Study on the factor reduction of rockburst risk in coal mines Authors: Jing Zhang, Beifang Wang, Duo Zhou Published: 2026 ABSTRACT AbstractRockburst is a typical dynamic hazard in coal mining that severely affects safe production. Existing risk evaluation systems for rockbursts involve numerous factors and overlap in information, which affects the accuracy of judgment. This study combines theoretical analysis, mathematical modeling, and field testing to analyze the influencing factors of rockburst risk from multiple dimensions. Based on a factor reduction algorithm, seven influencing factors of rockburst are analyzed to identify the key disaster-causing factors and construct a hazard assessment method. The results indicate that coal seam thickness variation, geological structures, and floor coal thickness exert dominant control over the rockburst risk in coal mines. Using the ZF1420 working face of the Yadian coal mine in Shaanxi as a case study, the risk evaluation results align with the actual conditions. The study also proposes classification-based prevention and control measures for rockbursts. The research provides theoretical and methodological support for the evaluation and prevention of rockburst risks.Keywords: Rockburst, Factor reduction, Determination degree, Risk evaluationSubject terms: Energy science and technology, Engineering, Environmental sciences, Natural hazards, Solid Earth sciences Introduction Rockburst is a dynamic phenomenon in underground coal mining, influenced by multiple geological and mining factors. It occurs when the coal-rock mass surrounding the mine shaft or working face suddenly releases accumulated elastic energy under the combined action of in-situ and mining-induced stresses1. This leads to sudden, intense, and catastrophic destruction. The occurrence of rockbursts is accompanied by loud noises and vibrations, which can instantly destroy working faces or roadways ranging from several meters to several hundred meters in length. This results in support failure, equipment damage, roadway blockages, and casualties, making rockburst one of the most severe dynamic hazards in coal mining2–5. China is one of the countries most severely impacted by rockburst disasters. According to data from the National Mine Safety Administration, by April 15, 2024, the number of rockburst-prone mines in China has increased to 138, distributed across 15 provinces. Among these, Shandong, Shaanxi, Heilongjiang, Gansu, and Inner Mongolia each have over 10 such mines (Fig. 1). Destructive rockbursts occur several hundred times each year. As mining depth increases, rockbursts are becoming more frequent and destructive under “three-high” in-situ conditions (high stress, high temperature, and high gas). Even mines with no previous history of rockbursts are now experiencing these phenomena. The prevention and control of rockbursts have become an urgent global scientific challenge in the field of coal mine safety production6. Overview of rockburst conditions in China: (a) spatial distribution of rockburst-prone coal mines in China; (b) temporal variation in the number of rockburst-prone coal mines in China over recent years. Research on rockburst prevention and control in Chinese coal mines began in the early 1980s. Over more than 40 years of development, significant achievements have been made through the relentless efforts of experts and technicians in the field7–11. Rockburst prevention and control encompasses risk evaluation, monitoring and early warning, control technologies, and safety protection, with the most critical and fundamental component being rockburst risk evaluation12. Bukowska13, considering the natural geological conditions of mining operations in the Upper Silesian coal basin, proposed a system for evaluating the probability of rockburst occurrence and assessing rockburst risk states. Pan et al.14, through the establishment of a coal-rock charge sensing experimental system, studied the charge sensing laws during the deformation and rupture process of coal-rock, and proposed a technique for predicting rockburst risk based on charge sensing. Zhang et al.15 developed an evaluation method and indicator system for the geological and dynamic conditions of rockbursts, and proposed a method for predicting rockbursts by monitoring fault activity with anchor cables. Rong et al.16 analyzed the main triggering factors of rockbursts at Wudong coal mine and proposed a multi-factor pattern recognition method for predicting rockburst risk in steeply inclined, thick coal seams, and performed risk evaluations and predictions for various regions within the Wudong mining area. Liu et al.17, using the isolated working face under complex geological conditions at Nanshan coal mine as the study background, proposed a comprehensive prediction method for rockburst risk, including both regional and local predictions, with the results aligning with actual field conditions. Zhu et al.18, after analyzing the multiple influencing factors of rockbursts, established a GA-ELM (Genetic Algorithm-Extreme Learning Machine) rockburst prediction model, improving the accuracy of rockburst predictions. Zeng et al.19 developed a fuzzy evaluation index for the influencing factors of rockburst risk and established a rockburst risk evaluation model. Wang et al.20, by analyzing the composition, movement patterns, and stress characteristics of the “S”-shaped overburden space structure, proposed a rockburst prediction method based on the theory of overburden spatial structure. This method macroscopically assesses the rockburst risk at each stage of the mining operation. Zhu et al.21, addressing the challenge of determining the weights of indicators in rockburst risk evaluation, proposed the CW-TOPSIS comprehensive evaluation model based on the AHP (Analytic Hierarchy Process) and entropy weight method, enabling accurate prediction of rockburst levels. Jiang et al.22, in response to rockburst accidents in a fully mechanized top coal caving face with soft-hard layered composite thick coal seams, constructed a “seismic-shock” dynamic disaster mechanical model and analyzed the mechanical conditions and interrelationships between mine tremors and rockbursts. The aforementioned scholars have made significant contributions to the evaluation of rockburst risks. However, the existing evaluation systems commonly face a shared problem: in the pursuit of comprehensiveness, the indicator systems continue to expand, resulting in complex evaluation factors, dispersed weights, and overlapping information. This not only increases the complexity of the evaluation but also potentially obscures key disaster-causing factors due to interference from secondary or redundant factors, thus affecting the accuracy and effectiveness of the evaluation. In light of this, the author employs a combination of theoretical analysis, mathematical modeling, and field testing to analyze the influencing factors of rockburst induction from multiple dimensions. By using factor reduction methods, the key factors affecting rockburst risk are identified, and a rockburst risk evaluation method is constructed. The method is applied to the Yadian coal mine as an engineering case study to perform risk evaluation and propose classification-based prevention and control measures for rockbursts. Unlike traditional multi-index weighted evaluation methods, this study adopts a “factor redundancy reduction” perspective and introduces factor space theory and a factor reduction algorithm to achieve quantitative identification and rule-based representation of key controlling factors of rockburst risk. While ensuring evaluation accuracy, this approach significantly reduces method complexity, providing a new research perspective for rapid assessment of rockburst risk. The research provides theoretical and technical support for the evaluation and prevention of rockburst risks in the Yadian coal mine and other mines with similar geological conditions. Analysis of influencing factors of rockburst risk Rockburst is a nonlinear dynamic hazard with a complex array of influencing factors. The factors affecting rockburst risk can be classified into two main categories: geological factors and mining factors. The geological conditions of the mine, coal seam, and mining area, along with the mining conditions, determine the different influencing factors of rockburst occurrence. These factors exhibit varying patterns in terms of time, space, and intensity. By combining relevant engineering research findings and the indicator system for rockburst risk classification, a comprehensive analysis of the influencing factors of rockburst risk is conducted. Geological factors Mining depth With increasing mining depth, the self-weight stress of the overlying strata acting on the coal-rock mass increases correspondingly, leading to a significant accumulation of elastic strain energy in the coal seam and bringing it closer to its ultimate bearing capacity. This mechanical response not only markedly increases the likelihood of rockburst initiation but also results in more intense energy release once a rockburst occurs. Statistical analyses indicate a positive correlation between mining depth and the probability of rockburst occurrence. When the mining depth H≤350 m, rockbursts rarely occur; for 350 m ≤ H≤500 m, the rockburst risk gradually increases; beyond 500 m, the rockburst risk rises sharply with further increases in mining depth (Fig. 2)23. It should be emphasized that while mining depth primarily dictates the magnitude of the initial vertical stress (σv = γH) derived from the overburden deadweight, rockburst susceptibility in engineering practice is not merely a monotonic function of depth. Instead, it is profoundly governed by the regional tectonic stress field. The orientation and magnitude of the maximum principal stress (σ1) dictate the characteristics of stress redistribution and the mode of energy accumulation around the longwall gateroads. In structural domains where σ1 aligns vertically, stress concentration and elastic energy storage primarily occur deep within the coal pillars and gateroad ribs, driven predominantly by overburden load transfer. Conversely, within highly active tectonic zones—such as fold-and-thrust belts, fault systems, or plate subduction margins—σ1 typically manifests horizontally, resulting in a lateral pressure coefficient (λ) significantly greater than unity24,25. Intensive horizontal compression not only alters the stress distribution in the surrounding rock, but also induces pronounced shear concentration and buckling deformation within the roof and floor strata, thereby amplifying the localized capacity for elastic energy storage and sudden release. Consequently, in such tectonic-stress-dominated environments, a elevated rockburst risk can persist even under shallow cover. Relationship between mining depth and rockburst occurrence. Coal seam thickness variation According to geomechanical theory, zones where the coal seam thins or where the dip angle changes abruptly are typically areas of structural stress concentration. When mining operations approach such zones, stress superposition is likely to occur, thereby triggering rockbursts. It should be noted that the magnitude and abruptness of coal seam thickness variation often have greater engineering significance for rockburst occurrence than the absolute coal seam thickness itself. In areas with sudden thickness changes (rapid thinning or thickening), the coal mass is prone to pronounced stress concentration under static loading conditions(Fig. 3). The influence of different local thickness variation patterns on stress distribution can be summarized as follows26,27: when the coal seam locally thins, vertical stress in the thin-seam zone increases, forming a stress concentration; when the coal seam locally thickens, vertical stress within the thickened zone decreases relatively, whereas vertical stress in the adjacent normally thick coal zones increases; the intensity of coal seam thickness variation directly affects the degree of stress concentration, with larger thickness gradients resulting in higher stress concentration factors. In particular, within coal seam thinning zones, shorter longitudinal extents of the thinned section lead to more pronounced stress concentration effects. Illustration of stress concentration due to variation in coal thickness. Geological structure Geological structures are the products of long-term tectonic movements and commonly include faults, folds, anticlines and synclines, coal seam thickness variation zones, and lithological transition zones. Within these structural zones, significant concentration of tectonic stresses—particularly horizontal tectonic stress—alters the internal stress state of the coal-rock mass, substantially increasing the likelihood of rockburst initiation. Specifically, pronounced stress concentration commonly occurs in the axial zones of synclines and anticlines, where the maximum principal stress is typically oriented horizontally (Fig. 4). This stress regime readily induces severe strata behavior or rockbursts characterized by intense deformation of the roof and floor, phenomena that are especially prominent during roadway excavation3,28. Faults, as typical geological discontinuities, exert a critical influence on the structural integrity and stress distribution of the coal-rock mass. The extent of this influence depends on fault properties, the mechanical characteristics of the surrounding rock, and the regional in-situ stress environment. When the shear stress acting on a fault plane reaches its shear strength, the fault enters a critical instability state, at which point even minor external disturbances may activate fault slip and trigger intense rockbursts29,30. It should be noted that the disturbance of the in-situ stress field caused by faults exhibits strong locality. Beyond the affected zone, the stress distribution gradually returns to its normal state. Stress state and danger of fold parts. Roof characteristics A thick and competent roof is one of the typical geological conditions conducive to rockburst occurrence. Statistical analyses indicate that coal seams prone to rockbursts are often overlain by one or more hard rock layers with thicknesses exceeding 10 m31. The influence of such thick and hard roofs on rockburst development can be explained by two primary mechanisms (Fig. 5)32,33. First, after coal extraction, the immediate roof collapses progressively with the advance of the working face, while the overlying thick and competent roof remains suspended. This suspension transfers the load of the overlying strata to the coal mass ahead of and on both sides of the working face, resulting in pronounced abutment pressure and lateral support pressure. The associated stress concentration markedly increases the elastic strain energy stored in the coal-rock mass, thereby significantly elevating the rockburst risk in the affected zones. Second, during the fracture or sliding instability of a thick and competent roof, a large amount of accumulated elastic energy is abruptly released, generating high-intensity dynamic loading disturbances. Such disturbances can readily trigger rockbursts or other dynamic disasters, such as large-scale roof weighting events. Mechanism of rockburst induced by hard roof. Mining factors Floor coal thickness Both theoretical studies and engineering practice indicate that rockburst occurrence is often accompanied by pronounced floor heave, and the presence of a thick floor coal layer tends to aggravate rockburst risks (Fig. 6)34. In the surrounding rock system of a roadway, the roof and ribs are typically constrained by support systems, whereas the floor is generally unsupported, making it the weakest structural component. When dynamic loads associated with rockbursts act on the roadway, energy is prone to be released through this weak zone, inducing either gradual deformation or instantaneous impact failure of the floor. According to the nature of the floor medium, floor conditions can be classified into three types: rock floor, thin coal floor (thickness<2 m), and thick coal floor (thickness>2 m). The first two types are more common, while thick coal floors mainly occur in working faces extracting thick and ultra-thick coal seams. The influence of floor lithology on rockburst development primarily depends on its inherent strength and the magnitude of the applied load35. Rock floors, characterized by high strength and strong bearing capacity, exhibit substantial resistance to dynamic loading. When the floor coal is thin, progressive failure occurs after roadway excavation, leading to a gradual reduction in bearing capacity, while the underlying rock strata remain the primary load-bearing structure. In contrast, under thick floor coal conditions, the coal mass becomes the main load-bearing body and is more susceptible to instability under dynamic loading. If the floor coal itself exhibits rockburst proneness, elastic strain energy can readily accumulate and be suddenly released, thereby further intensifying the risk of dynamic disasters. Relationship between the rockburst and the thickness of floor coal. Pressure-relief conditions Pressure-relief measures are a key mining technology for controlling rockburst risks, and their effectiveness directly determines the prevention and mitigation of dynamic disasters in coal mines. The essence of rockburst lies in the instantaneous release of elastic strain energy accumulated in the coal-rock mass, whereas pressure-relief operations aim to modify the surrounding rock stress field through artificial intervention, thereby promoting gradual energy release or redistribution and reducing rockburst risk36. Specifically, when high-intensity stress concentration zones exist ahead of a working face—such as those adjacent to geological structures, beneath thick and competent roofs, or around improperly designed coal pillars—energy accumulation cores are likely to form. In the absence of effective pressure-relief measures, the coal-rock mass is highly susceptible to dynamic instability and rockburst under mining-induced disturbance. Common pressure-relief techniques include large-diameter borehole drilling, deep-hole blasting for coal seam pressure relief, and hydraulic fracturing of the roof37. These methods alter the coal mass structure, weaken roof integrity, or induce fracture development, thereby disrupting the high-stress equilibrium state and achieving stress transfer, advance, or dissipation. Coal burst sounds Coal burst sounds emitted by the coal-rock mass under mining-induced stress, are one of the most direct and common precursors during the rockburst initiation process. The mechanism behind these sounds is that when the stress applied to the coal-rock mass approaches its ultimate strength limit, pre-existing fissures within the material begin to propagate and coalesce, creating new fractures. The elastic strain energy accumulated during this process is released in the form of sound waves, resulting in vibrations and audible noise38. The dynamic evolution of coal burst activity—characterized by frequency, energy, and location—serves as a critical indicator for assessing the rockburst risk in a working face. Generally, coal burst sounds transition from sporadic and dispersed to frequent and concentrated, with the intensity of the sounds increasing progressively. This indicates an increase in the stress level within the coal-rock mass, accelerating the accumulation of internal damage, and significantly raising the risk of rockburst. Application of factor reduction in rockburst risk factor analysis Factor reduction The factor space theory established by Wang states that factors are essential components constituting things, and the factors of things can be regarded as a mapping relation39. This mapping associates objects in the universe of discourse with specific attributes or states of their factors. Specifically, it can be expressed as follows: Let U be the universe of discourse (i.e., the set of objects under discussion), and F be the set of factors for the objects in the universe. For any factor f∈F, there exists a corresponding space X(f), called the state space of factor f, which represents the set of possible values (i.e., the set of states or attributes) that factor f can take. That is, for each u∈U and each f∈F, there exists a mapping φu,f :U×F→X(f)={a1,a2,⋯,an}, such that φu,f (u,f)∈X(f). To further distinguish, let R∈F be the outcome factor of particular interest. The set of condition factors C⊆F−{R} that influence (i.e., determine or affect) the value of R is referred to as the condition factor set. Decision degree expresses the extent to which condition factors influence or determine the occurrence of an outcome for an object. For a certain state s of condition factor fi, if all objects with state s under fi are included in the set of objects corresponding to a certain state or level t of outcome factor g, i.e., [t]={ui∣g(ui) = t}⊇[s], then [s] is called a decision class of fi. Here, [s] denotes the set of all objects with state s under fi, and [t] similarly denotes the set of all objects with state t under outcome factor g. The union of all decision classes of factor fi is called its decision domain. The ratio of the number of objects included in the decision domain of factor fi to the total number of objects is called the decision degree of fi, denoted as d(fi). This proportion quantifies the degree of influence of fi on outcome factor g. The formula for calculating the decision degree d(fi) is: where: —denotes the total number of objects in all decision classes of factor fi, with n being the number of all possible states of fi; |U|—represents the total number of objects in the universe. Factor reduction refers to the process of eliminating redundant factors from the set of conditional attributes. The reduction method based on decision degree can be implemented through the following three steps: Construct a factor analysis table denoted as (U,F, G), consisting of the object set U, the conditional factor set F, the result factor set G, as well as the state data of each object with respect to all conditional factors and the result factor; Construct a factor analysis table denoted as (U,F, G), consisting of the object set U, the conditional factor set F, the result factor set G, as well as the state data of each object with respect to all conditional factors and the result factor; Classify each conditional factor according to the state of the result factor g, calculate the decision degree d(fi) or each factor, select the factor corresponding to the maximum decision degree to partition the domain, and transform all decision classes into inference rules; Classify each conditional factor according to the state of the result factor g, calculate the decision degree d(fi) or each factor, select the factor corresponding to the maximum decision degree to partition the domain, and transform all decision classes into inference rules; Add the selected factor into the reduction set and remove the corresponding decision domain. Repeat the above procedure until the domain is empty. The selected factors constitute the final reduced set of conditional factors. Add the selected factor into the reduction set and remove the corresponding decision domain. Repeat the above procedure until the domain is empty. The selected factors constitute the final reduced set of conditional factors. Factor reduction has been effectively applied in the field of safety science and engineering, as it can eliminate redundant factors from the numerous influencing factors of safety accidents to identify the causative factors of accidents40. This provides a rapid method for determining causative factors during accident investigation and analysis. Construction of a factor analysis table for rockburst risk Based on existing theoretical frameworks and engineering practice, numerous factors influence rockburst occurrence. To simplify the analysis, seven principal factors were selected to construct a factor analysis matrix41,42. These include mining depth f1, coal seam thickness variation f2, geological structure f3, roof characteristics f4, floor coal thickness f5, pressure-relief conditions f6, and coal burst sounds f7, forming the conditional factor set F={f1, f2, f3, f4, f5, f6, f7}. To ensure the repeatability and reliability of factor reduction, fifteen coal mines were selected to constitute the research domain U={u1,u2,…,u15}. among which fourteen exhibit typical rockburst characteristics. The data were compiled from published literature and engineering statistics43–46. The sample selection followed three criteria: (1) clearly documented rockburst intensity levels; (2) complete geological and mining parameters; and (3) representative regional distribution. The classification of each influencing factor was conducted according to current rockburst evaluation standards, relevant studies, and engineering experience47–50. Mining depth was categorized into shallow (≤ 350 m), moderate (350–500 m), relatively deep (500–800 m), and deep (> 800 m), reflecting critical depth thresholds for rockburst occurrence. Coal seam thickness variation was divided into small, moderate, relatively large, and large based on thickness gradients and their impact on stress concentration. Geological structure complexity was classified as simple, moderate, relatively complex, and complex according to fault density, fold development, and structural influence range. Roof characteristics were graded by integrating roof strength and the thickness of competent strata. Floor coal thickness was categorized into rock floor (no bottom coal), thin (< 2 m), and thick (> 2 m). Pressure-relief conditions were classified based on the stress state of the coal ahead of the working face. Coal burst sounds were graded according to the frequency of sound emissions generated by coal fracturing under high stress. These classification schemes have been widely adopted in rockburst risk assessment studies and effectively capture the relative influence of each factor on rockburst occurrence. The value states of the seven factors with respect to each coal mine are defined as follows: Mining depth X(f1)={ shallow, medium, relatively deep, deep}={1,2,3,4}; Coal seam thickness variation X(f2)={ small, medium, relatively large, large}={1,2,3,4}; Geological structure X(f3)={ simple, moderate, relatively complex, complex}={1,2,3,4}; Roof characteristics X(f4)={ very good, good, moderate, poor}={1,2,3,4}; Floor coal thickness X(f5)={ none, thin, thick}={1,2,3,4}; Pressure-relief conditions X(f6)={ good, moderate, poor, none}={1,2,3,4}; Coal burst sounds intensity X(f7)={ none, few, frequent }={1,2,3}. Based on the outcome characteristics of rockburst risk, a single result factor is defined. According to existing rockburst classification standards and statistical records of historical occurrence intensity, the result factor G is categorized into four distinct states: 1-no rockburst risk, 2-weak risk, 3-moderate risk, and 4-strong risk. The determination degree of each factor fi is calculated, and the resulting rockburst risk factor analysis is summarized in Table 1. Table 1Analysis of rockburst risk factors.Universe UCondition factor FOutcome Gf1f2f3f4f5f6f7 u 1 24422234 u 2 34221333 u 3 31332413 u 4 41332133 u 5 14411134 u 6 32141411 u 7 21131122 u 8 32321333 u 9 21131122 u 10 23421134 u 11 42331213 u 12 41221133 u 13 31141412 u 14 12412134 u 15 32413133d(fi)0.330.070.4000.070.130.13 Analysis of rockburst risk factors. Factor reduction calculation As shown in Table 1, factor f3 has the highest determination degree and is therefore selected as the first splitting attribute, yielding four categories in total: Among them, the first category includes outcomes 1 and 2 and thus does not constitute a decision class; the second category includes outcome 3 and constitutes a decision class; the third category also includes outcome 3 and constitutes a decision class; the fourth category includes outcomes 3 and 4 and does not constitute a decision class. Accordingly, the following two inference rules can be formulated. Therefore, the decision domain is {2,3,4,8,11,12}. After removing the decision domain, Table 2 is obtained. Table 2Analysis of factors after removing the decision domain.Universe UCondition factor FOutcome Gf1f2f3f4f5f6f7 u 1 24422234 u 5 14411134 u 6 32141411 u 7 21131122 u 9 21131122 u 10 23421134 u 13 31141412 u 14 12412134 u 15 32413133d(fi)0.220.6700.560.330.110.22 Analysis of factors after removing the decision domain. As indicated in Table 2, factor f2 exhibits the highest determination degree and is therefore selected as the next splitting attribute, resulting in four categories: In this partition, the first category contains outcome 2 and thus constitutes a decision class; the second category contains outcomes 1, 3, and 4 and does not constitute a decision class; the third category contains outcome 4 and constitutes a decision class; the fourth category also contains outcome 4 and constitutes a decision class. Accordingly, the following inference rules can be derived: Therefore, the decision domain is {1,5,7,9,10,13}. After removing the decision domain, Table 3 is obtained. As shown in Table 3, the determination degree of factor f5 equals 1, indicating that the classification is complete. The corresponding inference rules can thus be formulated: Table 3Analysis of factors after removing the decision domain for the second time.Universe UCondition factor FOutcome Gf1f2f3f4f5f6f7 u 6 32141411 u 14 12412134 u 15 32413133d(fi)0.3300.330.3310.330.33 Analysis of factors after removing the decision domain for the second time. Overall, the analysis demonstrates that conditional factors f2, f3, and f5—namely coal thickness variation, geological structure, and floor coal thickness—have high determination degrees for rockburst risk in coal mines. Accordingly, the following conclusions can be drawn: when the geological structure is moderate or relatively complex, the hazard level is moderate; when the geological structure is simple and coal thickness variation is small, the hazard level is weak; when the geological structure is complex and coal thickness variation is relatively large, the hazard level is strong; when the geological structure is complex and coal thickness variation is large, the hazard level is strong; when the geological structure is simple, coal thickness variation is moderate, and no floor coal is present, the hazard level is none; when the geological structure is complex, coal thickness variation is moderate, and the floor coal is thin, the hazard level is strong; and when the geological structure is complex, coal thickness variation is moderate, and the floor coal is thick, the hazard level is moderate (Table 4). Table 4Inference rules for rockburst risk.f3f2f5 g 2333112434444121142244233 Inference rules for rockburst risk. Field application Engineering background Yadian Coal Mine is located in Xianyang City, Shaanxi Province. Fold structures are well developed within the mining field, whereas faults are scarce and no magmatic intrusions are present. The mine adopts a vertical shaft with single-level development, with a designed production capacity of 4.0 Mt/a. The ground elevation ranges from + 864 to + 1252 m. The main mineable coal seams are the No. 1 and No. 4 seams. The No. 4 seam occurs at depths of 427 ~ 850 m, with a dip angle of 0 ~ 7° and a thickness ranging from 0.15 to 20.87 m (average 12.07 m). The main roof consists of fine- to siltstone, while the immediate roof is dominated by mudstone and sandy mudstone. Rockburst tendency identification indicates that the No. 4 coal seam and its roof exhibit weak rockburst proneness. The ZF1420 longwall panel is located in the eastern wing of the first mining district at the + 388 m level and extracts the No. 4 coal seam. The western side of the panel is bounded by three down-dip roadways of the No. 4 seam; the northern side adjoins the mine boundary, where a boundary protective coal pillar is retained; the southern side borders the ZF1418 panel, with a 40 m section protective coal pillar; and the eastern side adjoins the second mining district. The ZF1420 panel adopts a three-entry layout, consisting of a belt entry, a return air entry, and a drainage roadway, all driven along the floor of solid coal. The designed panel length along strike is 1600 m, with a dip length of 200 m. Fully mechanized longwall mining with top-coal caving is employed. Field investigations indicate severe deformation of the surrounding rock in the retreat roadways, manifested by floor heave, rib squeezing, shoulder spalling, roof subsidence, and failure of rock bolts and cables, accompanied by intermittent coal bursting sounds. At present, Yadian Coal Mine has a relatively short mining history, with limited extracted and developed panels, and no rockburst accidents have been recorded. Therefore, the potential rockburst risk is inferred based on the occurrence characteristics observed in adjacent mines with longer mining histories. Gaojiapu Coal Mine, located to the east of Yadian Coal Mine, extracts the same No. 4 coal seam, which is deeply buried and characterized by extremely complex hydrogeological conditions, and is classified as a strongly rockburst-prone mine. Since coal exposure operations began in July 2014, Gaojiapu Coal Mine has experienced frequent high-energy events (> 10⁵ J). Given that Yadian Coal Mine is adjacent to Gaojiapu Coal Mine, extracts the same coal seam, and shares similar geological conditions, it is therefore subject to a potential risk of rockburst. Evaluation of rockburst risk of ZF1420 working face Taking the ZF1420 working face of Yadian Coal Mine as a case study, and based on the aforementioned factor reduction results, coal seam thickness variation (f2), geological structure (f3), and floor coal thickness (f5) were selected as the core discriminant indicators to evaluate rockburst risk. Field engineering parameters were primarily obtained from geological exploration data, mining design documents, and on-site monitoring information. Specifically, coal seam thickness variation was determined as “moderate” (f2 = 2) based on borehole logs and geological profiles; geological structure was classified as “complex” (f3 = 4) according to the development of faults and folds near the working face; and the measured floor coal thickness of approximately 2.4 m was categorized as “thick” (f5 = 3). These parameters were input into the rule base corresponding to the factor reduction model, yielding a result factor of G = 3, which corresponds to a “moderate rockburst risk.” To verify the reliability of the model, the result was compared with evaluations from working faces under similar conditions within the same mining area. The consistency in hazard classification confirms the robustness and engineering applicability of the proposed method. Compared with traditional methods such as the Analytic Hierarchy Process (AHP) and comprehensive index methods, the proposed approach eliminates the need for subjective weighting of factors. Instead, it directly extracts key controlling variables through factor reduction and determines hazard levels based on deterministic inference rules, thereby significantly reducing human-induced bias. By identifying dominant factors, the method simplifies a complex multidimensional index system into a small set of critical variables, enabling rapid preliminary hazard assessment under limited data conditions. Moreover, the method is not restricted to static evaluation of a single working face; it can be integrated with dynamic monitoring data (e.g., coal burst frequency and microseismic energy) to enable real-time updating. When monitoring indicators change, updating the corresponding factor states allows rapid reassessment of hazard levels, providing effective support for dynamic risk management in the field. Prevention measures for rockburst risk Based on the different coal burst (rockburst) risk levels obtained from the evaluation, targeted prevention and control measures for rockbursts in the Yadian Coal Mine are formulated. As shown in Table 5, combined with the geological conditions of the mine, different working faces or zones can be classified into rockburst risk zones. Where conditions permit, regional pressure-relief measures should be preferentially adopted, aiming to eliminate potential safety hazards over a broader area. For local zones or under specific conditions, corresponding localized pressure-relief measures should be implemented to reduce the level of risk. Table 5Classification of rockburst risk and prevention measures in Yadian Coal Mine.Rockburst risk levelPrevention measuresNo rockburst riskAll mining work can be carried out according to the operating regulationsWeak rockburst risk①Adopt a single local crisis relief measures②Strengthen the monitoring of the hazard state of rockburst in the operation, regular inspection, and prepare emergency materialsModerate rockburst risk①Take two or more combinations of local crisis relief measures②Strengthen the monitoring of the hazard state of rockburst in the operation, regular inspection, and prepare emergency materialsStrong rockburst risk①Take comprehensive and local measures to address the crisis②During the operation, the monitoring of the hazard state of rockburst should be strengthened, regular inspection should be carried out, the detection index should exceed the limit, mining operations should be stopped, and unnecessary personnel should be evacuated from the dangerous place③Comprehensive measures and methods under special conditions should be adopted according to expert advice. After the measures are taken, the next operation can be carried out only after expert appraisal④Strengthen the strong support and structural support of the roadway, take measures such as large-diameter drilling pressure relief, coal seam deep hole blasting pressure relief, roof hydraulic fracturing pressure relief Classification of rockburst risk and prevention measures in Yadian Coal Mine. ①Adopt a single local crisis relief measures ②Strengthen the monitoring of the hazard state of rockburst in the operation, regular inspection, and prepare emergency materials ①Take two or more combinations of local crisis relief measures ②Strengthen the monitoring of the hazard state of rockburst in the operation, regular inspection, and prepare emergency materials ①Take comprehensive and local measures to address the crisis ②During the operation, the monitoring of the hazard state of rockburst should be strengthened, regular inspection should be carried out, the detection index should exceed the limit, mining operations should be stopped, and unnecessary personnel should be evacuated from the dangerous place ③Comprehensive measures and methods under special conditions should be adopted according to expert advice. After the measures are taken, the next operation can be carried out only after expert appraisal ④Strengthen the strong support and structural support of the roadway, take measures such as large-diameter drilling pressure relief, coal seam deep hole blasting pressure relief, roof hydraulic fracturing pressure relief Discussion This study identifies coal thickness variation, geological structures, and floor coal thickness as the key controlling factors influencing rockburst risk through a factor reduction method. These findings are broadly consistent with existing research, while also providing an extension in terms of quantitative expression and factor selection methodology. Regarding geological structures, previous studies generally regard faults and folds as core triggers of rockbursts. Zhang et al. pointed out that structural stress concentration is a primary driving source of rockburst occurrence15. Jiang et al., based on a dynamic disaster mechanics model, demonstrated that fault activation can significantly enhance energy release intensity22. In the initial analysis, the tectonic factor (f3) exhibits the highest degree of determination (d = 0.40), markedly exceeding that of mining depth. Its preferential selection during the factor reduction process underscores that tectonic stress possesses a superior explanatory capacity for rockburst susceptibility. This finding further confirms that under intense tectonic stress conditions, the energy evolution within the rock mass is predominantly driven by tectonic fields. Consequently, the gravitational stress factor represented by mining depth shares a certain degree of information overlap with tectonic attributes, rendering its independent contribution relatively limited. For coal seam thickness-related factors, traditional studies have mainly focused on the influence of absolute coal seam thickness on burst tendency. However, recent research has increasingly recognized that variations in coal thickness, especially abrupt changes, play a more critical role in local stress redistribution. Rong et al. indicated that coal thickness variation zones are prone to stress concentration and rockburst occurrence16. The results of this study show that coal thickness variation (f2) becomes a key factor in the second reduction stage, with its decisiveness significantly increasing (d = 0.67). This suggests that, compared with absolute thickness, thickness variation is more sensitive in affecting rockburst risk, which is consistent with existing understandings. Regarding floor coal thickness, existing studies typically analyze it from the perspectives of roadway floor heave and energy release pathways, suggesting that thicker floor coal may act as a weak plane for energy dissipation51. In this study, floor coal thickness (f5) is identified as a decisive factor (d = 1) in the final reduction stage, indicating that under specific geological conditions, the floor structure plays a critical controlling role in rockburst risk. This finding supplements the insufficient attention given to floor-related factors in previous research. It should be noted that some factors traditionally considered important (such as mining depth and roof properties) were not included in the final key factor set in this study. This does not imply that their influence is negligible, but rather suggests that, under the selected dataset and research conditions, their effects may be indirectly reflected through other variables such as geological structures or coal thickness variation, indicating a certain degree of information redundancy. This finding indirectly demonstrates the advantage of the factor reduction method in identifying dominant factors and eliminating redundant information. In summary, the results of this study not only validate the fundamental understanding of burst controlling factors reported in previous research, but also achieve quantitative screening and structural optimization of key factors through a factor reduction approach. This provides a more concise and interpretable analytical framework for rockburst risk assessment. Conclusion The factor reduction method based on factor space theory effectively identifies the key factors controlling rockburst, overcoming the factor redundancy and weight dispersion inherent in traditional evaluation systems, thereby improving both accuracy and practicality. The factor reduction method based on factor space theory effectively identifies the key factors controlling rockburst, overcoming the factor redundancy and weight dispersion inherent in traditional evaluation systems, thereby improving both accuracy and practicality. Based on factor reduction, seven influencing factors of rockburst are analyzed to identify the key disaster-causing factors, and a rockburst risk evaluation method with explicit inference rules was established, enabling precise classification of working-face hazard levels. Based on factor reduction, seven influencing factors of rockburst are analyzed to identify the key disaster-causing factors, and a rockburst risk evaluation method with explicit inference rules was established, enabling precise classification of working-face hazard levels. Application of the model to the ZF1420 working face in Yadian Coal Mine classified it as having a moderate rockburst risk, consistent with the results obtained from the comprehensive index method. This agreement verifies the engineering applicability of the proposed approach. On this basis, graded prevention and control measures were proposed, providing a scientific and technical reference for rockburst mitigation in mines with similar geological conditions. Application of the model to the ZF1420 working face in Yadian Coal Mine classified it as having a moderate rockburst risk, consistent with the results obtained from the comprehensive index method. This agreement verifies the engineering applicability of the proposed approach. On this basis, graded prevention and control measures were proposed, providing a scientific and technical reference for rockburst mitigation in mines with similar geological conditions. Author contributions J.Z.: Conceptualization, Supervision, Project administration, Investigation, Writing—Original draft preparation. B.W.: Conceptualization, Data Curation, Funding acquisition. D.Z.: Validation, Writing—Review & Editing. All authors have read and agreed to the published version of the manuscript. Funding The works described in this paper are financially supported by National Natural Science Foundation of China (Grant No. 52374092), Basic Scientific Research Program of the Educational Department of Liaoning Province (Grant No. LJ212410147086). Data availability All data generated or analyzed during this study are included in this published article. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement All data generated or analyzed during this study are included in this published article.