Hazardous High-Energy Seismic Event Discrimination Method Based on Region Division and Identification of Main Impact Factors: A Case Study

Abstract

An investigation of risk factors has been identified as a crucial aspect of the routine management of rockburst. However, the identification methods for principal impact factors and the examination of the relationship between seismic energy and other source parameters have not been extensively explored to conduct dynamic risk management. This study aims to quantify impact risk factors and discriminate hazardous high-energy seismic events. The analytic hierarchy process (AHP) and entropy weight method (EWM) are utilized to ascertain the primary control factors based on geotechnical data and nearly two months of seismic data from a longwall panel. Furthermore, the distribution law and correlation relationship among seismic source parameters are systematically analyzed. Results show that the effect of coal depth, coal seam thickness, coal dip, and mining speed covers the entire mining process, while the fault is only prominent in localized areas. There are varying degrees of log-positive correlations between seismic energy and other source parameters, and this positive correlation is more pronounced for hazardous high-energy seismic events. Utilizing the linear logarithmic relationship between seismic energy and other source parameters, along with the impact weights of dynamic risks, the comprehensive energy index for evaluating high-energy seismic events is proposed. The comprehensive energy index identification method proves to be more accurate by comparing with the high-energy seismic events based on energy criteria. The limitations and improvements of this method are also synthesized to obtaining a wide range of applications.

1. Introduction

Rockburst is a common dynamic disaster phenomenon in the process of coal mine production [1,2,3,4,5,6]. In the process of coal mining, the elastic energy accumulated in the coal and rock mass in the high-stress state is suddenly released, resulting in significant damage to the surroundings [7]. This release is often accompanied by violent sound and vibration, posing a great threat to the personnel and equipment in the roadway, and its influence can even spread to the surface. At the same time when a rockburst occurs, it may also induce other mine disaster phenomena, such as coal and gas protrusion, coal dust explosion, coal rock layer water burst, etc. [6]. Consequently, rockburst has been identified as a principal factor impeding the safe production of coal mines.

Recent research has increasingly concentrated on the identification of main impact factors in many ways. Patyńska [8] analyzed the relationship between rockburst events and mining geological factors and concluded that the depth of mining is the most important influence factor. Gu et al. [9] investigated the impact factors of the Dongpang Coal Mine, encompassing geological and mining technology factors, and established a hierarchical structural model for dynamic hazard assessment. Wang and Li [10] established a rockburst criterion for the wall rock of an anchored tunnel under dynamic disturbance and analyzed the main control factors under the support condition. A database of water-surge disasters in longwall mining within Chinese coal mines was established by Zhao et al. [11], who also summarized the characteristics and causative factors of these accidents. Xu et al. [12] analyzed and summarized the basic characteristics and causative factors of rockburst damage in a large number of roadways, explored the main sources of energy released by rockbursts and the main controlling factors, and put forward the classification method and the principle of prevention and control in stages. Nan and Zhou [13] analyzed the degree of influence of hazardous factors in complex formations of extra-thick coal seams, such as split rock and alternating soft and hard coal conditions. The degree of dynamic hazard, in terms of stress source and path, was analyzed by Qiu et al. [14] through combining the occurrence characteristics of the deep mining face and mining conditions. However, it has been noted that existing studies predominantly focus on the rockburst mechanism and depend on subjective judgment for analyzing the principal control factors, failing to quantify the weights of hazardous factors.

Source parameters, such as frequency, energy, source radius, apparent volume, and apparent radius, can be utilized to characterize the degree of damage of the source in terms of rupture intensity, scale of disturbance, and degree of stress adjustment. This enables the assessment of the risk posed by high-energy events or the conduct of early warning predictions of dynamic hazards. The efficiency of seismic event counts and average released energy for predicting dynamic hazard was analyzed by Wang et al. [15] based on 80 days of seismic data from a coal mine in China. The pattern of change in the time series of source parameters (energy density, energy index, etc.) as a precursor feature for rockburst early warning was analyzed by Yu et al. [16], and the intrinsic relationship between seismic activity and microfractures was studied. The effectiveness of coal seam blasting for mining coal seams under complex coal seam conditions in the Upper Silesian Coal Basin was evaluated by Wojtecki et al. [17] using source parameters. Mao et al. [18] investigated the sensitivity of rockburst warnings to source parameters using Bayesian networks. Source parameters—such as seismic energy, seismic moment, and apparent stress—are used by Ma et al. [19] to reveal the evolutionary characteristics of hazard types and seismic paths. The variation characteristics of the daily seismic event count, average energy level, cumulative apparent volume, and b-value of the Jinping II hydropower station tunnels were statistically analyzed by Liu et al. [20], who concluded that the accuracy of rockburst prediction can be improved by comprehensively analyzing the multi-parameter precursors of seismicity. Apparent stress, apparent volume, and main frequency value were adopted by Xue et al. [21] to analyze the activity state inside the enclosing rock. It was found that the combination of these three parameters can better describe the different damage characteristics of the surrounding rock in real-time. Different seismic source parameters characterize the risk of dynamic hazards in different ways, and existing studies have included only some of them in the risk assessment. Methods of quantifying the contribution of different seismic source parameters to the risk assessment have not been sufficiently investigated, making it difficult to comprehensively quantify the risk of rockbursts.

A new seismic-based method is proposed in this study to determine main impact factors and discriminate hazardous high-energy seismic events. Geotechnical and seismic data from a burst-prone longwall panel in a Chinese coal mine were used to refine the method. The Analytic Hierarchy Process (AHP) [22] and entropy weight method (EWM) [23] were utilized to quantify dynamic risk induced by typical geotechnical factors. The distribution law and correlation relationship among the seismic energy and other source parameters are investigated, and the comprehensive energy index evaluation method is established to distinguish hazardous high-energy seismic events. Method validation is performed using nearly two months of seismic data in the longwall mining.

2. Overview of the Research Site

2.1. Geotechnical Conditions and Seismic Monitoring

The KZ Coal Mine, situated in Xuzhou City, Jiangsu Province, China, predominantly mines the No. 7 and No. 8 coal seams [24]. The No. 7 coal seam has an average thickness of 5 m, predominantly comprising magma rock gangue. Positioned beneath the No. 7 seam, the No. 8 coal seam exhibits an average thickness of 4 m, with its distance from the overlying No. 7 seam ranging between 4.17 and 40.18 m. Characterized by a monoclinic structure inclined towards the northwest, the mine’s field encompasses a fully covered coalfield, exhibiting gentle, wave-like undulations. Within the coalfield, 49 faults have been identified, each exhibiting a displacement exceeding 20 m out of a total of 277 faults, of which 98.9% are normal faults and 1.1% are reverse faults. The predominant fault orientation is northeast, with fault dips mainly grouping into northwest and southeast directions. In the mine area, the faults exhibit a northeast strike, forming a step block pattern with northwest ascending and southeast descending features.

The Seismic Observation System (SOS), designed by the Central Mining Institute of Poland, is used in the KZ Coal Mine for activity seismic monitoring [7]. This system is capable of capturing vibration signals with energies exceeding 100 J, frequencies ranging from 1 to 600 Hz, a sampling rate of 500 Hz, and an anti-interference capability of less than 30 DB. The geophone, a vertically oriented detection unit, can transmit data at rates up to 1 MB/s, offering 16-bit A/D conversion, and is typically secured to floor bolts using a special protective device. As depicted in Figure 1, geophones are classified into two categories: fixed and moved. Fixed geophones, specifically numbers 1, 2, 3, 4, 5, 13, 14, and 15, are permanently installed at predefined locations within the mine. In contrast, moved geophones, numbers 11 and 12, are relocated as necessary to improve monitoring quality in tandem with changes in the extraction area.

2.2. The Case Study Panel

As seen in Figure 1, Longwall (LW) 7434, positioned adjacent to Goaf 7432 to the south, is bounded by the solid coal area to the west and north and the mining area downhill to the east. The mining depth for LW7434 ranges from 796 to 900 m. In terms of dimensions, LW7434 extends approximately 1237 m along the strike direction and 197 m in width along the inclination. This longwall extracts coal from the No. 7 seam, which has an average thickness of approximately 4.2 m and an inclination angle ranging from 18° to 25°, with an average of around 23°. Within LW7434, 27 faults have been exposed, among which four exhibit a displacement exceeding 3.5 m. Figure 2 depicts the general stratigraphic column.

LW7434 is susceptible to rockburst risks due to factors such as its considerable burial depth (resulting in high static load stress levels), variations in coal seam thickness and inclination, fault structures, and fluctuations in the mining rate. From the start of mining in February 2018 to the end of December 2019, 28,565 seismic events were monitored within LW7434, with 363 of these events exhibiting seismic source energies exceeding 9 × 10³ J.

3. Methodology

3.1. Framework of the Methodology

To facilitate understanding of the high-energy seismic event discrimination method, which is based on regional division and identification of primary impact factors, the framework of the methodology is illustrated in Figure 3. Initially, the main control factors in various zones of the longwall are identified based on the impact risk factors’ weight ordering, utilizing the entropy weight method (EWM) for final determination. Subsequently, seismic source parameters (SSPs) are calculated, and their correlation with the seismic source energy is analyzed during the longwall mining process. Ultimately, the comprehensive energy index (CEI) is formulated to discern high-energy seismic events, considering each SSP, and to categorize these events accordingly. SSPs include source energy, seismic moment, stress drop, apparent stress, source radius, and apparent volume, with their calculation methods and physical meanings detailed in

Table A1. Common source parameters and physical meanings.

Source Parameter	Parameter Meaning	Formula for the Calculation	Parameter Information
Corner frequency	Preliminarily judge the size of the mining-induced earthquake scale; the larger the scale, the richer the low-frequency components of the spectrum.	The frequency corresponding to the intersection of the progressive high frequency trend and low frequency level of the source amplitude spectrum in logarithmic coordinates.	ρ = medium density in source area; c = wave velocity in source area; R = the distance between source and station; Ω0 = level of spectrum at low frequencies; F = radiation pattern (P-waves = 0.52, S-waves = 0.63); k = constant for the Brune model (2.34); fc = corner frequency; μ = shear modulus in source area; Es = seismic energy.
Scalar seismic moment	Describe the source intensity defined by the source couple model.	$M_0=\frac{4\pi \rho c^{3}R{\mathrm{\Omega }}_0}{F}$
Source radius	Characterize the influence range of source rupture.	$r_c=\frac{kc}{2\pi f_c}$
Apparent volume	Characterize the volume of inelastic deformation of the source.	$v_a=\frac{{M_0}^2}{2\mu E_s}$
Stress drop	Characterize the level of stress reduction before and after an earthquake.	$\Delta \sigma =\frac{7M_0}{16{r_c}^3}$
Apparent stress	Characterize the stress level at the source after the occurrence of an earthquake.	${\sigma }_{app}=\mu \frac{E_s}{M_0}$

of Appendix B. The detailed SSP calculations are discussed in previous research [25].

3.2. Method for Identification of Main Impact Factors

The analytic hierarchy process (AHP) [22] can be applied to multi-objective fuzzy evaluation decision-making and has now become an important tool for statistical analysis in many disciplines. AHP has the characteristics of combining qualitative and quantitative analysis. However, it also has certain limitations—determining the weights of different factors according to human-subjective factors has a greater impact and does not apply to the analysis of higher-precision decision-making [26]. Therefore, in this section, after using the AHP to initially determine the weight matrix of impact risk factors, the EWM [23] is used to further determine the weight ranking of each factor. The specific steps are shown below:

Step 1: Modeling of hierarchical structures. To thoroughly analyze the research objectives and elucidate the interrelationships and the extent of influence of the factors within the evaluation system, it is imperative to delineate the hierarchical relationships, both upper and lower, of the evaluation indices. This delineation should be based on the attributes and the positioning of each factor within the evaluation system, ultimately resulting in a tree-diagram-like subordinate structural relationship.

Step 2: Constructing judgment matrices. This paper adopts a single-level model structure. The judgment matrix can be constructed as shown in Equation (1).

$$P=\left[\begin{array}{cccc}{a}_{11}& a_{12}& \cdots & a_{1j}\\ a_{21}& a_{22}& \cdots & a_{2j}\\ \cdots & \cdots & \cdots & \cdots \\ a_{i1}& a_{i2}& \cdots & a_{ij}\end{array}\right]$$

(1)

where a_i, a_j (i, j = 1, 2, 3, …, n) are the indicator factors for the evaluation. The meaning of a_ij is a_i relative to the relative influence or importance of a_j, from a_ij. Within the hierarchical evaluation system, the degree of influence or importance of all factors at the same level is assessed. Furthermore, based on the criteria for value assignment in the consistency matrix (

Table 1. Scale value meaning of analytic hierarchy process.

Scale	Meaning
1	Indicates that the two factors are of equal importance compared to each other
3	Indicates that one factor is slightly more important than the other when comparing two factors
5	Indicates that one factor is significantly more important than the other when comparing two factors
7	Indicates that one factor is more strongly important than the other when comparing two factors
9	Indicates the extreme importance of one factor over the other when comparing two factors
2, 4, 6, 8	The median level of influence represented by the two neighboring scales above
Multiplicative inverse	aij of factor i compared to j, then the judgment of factor j compared to i aij = 1/ai

), the individual evaluation factors are quantified. This quantification leads to the formation of the final judgment matrix, which is utilized for evaluating the influence factors.

Step 3: Compute the weight vector. Utilizing the judgment matrix, compute the eigenvector corresponding to the largest eigenvalue as shown in Equation (2). Normalization is then performed to obtain the weight ordering of each factor.

$$Pw={\lambda }_{\max }w$$

(2)

where ω is the eigenvector and λ_max is the corresponding largest eigenvalue.

Step 4: Consistency check. To verify the accuracy of the weighting coefficients, the results should be examined using the consistency ratio CR (Equation (3)). The judgment matrix is deemed to have acceptable consistency and passes the consistency test if CR < 0.1.

$$CR=\frac{CI}{RI}=\frac{{{\displaystyle \lambda }}_{\max }-n}{\left(n-1\right)RI}$$

(3)

where n denotes the order of the judgment matrix. CI is a consistency indicator. RI denotes the random consistency index, and the values are shown in

Table 2. Average random consistency indicator.

Ordinal Number (n)	1	2	3	4	5	6	7	8
RI	0	0	0.52	0.89	1.12	1.26	1.36	1.14

.

After the initial determination of the weights of the rockburst risk factors, the EWM was used to determine the weight ordering steps of the factors as follows:

Step 5: Construct the original matrix. Assuming that there are i evaluation objects and j evaluation indicators, the original matrix is as follows:

$$X=\left[\begin{array}{cccc}{x}_{11}& x_{12}& \cdots & x_{1j}\\ x_{21}& x_{22}& \cdots & x_{2j}\\ \cdots & \cdots & \cdots & \cdots \\ x_{i1}& x_{i2}& \cdots & x_{ij}\end{array}\right]$$

(4)

where x_ij is the weight of the j-th indicator for the i-th element in the judgment matrix.

Step 6: Non-negativization of data. Normalization is required when the evaluation indicators are inversely proportional and moderate. Additionally, data leveling must be performed to prevent the meaninglessness of the logarithm during entropy value calculation. The resulting data matrix is shown in Equation (5); the r_ij is the indicator of x_ij after the normalization.

$$X=\left[\begin{array}{cccc}{r}_{11}& r_{12}& \cdots & r_{1j}\\ r_{21}& r_{22}& \cdots & r_{2j}\\ \cdots & \cdots & \cdots & \cdots \\ r_{i1}& r_{i2}& \cdots & r_{ij}\end{array}\right]$$

(5)

For the positive impact indicators:

$$r_{ij}=\frac{x_{ij}-x_i^{\min }}{x_i^{\max }-x_i^{\min }}+1$$

(6)

For the negative impact indicators:

$$r_{ij}=\frac{x_i^{\max }-x_{ij}}{x_i^{\max }-x_i^{\min }}+1$$

(7)

Step 7: The weight of the ith sample under the jth indicator for that indicator is calculated as follows:

$$f_{ij}=\frac{r_{ij}}{{\displaystyle \sum _{i=1}^n{r}_{ij}}}$$

(8)

where r_ij denotes the value of the jth indicator of the ith evaluation indicator object; n denotes the number of evaluated objects.

Step 8: The entropy value of the jth indicator, H_j, is calculated as follows:

$$H_j=-k{\displaystyle \sum _{i=1}^n{f}_{ij}\ln }{f}_{ij}$$

(9)

where k = 1/lnn, 0 ≤ Hj ≤ 1.

Step 9: Indicator weights are calculated according to Equation (10):

$$w_j=\frac{1-H_j}{{\displaystyle \sum _{j=1}^m\left(1-H_j\right)}}\left(j=1,2\dots m\right)$$

(10)

where the coefficient of variation for indicator x_j, denoted as 1-H_j, increases as the difference of x_ij becomes larger; this indicates a greater contribution of x_j to the comprehensive evaluation. Conversely, a lack of difference in x_j among the evaluated objects results in H_j equating to 1 and the coefficient of variation becoming 0. If the difference of x_ij is smaller, H_j increases, leading to a decrease in the coefficient of variation. The term m represents the number of evaluation indexes.

Step 10: The weights of the evaluation indicators calculated by the entropy method are normalized to regain the weight set A = [a₁, a₂, …, a_m] and used as the final weight ranking of the risk factors.

3.3. Method for High-Energy Seismic Events Discrimination

Seismic energy has emerged as the primary indicator for identifying high-energy events, yet it remains the sole criterion. This method reflects only the energy magnitude radiated from the seismic source’s rupture and does not adequately describe the seismic source medium’s perturbation scale or the stress adjustment at the rupture surface. Moreover, it overlooks the impact of other SSPs on dynamic hazard assessment. Thus, while the seismic energy-based criterion is relatively reliable, it serves merely as a sufficient condition for identifying hazardous high-energy events. This evaluation approach also risks overlooking potentially hazardous events and cannot precisely differentiate the relative hazards of mining seismic events with comparable seismic energy levels.

This section, predicated on the energy conditions for identifying hazardous high-energy events, examines the correlation between SSPs, selects appropriate evaluation parameters for these events, computes the fitting functions and influence weights for evaluation parameters and seismic energy, and formulates identification criteria based on the seismic energy and SSPs correlation. Consequently, hazardous high-energy events can be assessed accurately and judiciously. Furthermore, in light of the current criteria for evaluating the energy of hazardous high-energy events (referenced in

Table 3. Energy criterion of hazardous high-energy events.

Seismic Energy/J	Risk Level
Es < 104	Low
104 ≤ Es < 105	Moderate
Es ≥ 105	Strong

), a comprehensive energy index (CEI) is formulated, as outlined in

Table 4. Comprehensive energy index (CEI) of hazardous high-energy events.

CEI (CEI = lgEs)	Risk Level
CEI < 4	Low
4 ≤ CEI < 5	Moderate
CEI ≥ 5	Strong

, incorporating various SSPs.

The calculation process for the CEI involves determining the correlation of the SSPs through analytical calculations of the fitting function and influence weight of SSP with seismic energy. Subsequently, the energy index corresponding to each SSP is derived from the correlation relationship between each SSP of the seismic source and the seismic energy, as delineated in Equation (11).

$$W_i=\frac{\lg m_i-b_i}{a_i}$$

(11)

where a_i is the slope of the corresponding fit function and b_i is the intercept of the corresponding fit function.

Upon establishing the fitted relationship between SSPs and seismic energy, the strength of the linear correlation between these two data sets can be measured further by the correlation coefficient R, calculable through Equation (12).

$$R=\frac{\mathrm{Cov}(\mathrm{X},\mathrm{Y})}{\sqrt{\mathrm{Var}\left[\mathrm{X}\right]\times \mathrm{Var}\left[\mathrm{Y}\right]}}$$

(12)

where Cov(X, Y) is the covariance of the two sets of data and Var[X], Var[Y] are the variances of each of the two sets of data.

The correlation coefficient calculations for each parameter indicate that the seismic moment, stress drop, and apparent stress are strongly related to the seismic energy and thus can serve as key parameters for subsequent dynamic hazard evaluations. In contrast, the apparent volume and source radius exhibit low responses to the seismic energy, qualifying them as non-key mechanical parameters for further evaluations. Additionally, the influence weights of each parameter on the dynamic seismic hazard are calculable using Equation (13).

$$w_i=r_i/{\displaystyle \sum _{i=1}^n{r}_i}$$

(13)

The CEI is calculated using Equation (14), based on the influence weights of SSPs and their fitting results with source energy. Seismic event risk is evaluated through the combined CEI and the corresponding risk levels detailed in Table 4.

$$CEI={\displaystyle \sum _{i=1}^n{w}_i{W}_i}$$

(14)

4. A Case Study of the LW7434

This section delineates the method proposed in this paper for discriminating hazardous high-energy seismic events, based on the engineering geological conditions of LW7434 and nearly two months of monitored seismic data collected during the mining period.

4.1. Relationship between Main Impact Factors and the Seismic Events

Research indicates that regions exhibiting abnormally high stress, such as those affected by geological and tectonic anomalies, tend to experience numerous seismic events during mining. Consequently, areas with stress anomalies can be preliminarily identified through the distinctive characteristics of these seismic events [26]. In this section, seismic data collected from LW7434 between February 2018 and December 2019 serve as the basis for a statistical analysis of the geological and technical factors significantly influencing rockburst risks. This section aims to identify and determine the principal factors impacting LW7434, thereby providing a foundation for establishing the main control factors in subsequent analyses.

Figure 4 illustrates the correlation between variations in coal seam depth and the distribution of seismic events. Seismic events within each mining depth range exhibit varying degrees of clustering. Specifically, within the depths of 790–830 m, 830–870 m, and 870–910 m, there were 80, 165, and 118 seismic events, representing 22%, 45%, and 33% of the total events, respectively. Additionally, areas with mining depths exceeding 830 m demonstrate a significant increase in seismic event accumulation and distribution. Furthermore, the distribution of seismic events is positively correlated with the burial depth of coal seams in specific areas. A comparative analysis of high mining depth areas during the initial and final phases of mining reveals distinct patterns of agglomeration and dispersal, respectively, suggesting that coal seam depth does not exert an absolute influence on dynamic hazards.

Figure 5 depicts the correlation between variations in coal seam thickness and seismic event distribution in LW7434. The thickness of the coal seam increases progressively from west to east. In areas where the coal seam thickness exceeds 4.4 m, 243 seismic events were recorded, constituting 67% of the total. However, in zones where the thickness surpasses 4.6 m, only 20 seismic events were observed, indicating a lack of a definitive correlation between coal seam thickness and seismic event distribution. In addition, regions with significant fluctuations in coal seam thickness exhibited concentrated seismic activity, suggesting that the extent of thickness variation significantly affects seismic event distribution. Therefore, while the impact of coal seam thickness on mine pressure is somewhat limited, the correlation between changes in coal seam thickness and seismic event distribution is notably stronger.

Figure 6 illustrates the relationship between variations in coal dip and seismic event distribution. The inclination map, presented in Figure 6a, reveals that seismic activity is primarily concentrated in areas with a dip range of 20° to 25°, where a total of 329 seismic events occurred, comprising 90.6% of the overall event count. Nonetheless, in the initial stages of mining, seismic events are more scattered across regions with steeper dip angles, indicating that the dip angle has a localized correlation with seismic event distribution. Consequently, it is difficult for the dip angle to significantly influence the seismic event distribution across the entire area.

The inclination map along the strike direction, as shown in Figure 6b, indicates that the overall strike of LW7434 falls within the range of 3 to 5°. In this region, 142 seismic events occurred in areas with a strike dip exceeding 4°, and 75 events occurred in areas where the dip exceeded 4.5°, constituting 39% and 21% of the total seismic events, respectively. Seismic events are more evenly distributed across the various strike dip ranges, demonstrating a lower correlation between the distribution of strike dips and seismic events. Therefore, in the quantification of subsequent weights, the inclination of the coal bed should be considered with moderate importance.

Figure 7 illustrates the relationship between the variation of key parameters and the seismic event distribution associated with faults during the mining of LW7434. The correlation between fault density and seismic event distribution is more pronounced than that with fault strength. Notably, when a high fault density zone extends throughout the entire longwall, a significant accumulation of seismic events is observed, indicating that the number of fault structures exerts a considerable influence on the distribution of seismic events.

Figure 8 displays the relationship between the monthly advancement of LW7434 and the frequency of seismic events. It reveals that from October 2018 to March 2019 and from July 2019 to December 2019, there is a notable correlation between the longwall mining speed and the frequency of seismic events. This observation suggests that longwall mining speed significantly impacts localized areas, while its influence varies considerably across the entire longwall range. Consequently, the significance of mining speed in identifying the main control factors of seismic hazard differs.

It follows from the above statement that the coal depth, coal seam thickness, coal dip, and mining speed are the overall factors, which means that these factors pose different levels of dynamic hazards during mining. In addition, the influence of mining speed covers the whole mining stage of the longwall, while the influence of fault structure is only prominent in the local area.

4.2. Identification and Weighting of Main Impact Factors

4.2.1. Regional Division

Based on the analysis of Section 4.1 on the distribution of each impact factor and seismic events distribution in LW7434, the longwall is divided into five regions according to the strength of the influence degree of each factor, as shown in Figure 9, and the influence degree of each factor in different regions of the longwall is shown in

Table 5. Scale value of rock burst impact factors of LW7434.

Region No.	Depth	Coal Seam Thickness	Coal Dip	Mining Speed	Fault
A	1	1	1	3	4
B	1	3	1	3	5
C	1	1	1	6	5
D	1	1	1	4	7
E	1	2	1	6	7
F	1	2	1	6	4

.

4.2.2. Quantification of the Weights of the Factors in Each Region of LW7434

The judgment matrix of influence factors in each region was constructed based on the correlation between the influence factors and the distribution of seismic events in each region, utilizing the hierarchical analysis method comparison scale criterion. The judgment matrix of the impact factors in different regions of LW7434 is displayed in

Table 6. Judgment matrix of the impact factors in different regions of LW7434.

Region A	Factor	Depth	Coal seam thickness	Coal dip	Mining speed	Fault
	Depth	1	1	1	1/3	1/4
	Coal seam thickness	1	1	1	1/3	1/4
	Coal dip	1	1	1	1/3	1/4
	Mining speed	3	3	3	1	1/2
	Fault	4	4	4	2	1
Region B	Factor	Depth	Coal seam thickness	Coal dip	Mining speed	Fault
	Depth	1	1	1/3	1/3	1/5
	Coal seam thickness	1	1	1/3	1/3	1/5
	Coal dip	3	3	1	1	1/2
	Mining speed	3	3	1	1	1/2
	Fault	5	5	2	2	1
Region C	Factor	Depth	Coal seam thickness	Coal dip	Mining speed	Fault
	Depth	1	1	1	1/4	1/6
	Coal seam thickness	1	1	1	1/4	1/6
	Coal dip	1	1	1	1/4	1/6
	Mining speed	4	4	4	1	1/2
	Fault	6	6	6	2	1
Region D	Factor	Depth	Coal seam thickness	Coal dip	Mining speed	Fault
	Depth	1	1	1	1/4	1/7
	Coal seam thickness	1	1	1	1/4	1/7
	Coal dip	1	1	1	1/4	1/7
	Mining speed	4	4	4	1	1/2
	Fault	7	7	7	2	1
Region E	Factor	Depth	Coal seam thickness	Coal dip	Mining speed	Fault
	Depth	1	1	1/2	1/6	1/7
	Coal seam thickness	1	1	1/2	1/6	1/7
	Coal dip	2	2	1	1/4	1/4
	Mining speed	6	6	4	1	1
	Fault	7	7	4	1	1
Region F	Factor	Depth	Coal seam thickness	Coal dip	Mining speed	Fault
	Depth	1	1	1/2	1/4	1/6
	Coal seam thickness	1	1	1/2	1/4	1/6
	Coal dip	2	2	1	1/2	1/3
	Mining speed	4	4	2	1	1/2
	Fault	6	6	3	2	1

.

The results of the calculations for the different regions are listed below:

• Calculated results for region A: CI = 0.047, RI = 1.12, λ_max = 5.047, CR = 0.007 < 0.1, which satisfies the consistency test, and the weight matrix W = [0.099, 0.099, 0.099, 0.274, 0.429] is calculated to obtain the weights of the depth, coal seam thickness, coal dip, mining speed, and fault.
• Calculated results for region B: CI = 0.011, RI = 1.12, λ_max = 5.011, CR = 0.011 < 0.1, which satisfies the consistency test, and the weight matrix W = [0.076, 0.076, 0.220, 0.220, 0.409] is calculated to obtain the weights of the depth, coal seam thickness, coal dip, mining speed, and fault.
• Calculated results for region C: CI = 0.025, RI = 1.12, λ_max = 5.025, CR = 0.006 < 0.1, which satisfies the consistency test, and the weight matrix W = [0.076, 0.076, 0.076, 0.288, 0.484] is calculated to obtain the weights of the depth, coal seam thickness, coal dip, mining speed, and fault.
• Calculated results for region D: CI = 0.006, RI = 1.12, λ_max = 5.006, CR = 0.001 < 0.1, which satisfies the consistency test, and the weight matrix W = [0.071, 0.071, 0.071, 0.277, 0.510] is calculated to obtain the weights of the depth, coal seam thickness, coal dip, mining speed, and fault.
• Calculated results for region E: CI = 0.021, RI = 1.12, λ_max = 5.021, CR = 0.005 < 0.1, which satisfies the consistency test, and the weight matrix W = [0.058, 0.058, 0.106, 0.377, 0.401] is calculated to obtain the weights of the depth, coal seam thickness, coal dip, mining speed, and fault.
• Calculated results for region F: CI = 0.024, RI = 1.12, λ_max = 5.024, CR = 0.005 < 0.1, which satisfies the consistency test, and the weight matrix W = [0.071, 0.071, 0.141, 0.268, 0.449] is calculated to obtain the weights of the depth, coal seam thickness, coal dip, mining speed, and fault.

4.2.3. Determination of the Degree of Dynamic Hazard in Each Area of LW7434

The evaluation matrix for the degree of factors on the dynamic risk in each region of the longwall is constructed by selecting five factors: depth, coal seam thickness, coal dip, mining speed, and fault. This matrix is shown in

Table A2. Judgment matrix of the impact factors of LW7434.

	Factor	Depth	Coal Seam Thickness	Coal Dip	Mining Speed	Fault
Region
A		0.099	0.099	0.099	0.274	0.429
B		0.076	0.220	0.076	0.220	0.409
C		0.076	0.076	0.076	0.288	0.484
D		0.071	0.071	0.071	0.277	0.510
E		0.058	0.058	0.106	0.377	0.401
F		0.071	0.071	0.141	0.268	0.449

of Appendix B. The weights of each main impact factor were calculated according to Equations (5)–(8), as shown in

Table A3. Different factors in each region for the proportion of this indicator of LW7434.

	Factor	Depth	Coal Seam Thickness	Coal Dip	Mining Speed	Fault
Region
A		0.220	0.166	0.174	0.161	0.160
B		0.169	0.370	0.134	0.129	0.152
C		0.169	0.128	0.134	0.169	0.180
D		0.157	0.119	0.125	0.163	0.190
E		0.129	0.097	0.186	0.221	0.150
F		0.157	0.119	0.248	0.157	0.167

of Appendix B.

The entropy value H_j and indicator weight w_j of each risk factor are calculated based on the weights of the impact factors in each region and according to Equations (9) and (10), with n = 6 representing the number of dividing regions (illustrated in Figure 9). These values are then normalized, and the results are presented in

Table A4. Entropy of each influence factor of LW7434.

Factor	Depth	Coal Seam Thickness	Coal Dip	Mining Speed	Fault
Entropy value Hj	1.105	1.034	1.094	1.105	1.111

and

Table A5. The weight of each influence factor of LW7434.

Factor	Depth	Coal Seam Thickness	Coal Dip	Mining Speed	Fault
Indicator weight wj	0.23	0.07	0.21	0.23	0.25

of Appendix B.

After calculating the weights of the influence degree of each factor, a comprehensive evaluation of the impact factors on different areas is conducted according to Equation (14), with the findings presented in

Table A6. The influence degree of impact factors on each region of LW7434.

Region	A	B	C	D	E	F
influence degree	0.222	0.203	0.227	0.228	0.228	0.225

of Appendix B. It is evident that the D and E areas exhibit the largest degree of influence, while the B area has the smallest.

The synthesis of the analysis indicates that coal depth, coal seam thickness, coal dip, and mining speed are significant overall factors affecting the dynamic risk in LW7434. The first three predominantly influence local geological anomalies, whereas mining speed’s impact extends across the entire longwall. Fault acts as a local factor with a particularly pronounced effect during specific mining stages of the longwall. Analysis of the parameters associated with each factor and the seismic event distribution characteristics reveals a stronger influence of fault and mining speed on rockburst incidents compared to the relatively weaker impact of coal depth, coal seam thickness, and coal dip.

4.3. High-Energy Seismic Events Identification Results

The seismic data from the LW7434 during the mining period were used to calculate the SSPs and their relationships with the seismic energy are fitted to obtain the correlations between each SSP and the seismic energy. The results are shown in Figure 10 and

Table A7. The fitting results of SSPs and source energy.

SSPs	Fitting Functional Relationships	Standard Error of ai	Standard Error of bi
M0 vs. Es	lg M0 = 0.525 1g Es + 6.967	±0.026	±0.092
Δσ vs. Es	lg Δσ = 0.366 lg Es + 2.687	±0.029	±0.103
σapp vs. Es	lg σapp = 0.475 lg Es + 2.878	±0.026	±0.092
rc vs. Es	lg rc = 0.053 lg Es + 1.307	±0.037	±0.097
Va vs. Es	lg Va = 0.051 lg Es + 3.789	±0.052	±0.184

of Appendix B. The correlation coefficients between each SSP and the seismic energy are calculated and shown in

Table A8. The correlation coefficients between SSPs and source energy.

SSP	Es	M0	Δσ	σapp	rc	Va
R	1.000	0.746	0.658	0.727	0.329	0.228

of Appendix B.

The results of the correlation coefficient calculations for each parameter indicate that the seismic moment, stress drop, and apparent stress are strongly related to the seismic energy and thus can serve as key mechanical parameters for subsequent dynamic hazard evaluations. In contrast, the apparent volume and source radius exhibit low responses to the seismic energy, qualifying them as non-key mechanical parameters for further evaluations. Additionally, the influence weights of each parameter on the dynamic seismic hazard are calculable using Equation (13), with the calculated results presented in

Table A9. Influence weight of each SSP.

SSP	Es	M0	Δσ	σapp	rc	Va
Weight	0.271	0.202	0.178	0.197	0.089	0.062

of Appendix B.

In addition, we compare and analyze the identification results of hazardous high-energy seismic events based on the energy and CEI described in Section 3.3, which considers various source parameters. The analysis is conducted with data from 100 seismic events monitored during the near-fault mining of LW7434.

Seismic data from LW7434 during 21 October 2018 and 1 February 2019, were analyzed. The distributions of hazardous high-energy events as determined by energy and CEI are depicted in Figure 11a,b, respectively. In the energy-based discrimination result (Figure 11a), all seismic events are categorized as weak hazard class. Figure 11b,c show that 95 events were classified as weak hazards, and five as moderate hazards by the CEI. Furthermore, Figure 11d reveals that out of 100 seismic events, those categorized as hazardous high-energy events by the CEI have higher values in source parameters like energy, seismic moment, stress drop, source radius, and apparent volume, with the exception of apparent stress. This suggests that the rupture intensity, size of disturbance, and stress adjustment at the source of hazardous high-energy events are significantly more pronounced than in general seismic events. Relying solely on one source parameter, such as apparent stress, for classifying hazardous high-energy events could lead to misclassification.

5. Discussion

The study addresses a critical gap in current research by moving beyond subjective judgment in identifying principal control factors of seismic events. By employing a combination of AHP and EWM, the research provides a more objective and quantifiable approach to assess the dynamic risks associated with mining-induced seismicity. By conducting a systematic analysis of the correlation between seismic energy and various source parameters, the study establishes a comprehensive energy index that offers a more accurate and nuanced assessment of seismic hazards compared to traditional methods relying solely on seismic energy. The practical implications of this research are significant for enhancing the safety and operational efficiency of mining activities, particularly in regions prone to high-energy events. The CEI, supported by AHP and EWM, can be integrated into existing monitoring systems to improve rockburst prediction and management. This is particularly significant for the management and prevention of rockbursts.

However, because the energy criterion for hazard determination in this method heavily relies on historical data from the mine site over many years, the criterion must be modified according to the conditions of the mine site when the method is applied to a new location. Additionally, a linear relationship is used to fit the correlation between source parameters and seismic energy, considering the effects of all source parameters. Therefore, the success of this linear fitting also influences the effectiveness of the method’s implementation. Moreover, the different source mechanisms of high-energy events result in diverse source parameter responses [27]. Future studies should focus on a detailed examination of the source mechanisms of various high-energy events and the categorization of precursor event source parameters based on these mechanisms, which will be crucial for further improving the method.

6. Conclusions

Quantifying the weights of impact risk factors in the mining area and early identification of high-energy events hold significant importance for targeted prevention and control measures. This paper proposes the use of the analytic hierarchy process (AHP) and the entropy weight method (EWM) to ascertain the primary control factors of rockburst risk. It systematically analyzes the distribution law and the correlation relationship among the seismic source parameters, determining the linear relationship between each source parameter and seismic energy. Subsequently, a comprehensive energy index evaluation method for rockburst risk is established, leading to the following conclusions:

• In the case study, the coal depth, coal seam thickness, coal dip and mining speed are the overall factors. The influence of mining speed covers the whole mining stage of the longwall, while the influence of fault structure is only prominent in the local area.
• Through the linear logarithmic fitting of seismic energy and other source parameters, it is found that there are different degrees of positive logarithmic correlations between seismic source energy and other source parameters, and this positive correlation is more obvious for the hazardous high-energy seismic events.
• The correlation coefficients between seismic source energy and other source parameters show that seismic moment, stress drop, and apparent stress have a medium correlation with seismic energy, while seismic radius and apparent volume have a weak correlation with seismic energy. Based on the correlation coefficients of each source parameter, the weights of the influence of each parameter on the risk of high-energy seismic events were determined.
• An evaluation method for high-energy seismic events called the comprehensive energy index method is proposed based on the linear logarithmic relationship between seismic energy and other source parameters and the influence weights of dynamic risk. It is found that the comprehensive energy index identification method is more accurate by comparing it to the high-energy seismic events based on the energy criterion. The advantages, limitations and improvements of this method are also systematically synthesized.