geometries and the assessment of long-term structural stability in underground mining operations.

Predicting shear-behavior transitions in block-in-matrix rocks (bimrocks) requires a descriptor that links mesoscale localization to macroscopic strength mobilization. Volumetric block proportion (VBP) specifies composition, whereas conventional interpretations based on peak friction-angle increments do not directly quantify the geometry of the active shear path. Here, equivalent shear-slip roughness is defined from the statistical centerline of a displacement-gradient-derived slip field and used to characterize the VBP-dependent transition. A three-dimensional Voronoi-expansion method and a heterogeneous discrete-element model are used to resolve stress-displacement responses, crack evolution, contact-force networks, and slip fields for VBP = 10-70%. Increasing VBP changes the response from localized matrix-through fracture with post-peak softening to distributed interface-guided shearing and stable frictional/block-rearrangement bearing. Concurrently, the residual equivalent slip path becomes progressively more tortuous, while load transfer evolves from matrix cohesion to frictional contact and block interlocking. The proposed roughness therefore complements VBP by quantifying how the imposed block proportion is expressed in the evolved localization geometry. Although evaluated from the post-deformation field, the residual numerical roughness converges toward the first-order theoretical estimate, indicating that the fully mobilized slip state is consistent with the model assumptions across the investigated VBP range. These results provide a measurable structure-mechanics link and support future roughness-informed strength formulations. Block morphology, orientation, gradation, anisotropy, and scale can be incorporated as refinements in broader validation.

Accurate identification of crack types in rock masses is critical for understanding damage mechanisms and ensuring the structural safety of rock engineering. This study presents a novel unsupervised classification framework based on Gaussian mixture modeling (GMM) for distinguishing acoustic emission (AE) signatures associated with different fracture modes in sandstone samples that contain prefabricated fissures at varying inclination angles. The frequency-domain characteristics of the AE signals were extracted using fast Fourier transform (FFT), while the RA–AF (rise time/amplitude versus average frequency) parameter space was employed to characterize the crack mechanisms. To increase classification accuracy and model robustness, the Bayesian information criterion (BIC) was introduced to determine the optimal number of Gaussian components. Experimental results from uniaxial compression tests reveal that fissure inclination significantly affects crack evolution behavior: low-angle fissures favor shear and hybrid cracks, whereas high-angle fissures cause tensile failure. The proposed GMM-based method effectively identifies tensile, shear, and hybrid cracks with increased objectivity and accuracy, outperforming traditional empirical RA–AF thresholding techniques. This research provides a reliable and generalizable approach for AE signal classification, which presents theoretical insights and practical support for real-time monitoring, early warning, and structural health assessment in fractured rock masses.