“Four-in-One” Coal Mine Safety Management Method for Coal Mines Based on Time and Space Characteristics of Potential Safety Hazards

Abstract

Safety hazards in coal mines are characterized by significant spatiotemporal heterogeneity, dynamic evolution, and multi-actor coupling. Traditional safety management models, which center on periodic inspections and accident rectification, struggle to achieve proactive risk identification and full-process control. To address this issue, this paper proposes a multi-scale, collaborative “four-in-one” safety management framework oriented toward the hazard lifecycle, based on the spatiotemporal evolution patterns of safety hazards. This framework integrates systems safety theory with the safety philosophy of socio-technical systems, viewing safety hazards as an evolutionary process shaped by the combined effects of spatial exposure, human behavior, organizational management, and dynamic states. It establishes a comprehensive safety governance system comprising precise risk identification, active personnel participation, closed-loop accountability governance, and intelligent dynamic feedback. By integrating the “Area–Point–Number” risk classification method; the “Two-way Risk Purchasing” incentive mechanism; the “Six-level, Six-step, and Three-chain” closed-loop management model; and the Hazard Alert System, the framework achieves the coordinated optimization of risk identification, hazard management, and information feedback. Application validation based on safety hazard data from a coal mine between 2017 and 2020 demonstrates that this method enhances the ability to identify potential risks and effectively reduces major hazard types, such as management deficiencies, unsafe behaviors, and unsafe conditions. The research findings indicate that this framework overcomes the limitations of traditional safety management—such as a single-entity approach, static inspections, and passive responses—and facilitates a shift in coal mine safety governance from hazard control to risk prevention and from manual, experience-based management to intelligent, collaborative decision-making, thereby providing a new theoretical approach for enhancing the safety resilience of complex coal mine production systems.

1. Introduction

As China’s most important primary energy source for a long time, coal plays an irreplaceable role in ensuring energy security, supporting economic and social development, and driving the industrialization process. However, due to the complex operating environment, variable production processes, diverse types of disasters, and the high degree of coupling among personnel, equipment, the environment, and management systems, coal mining remains one of the industries with the highest safety risks in the industrial sector [1,2]. As the development of smart mines continues to advance, coal mine safety management is evolving from traditional manual inspection models toward a model characterized by “intelligent sensing, digital analysis, and collaborative decision-making.” However, due to the complex nature of the mining environment and the high degree of randomness in human behavior, relying solely on equipment monitoring data makes it difficult to fully identify management deficiencies and human-induced risks. Therefore, how to achieve the precise identification, dynamic control, and root cause remediation of safety hazards has become a key scientific issue in enhancing the inherent safety of coal mines [3].

Numerous coal mine accident cases and related studies indicate that accidents are not the result of a single factor but rather the combined effect of long-term risk accumulation, uncontrolled hidden hazards, and management failures. Among these, unsafe human behavior, unsafe equipment conditions, and management deficiencies are major contributing factors to accidents [4,5]. Modern accident causation theories further point out that while the apparent direct causes of accidents often stem from deviant human behavior and equipment malfunctions, the underlying causes are closely related to imperfect safety management systems and inadequate risk control mechanisms [6]. Therefore, the core objective of coal mine safety management should gradually shift from traditional accident control to the comprehensive management of hidden hazards throughout the entire process, achieving the proactive prevention of safety risks through the strengthened coordinated control of multiple factors, including personnel, equipment, the environment, and management. In recent years, scholars both domestically and internationally have conducted extensive research on the optimization of coal mine safety management systems, the control of safety behavior, and the management of hazard sources. Wang et al. [7] utilized methods such as Principal Component Analysis, Binary Logistic Regression, and Poisson Regression to analyze the key factors influencing unsafe behavior among coal mine workers, revealing the significant impact of human factors on accident occurrence. Liu et al. [8] explored optimization strategies for coal mine safety knowledge management from the perspective of miners’ utility, proposing that emphasis should be placed on strengthening “safety knowledge regarding rules and regulations,” placing greater importance on “job satisfaction,” and paying special attention to miners’ welfare and benefits. Regarding hazard source control, relevant studies have proposed methods for the classification, identification, graded control, and dynamic management of hazard sources, aiming to prevent accidents by reducing the probability of hazard sources getting out of control [9]. At the same time, hazards arising from the stockpiling of solid mining waste—such as landslides, debris flows, spontaneous combustion, and heavy metal leaching pollution—have been incorporated into the identification and routine management of major hazard sources in coal mines and constitute an indispensable component of coal mine hazard source control [10,11]. Bird et al. [12], the founders of modern accident causal chain theory, proposed that the direct causes of accidents are unsafe human behavior and unsafe conditions, while the root cause lies in management deficiencies. The aforementioned studies have provided an important foundation for the development of coal mine safety management theory; however, existing research still primarily focuses on the analysis of individual influencing factors or the construction of static management models. There is insufficient attention paid to the evolutionary patterns of safety hazards across different time periods and spatial areas, and there is a lack of research on methods for precise, multi-stakeholder collaborative control based on the spatiotemporal distribution characteristics of these hazards. At the same time, as the philosophy of coal mine safety management shifts from “accident prevention” to “risk pre-control,” safety hazards—as a critical intermediary linking risk and accidents—require urgent and in-depth research into their dynamic patterns and governance models. Safety hazards exhibit distinct spatiotemporal clustering characteristics, with significant differences in hazard types and occurrence frequencies across different production stages, work areas, and management processes. Therefore, identifying key control targets based on the spatiotemporal characteristics of hazards and establishing a comprehensive safety management system that covers personnel behavior, equipment status, management mechanisms, and on-site supervision are of great significance for enhancing the precision and effectiveness of coal mine safety management.

Based on this, this paper takes the SS coal mine as its research subject. By analyzing the spatiotemporal distribution characteristics of safety hazard data in recent years, it proposes a “four-in-one” safety management approach. This approach establishes a multidimensional safety control system encompassing personnel behavior control, equipment condition monitoring, management deficiency remediation, and on-site collaborative supervision and verifies its effectiveness through on-site application. This study aims to explore a coal mine safety management model driven by hazard characteristics, featuring full staff participation and dynamic collaboration, thereby providing a theoretical basis and practical reference for enhancing coal mine safety risk prevention and control capabilities and promoting innovation in mine safety management systems.

2. Correspondence Analysis

2.1. Potential Safety Hazards of SS Coal Mine

According to the statistics of the SS coal mine’s potential safety hazard database from 2015 to 2019, human unsafe behavior accounted for 45.8%, the unsafe condition accounted for 31.2%, and the two types of potential safety hazards accounted for 77%. Therefore, if people and things can be managed well, coal mine accidents will be effectively controlled [13,14] (see Figure 1).

2.2. Correspondence Analysis Model of Temporal and Spatial Distribution of Coal Mine Potential Safety Hazards

Correspondence analysis is a multivariate statistical analysis method developed by Professor Bezecri based on R-type factor analysis and Q-type factor analysis. The correspondence analysis method combines R-type factor analysis and Q-type factor analysis through the transition matrix and uses the same factor axis to reflect the variable points and category points in the contingency table on the same plane [15]. Through the spatial distance between variable category points and sample points, the relationship between them can be reflected intuitively [16].

In order to gain a deeper understanding of the spatiotemporal characteristics of potential safety hazards in coal mine safety production, we conducted statistics and analysis on the types of safety hazards detected in 2018. Through data preprocessing, the two-dimensional contingency table of “Potential safety hazard—Quarter” belongs to $\mathrm{X}={\left({\mathrm{x}}_{ij}\right)}_{4\times 8}$, contingency table column vectors consist of eight types of security risks, and row vectors are four quarters. According to the data processing process of the corresponding analysis method [15], the modeling is as follows:

(1)

The original data matrix is transformed, and each element of the data set is divided by the sum of each element to obtain the probability matrix to ensure that the row and column vectors have the same non-zero eigenvalue, as follows:

$$P=\left(p_{ij}\right)={\displaystyle \frac{{\mathrm{x}}_{ij}}{\mathrm{x}}}$$

(1)

Here,

$$\mathrm{x}={\displaystyle \sum _{i=1}^4{\displaystyle \sum _{j=1}^8{\mathrm{x}}_{ij}}}$$

(2)

where P is the joint probability matrix, p_ij is the joint probability of cell i, j, x is the total number of hazards, and x_ij is the number of hazards in category j during quarter i.

(2)

A transformation matrix $Z={\left(z_{ij}\right)}_{4\times 8}$ is constructed, and the formulas are as follows:

$$z_{ij}={\displaystyle \frac{p_{ij}-p_i·p_j}{\sqrt{p_i·p_j}}}$$

(3)

Here,

$$p_i={\displaystyle {\displaystyle \sum _{j=1}^8}{p}_{ij}};p_j={\displaystyle {\displaystyle \sum _{i=1}^4}{p}_{ij}}$$

(4)

where Z is the standardized residual matrix, z_ij represents the strength of the association between the quarter and the hazard type, and p_i and p_j are the marginal probabilities in the i-th row and j-th column, respectively.

(3)

R-type and Q-type factor analyses are performed. The eigenvalues and eigenvectors of the covariance matrix ZTZ are calculated. The first y feature values and feature vectors are selected based on the cumulative percentage of feature values. Thus, the y-dimensional factor loading matrix U is obtained:

$$\mathrm{U}=\left|\begin{array}{lllll}{\mathrm{a}}_{11}\sqrt{{\gamma }_1}\hfill & {\mathrm{a}}_{12}\sqrt{{\gamma }_2}\hfill & {\mathrm{a}}_{13}\sqrt{{\gamma }_3}\hfill & {\mathrm{a}}_{14}\sqrt{{\gamma }_4}\hfill & \dots {\mathrm{a}}_{1\mathrm{y}}\sqrt{{\gamma }_{\mathrm{y}}}\hfill \\ {\mathrm{a}}_{21}\sqrt{{\gamma }_1}\hfill & {\mathrm{a}}_{22}\sqrt{{\gamma }_2}\hfill & {\mathrm{a}}_{23}\sqrt{{\gamma }_3}\hfill & {\mathrm{a}}_{24}\sqrt{{\gamma }_4}\hfill & \dots {\mathrm{a}}_{2\mathrm{y}}\sqrt{{\gamma }_{\mathrm{y}}}\hfill \\ {\mathrm{a}}_{31}\sqrt{{\gamma }_1}\hfill & {\mathrm{a}}_{32}\sqrt{{\gamma }_2}\hfill & {\mathrm{a}}_{33}\sqrt{{\gamma }_3}\hfill & {\mathrm{a}}_{34}\sqrt{{\gamma }_4}\hfill & \dots {\mathrm{a}}_{3\mathrm{y}}\sqrt{{\gamma }_{\mathrm{y}}}\hfill \\ {\mathrm{a}}_{41}\sqrt{{\gamma }_1}\hfill & {\mathrm{a}}_{42}\sqrt{{\gamma }_2}\hfill & {\mathrm{a}}_{43}\sqrt{{\gamma }_3}\hfill & {\mathrm{a}}_{44}\sqrt{{\gamma }_4}\hfill & \dots {\mathrm{a}}_{4\mathrm{y}}\sqrt{{\gamma }_{\mathrm{y}}}\hfill \end{array}\right|$$

(5)

Similarly, the Q-type factor analysis load matrix V is obtained:

$$\mathrm{V}=\left|\begin{array}{lllll}{\mathrm{b}}_{11}\sqrt{{\gamma }_1}\hfill & {\mathrm{b}}_{12}\sqrt{{\gamma }_2}\hfill & {\mathrm{b}}_{13}\sqrt{{\gamma }_3}\hfill & {\mathrm{b}}_{14}\sqrt{{\gamma }_4}\hfill & \dots {\mathrm{b}}_{1\mathrm{y}}\sqrt{{\gamma }_{\mathrm{y}}}\hfill \\ {\mathrm{b}}_{21}\sqrt{{\gamma }_1}\hfill & {\mathrm{b}}_{22}\sqrt{{\gamma }_2}\hfill & {\mathrm{b}}_{23}\sqrt{{\gamma }_3}\hfill & {\mathrm{b}}_{24}\sqrt{{\gamma }_4}\hfill & \dots {\mathrm{b}}_{2\mathrm{y}}\sqrt{{\gamma }_{\mathrm{y}}}\hfill \\ {\mathrm{b}}_{31}\sqrt{{\gamma }_1}\hfill & {\mathrm{b}}_{32}\sqrt{{\gamma }_2}\hfill & {\mathrm{b}}_{33}\sqrt{{\gamma }_3}\hfill & {\mathrm{b}}_{34}\sqrt{{\gamma }_4}\hfill & \dots {\mathrm{b}}_{3\mathrm{y}}\sqrt{{\gamma }_{\mathrm{y}}}\hfill \\ {\mathrm{b}}_{41}\sqrt{{\gamma }_1}\hfill & {\mathrm{b}}_{42}\sqrt{{\gamma }_2}\hfill & {\mathrm{b}}_{43}\sqrt{{\gamma }_3}\hfill & {\mathrm{b}}_{44}\sqrt{{\gamma }_4}\hfill & \dots {\mathrm{b}}_{4\mathrm{y}}\sqrt{{\gamma }_{\mathrm{y}}}\hfill \end{array}\right|$$

(6)

where U and V are the R-type and Q-type factor load matrices, respectively; a_ij and b_ij are the R-type and Q-type eigenvector coefficients, respectively; and γ_i is the i-th eigenvalue.

(4)

An empirical study is conducted on the temporal and spatial distribution of potential safety hazards in the SS coal mine. During the research process, the 1142 potential safety hazards in the investigation records of the SS coal mine were taken as a sample set for this study, as shown in

Table 1. SS coal mine “Potential safety hazards—Quarter” consolidated table.

Quarter	Coal Mining	Driving	Electromechanical	Transport	Ventilation	Monitor	Geodesy	Other
First—1	10	18	48	53	55	2	2	4
Second—2	12	38	67	43	45	1	8	5
Third—3	20	65	131	95	25	5	2	11
Fourth—4	33	55	116	132	29	2	3	7

and

Table 2. SS coal mine “Potential safety hazards—Area” consolidated table.

Area	Ventilation and Prevention	Coverboard	Electromechanical	Transport	Fire Safety	Monitor	Water Hazard	Other
Coal face—1	16	18	145	35	25	3	0	9
Excavation roadway—2	28	121	121	98	81	5	8	6
Auxiliary roadway—3	21	37	71	122	43	2	5	10
Other—4	10	0	25	68	5	0	2	2

.

(5)

Model analysis

The problem is solved through the corresponding analysis function of IBM SPSS20.0 [17]. In the solution process, the chi-square distance measurement method is used to configure the model.

The results of the analysis mapping potential safety hazards to specific areas show that the chi-square test for the two-dimensional contingency table of “Safety hazard categories—Quarter” reached a significant level (Sig. < 0.001), indicating that the distribution of different types of safety hazards across quarters is not random but exhibits distinct temporal correlations (

Table 3. Calculation of correspondence analysis dimension of potential safety hazard category and quarter in SS coal mine.

Dimension	Sig.	Inertia Ratio
		Explanation	Accumulation
1	-	0.718	0.718
2	-	0.201	0.918
3	-	0.082	1.000
Total	<0.001	1.000	1.000

). As shown in Table 3, in the column analysis of dimensions, the interpretation capabilities of the first and second dimensions are 0.718 and 0.201 respectively, and the cumulative interpretation degree of the two dimensions is 91.8%, and the interpretation effect is good. Overall, the four quarters are distributed in four quadrants, indicating that there are obvious differences in the types of potential safety hazards in different quarters. Specifically, the potential safety hazards in the fourth quarter are mainly Transport and VAP (Ventilation and Prevention) [18]. The potential safety hazards in the third quarter are Electromechanical, Monitor, Coverboard and Other. The potential safety hazards in the second quarter are Water Hazards, mainly in the first quarter, and Fire Safety (see Figure 2).

As shown in

Table 4. Calculation of correspondence analysis dimension of potential safety hazard category and area in SS coal mine.

Dimension	Sig.	Inertia Ratio
		Explanation	Accumulation
1	-	0.540	0.540
2	-	0.446	0.985
3	-	0.015	1
Total	<0.001	1	1

, similarly to the “Safety hazard categories—Quarter” analysis, the chi-square test for the “Safety hazard categories—Area” two-dimensional contingency table reached statistical significance (p < 0.001), also indicating a clear spatial association between the two. From the dimension interpretation column analysis, the explanatory power of the first dimension and the second dimension is 0.540 and 0.446 respectively (Table 4). The cumulative interpretation degree of the two dimensions is 98.5%, and the interpretation effect is good. The four regions are distributed in four quadrants, indicating that there are obvious differences in the types of potential safety hazards in different regions. Specifically, the potential safety hazards in the coal mining face of the SS coal mine are mainly Electromechanical and Other. The potential safety hazards in excavation roadways are Fire Safety, Coverboard, and Monitor. The potential safety hazards in auxiliary roadways are Water Hazards. The potential safety hazards in other areas is Transport. The distribution characteristics of VAP (Ventilation and Prevention) are not obvious (Figure 3).

Through the analysis above, it can be seen that the potential safety hazards detected by coal mines have significant time and space characteristics [19]. By taking targeted measures to conduct key investigations on people and objects at a specific time and area, the probability of accidents due to potential safety hazards will be minimized. Accordingly, we proposed a four-in-one security management and control method based on people and things.

3. A “Four-in-One” Collaborative Safety Management Framework Based on the Spatiotemporal Evolution Characteristics of Hazards

Traditional coal mine safety management primarily relies on passive control measures such as periodic inspections, special rectification campaigns, and accident accountability. At its core, this is a reactive model based on accident outcomes, making it difficult to adapt to the complex characteristics of coal mine production—including the dynamic evolution of risks, the spatial migration of hidden hazards, and the coupled interactions among multiple stakeholders. According to systems safety theory, coal mine accidents are not caused by a single factor but result from the simultaneous failure of multiple safety barriers, including human behavior, equipment condition, environmental conditions, and organizational management. Therefore, the key to safety management lies in establishing a comprehensive governance system covering risk perception, behavioral control, organizational coordination, and dynamic feedback. Based on this, this paper proposes a “four-in-one” collaborative safety management framework oriented toward the lifecycle of safety hazards. Based on the spatiotemporal evolution patterns of safety hazards, this framework treats hazards as complex evolving entities driven by spatial exposure characteristics, human behavioral factors, organizational accountability mechanisms, and dynamic state changes. By integrating precise risk identification, active employee participation, closed-loop accountability control, and digital risk feedback, the framework facilitates a shift from “accident control” to “risk prevention and control” and from “manual supervision” to “collaborative governance.”

The “Four-in-One” framework includes: (1) an “Area–Point–Number” (APN) mechanism for precise risk identification based on the spatial distribution and occurrence frequency of hazards; (2) a “Two-way Risk Purchasing” (TWR) mechanism for active participation based on incentives for employee safety behavior; (3) a “Six-level, Six-step, and Three-chain” (SST) full-process closed-loop governance mechanism based on the integration of the chain of responsibility; and (4) a Hazard Alert System (HAS) for graded hazard early warning based on the integration of hazard data. These four modules correspond to the risk perception, behavioral regulation, organizational control, and information feedback functions within the safety system, respectively, forming a multi-layered safety barrier.

3.1. APN-Based Mechanism for Precise Risk Identification Based on Spatiotemporal Heterogeneity of Hidden Hazards

Coal mine production systems exhibit significant spatial heterogeneity; risk exposure levels vary markedly across different production areas due to differences in mining processes, equipment condition, the intensity of personnel activity, and environmental conditions. However, traditional standardized inspection models typically employ fixed cycles and uniform standards, making it difficult to precisely align safety supervision resources with actual risk levels. To address these issues, this paper introduces the APN risk identification mechanism, which combines the spatial distribution patterns of hidden hazards with their occurrence frequency characteristics to construct a three-dimensional risk identification model comprising “Risk areas–Key locations–Dynamic frequency.” In this model, “areas” describe the overall risk level of production units; “locations” identify key risk sources affecting production safety; and “frequency” reflects the probability of hazard exposure and dynamic changes in risk, thereby shifting safety inspections from an experience-driven to a risk-driven approach, as shown in

Table 5. Frequency of inspections corresponding to different inspection points.

Node	Description	Inspection Times	Remark
AI	Class A area level I checkpoint	1–3	per day
AII	Class A area level II checkpoint	1–3	per day
AIII	Class A area level III checkpoint	1–3	per day
BI	Class B area level I checkpoint	1–3	per week
BII	Class B area level II checkpoint	1–3	per week
BIII	Class B area level III checkpoint	1–3	per week
CI	-	-	-
CII	Class C area level II checkpoint	2–4	per half month
CIII	Class C area level III checkpoint	2–4	per half month
DI	-	-	-
DII	-	-	-
DIII	Class D area level III checkpoint	3–5	per month

and

Table 6. Inspection frequencies for each professional inspection team every quarter.

Node	Description	Number of Inspection Days per Quarter
		First	Second	Third	Fourth
AJ	Follow-up Safety Inspector	7 × 4 × 3	7 × 4 × 3	7 × 4 × 3	7 × 4 × 3
CM	Coal Mining Inspection Team	2 × 4 × 3	2 × 4 × 3	1 × 4 × 3	3 × 4 × 3
JJ	Tunneling Inspection Team	1 × 4 × 3	2 × 4 × 3	3 × 4 × 3	1 × 4 × 3
JD	Mechanical and Electrical Inspection Team	1 × 4 × 3	2 × 4 × 3	3 × 4 × 3	2 × 4 × 3
YS	Transportation Inspection Team	2 × 4 × 3	2 × 4 × 3	1 × 4 × 3	3 × 4 × 3
TF	Ventilation Inspection Team	1 × 4 × 3	1 × 4 × 3	3 × 4 × 3	1 × 4 × 3
FC	Dust Inspection Team	1 × 4 × 3	1 × 4 × 3	3 × 4 × 3	1 × 4 × 3
JC	Monitoring Inspection Team	1 × 4 × 3	2 × 4 × 3	3 × 4 × 3	1 × 4 × 3
DC	Geophysical Inspection Team	2 × 4 × 3	3 × 4 × 3	2 × 4 × 3	1 × 4 × 3
FP	Firing Inspection Team	1 × 4 × 3	1 × 4 × 3	1 × 4 × 3	1 × 4 × 3
ZL	Quality Supervision Team	1 × 4 × 3	1 × 4 × 3	1 × 4 × 3	1 × 4 × 3
GG	Public Supervision Team	1 × 4 × 3	1 × 4 × 3	1 × 4 × 3	1 × 4 × 3

.

Based on differences in risk levels, production areas are divided into four management units: A, B, C, and D. Category A areas correspond to high-risk concentration zones and are subject to high-frequency dynamic supervision; Category B areas receive focused attention; and C and D areas are managed using periodic and routine strategies, respectively. This classification mechanism avoids the problem of evenly distributing supervisory resources found in traditional safety management, enabling limited management resources to be prioritized for high-risk areas [20]. Furthermore, by integrating equipment hazard levels, job-specific risk characteristics, and historical hazard data, risk objects within each area are classified into Level I, II, and III critical points [21]. Level I points primarily include mining faces, major transportation systems, and critical safety facilities—objects that directly impact production safety; Level II points mainly involve auxiliary production equipment and support facilities; and Level III points primarily include general infrastructure and management objects. By establishing a spatial risk stratification mechanism, the approach facilitated a shift from the broad-brush regional management of safety hazards to precise control at key nodes [22,23].

3.2. The TWR Active Participation Governance Mechanism Based on Behavioral Incentives

Unsafe human behavior is a major contributing factor to coal mine accidents; however, traditional safety management models typically treat employees as subjects of supervision, resulting in a lack of motivation among frontline personnel to proactively identify risks and report potential hazards. To address this issue, this paper proposes the TWR active participation mechanism (Figure 4), which establishes a mechanism for evaluating the value of potential hazards and providing incentive-based feedback to transform employees’ roles from “safety enforcers” to “risk management participants” [24]. The core of the TWR mechanism lies in linking the value of hazard information to employees’ behavioral contributions. By combining financial incentives, accountability, and information feedback, it enhances on-site personnel’s initiative in risk identification. On the one hand, frontline employees, who are constantly immersed in the production environment, possess a stronger on-site awareness of equipment anomalies, deviations in work practices, and environmental changes, enabling them to detect latent risks that managers may struggle to identify in a timely manner; on the other hand, by quantitatively evaluating the severity of hazards, the difficulty of remediation, and the potential impact of accidents, the mechanism quantifies the value of employees’ safety contributions. This mechanism breaks away from the traditional one-way management model—where “identifying hazards is the responsibility of managers”—and establishes a collaborative mechanism of “employees identify risks—the organization evaluates risks—the system addresses risks,” thereby strengthening human factor control capabilities in coal mine safety management [25,26].

3.3. SST Full-Process Closed-Loop Governance Mechanism Based on Chain-of-Responsibility Coordination

During the process of addressing potential hazards, blurred lines of responsibility, disjointed rectification processes, and a lack of continuous monitoring of remediation outcomes are key factors leading to the recurrence of such hazards. To address these issues, this paper proposes an SST full-process closed-loop governance mechanism that achieves comprehensive control over the entire lifecycle of potential hazards through the coordination of the responsibility system, governance process, and acceptance mechanism.

As shown in Figure 5, the SST mechanism comprises a “six-tier accountability system,” a “six-step remediation process,” and a “three-chain coordination mechanism” [27,28,29]. The “six tiers” cover accountability entities such as individual positions, work crews, teams, specialized departments, coal mines, and the corporate group, ensuring that management responsibilities are delegated step by step; the “six steps” include hazard identification, risk assessment, remediation implementation, effectiveness evaluation, re-inspection and confirmation, and closed-loop closure, enabling continuous control over the remediation process [30,31,32]; The “three-chain” mechanism consists of the hazard discovery chain, the responsibility implementation chain, and the rectification acceptance chain, ensuring the effective coordination of information, responsibilities, and execution processes. Compared to traditional hazard rectification models, the SST mechanism breaks away from a phased management approach that treats the completion of rectification as the endpoint. Instead, it integrates the hazard management process into a dynamic control system, enabling traceable management throughout the entire process—from discovery and remediation to elimination—and enhancing the practicality and operability of the safety management system.

3.4. HAS Dynamic Risk Feedback Mechanism Based on Hazard Data Integration

As the development of smart mines continues to advance, traditional manual recording and static statistical methods are no longer sufficient to meet the need for rapid risk response in complex mining environments. Therefore, this paper proposes an HAS dynamic risk feedback mechanism that enables the digital management of safety information through hazard data coding, status tracking, and tiered early warning. With the hazard database at its core, the HAS structurally manages information such as hazard locations, risk levels, responsible parties, rectification deadlines, and remediation status and implements dynamic early warnings based on risk levels and time constraint rules. When a hazard is not remediated within the specified timeframe, the system automatically triggers a warning of the corresponding level and prompts the responsible party to respond promptly [33]. The essence of the HAS lies in establishing an information feedback mechanism focused on the hazard lifecycle, thereby transitioning safety management from manual intervention to a data-driven approach. This mechanism not only improves the timeliness of hazard remediation but also provides a data foundation for subsequent risk trend analysis and the optimization of safety decision-making (see Figure 6).

4. Discussion

Safety hazards in coal mines exhibit distinct characteristics of dynamic evolution; their emergence and development are influenced by a combination of factors, including human behavior, equipment condition, environmental conditions, and organizational management. Traditional safety management models typically focus on corrective actions and control measures after hazards are identified, making it difficult to achieve dynamic risk management throughout the entire process. To address these issues, this paper constructs a “four-in-one” safety management system comprising APN, TWR, SST, and HAS, based on the spatiotemporal distribution characteristics of safety hazards, and validates its effectiveness through practical application.

4.1. Mechanism of Synergy Among APN, SST, TWR, and HAS

The “four-in-one” safety management system is not a simple combination of traditional safety management measures but rather a multi-scale, multi-stakeholder collaborative governance model built around the lifecycle of safety hazards. Specifically, APN is responsible for spatial risk identification and precise resource allocation; TWR enhances personnel’s proactive awareness capabilities; SST implements the closed-loop control of organizational accountability; and the HAS provides digital feedback and dynamic response support. Together, these four components form a full-process control chain encompassing “risk identification—proactive detection—accountability-based governance—dynamic feedback.”

(1)

APN enables the precise spatial identification of safety risks and serves as the foundational layer of the “Four-in-One” system.

During coal mine production, there are significant differences in risk exposure levels across different areas, workstations, and equipment units; traditional, uniform inspection models struggle to optimize the allocation of safety resources. Based on the spatial clustering characteristics and temporal frequency patterns of hazard occurrence, the APN method correlates production areas, key risk points, and inspection frequencies, thereby shifting the allocation of safety supervision resources from an experience-based approach to a risk-driven one. This method not only enhances the ability to identify risks in key areas and critical processes but also provides clear targets and priorities for subsequent hazard remediation.

(2)

SST facilitates organizational coordination in the hazard remediation process and serves as the execution layer of the “Four-in-One” system.

The effectiveness of hazard remediation depends not only on detection capabilities but also on the implementation of responsibilities and process control. By establishing a multi-level accountability system encompassing workstations, work crews, teams, specialized departments, coal mines, and the corporate group, the SST method systematically links processes such as hazard screening, risk identification, rectification implementation, and follow-up verification and closure, thereby enabling traceable management throughout the entire hazard remediation process. Compared to traditional single-step rectification models, SST strengthens internal information flow and accountability within the organization, effectively reducing the recurrence of hazards caused by ambiguous responsibilities and management failures.

(3)

TWR reinforces employee proactive participation and serves as a key driver for the transformation of safety management from passive compliance to active governance. Coal mine production sites are highly dynamic, and relying solely on management inspections makes it difficult to fully grasp real-time risk conditions. Frontline employees, who are constantly immersed in the production environment, possess a natural advantage in identifying latent risks. By establishing a hazard value assessment and incentive-based feedback mechanism, TWR transforms employee safety behaviors into corporate risk control resources, thereby increasing the enthusiasm of frontline personnel to participate in hazard identification and risk management. This mechanism embodies the crucial role of “human adaptive capacity” in safety management, as outlined in socio-technical systems theory.

(4)

The HAS implements a closed-loop information system for safety management and serves as the dynamic feedback layer of the “Four-in-One” system. In complex coal mine production systems, addressing potential hazards requires timely information transmission and status feedback. Through hazard coding, accountability assignment, and a tiered early warning mechanism, HAS enables dynamic tracking of hazard statuses and timely management responses. Essentially, the system establishes an information feedback mechanism focused on the hazard lifecycle, shifting safety management from manual processes to data-driven decision-making, thereby enhancing the timeliness and reliability of risk control.

In summary, the “Four-in-One” management framework does not merely integrate existing management measures; rather, it starts from the evolution process of latent hazards to achieve synergistic coupling among different management mechanisms across both temporal and spatial scales. APN, TWR, SST, and the HAS correspond to the four key links in safety management—“risk localization, stakeholder participation, responsibility control, and dynamic feedback”—respectively, forming a multi-layered defense mechanism consistent with systems safety theory. By enhancing the perception, response, and recovery capabilities of the coal mine safety system, this framework facilitates a transition from accident prevention to risk-resilient governance. Specifically, APN addresses the spatial identification issue of “where risks are located,” enabling precise risk localization; TWR addresses the behavioral issue of “who proactively identifies risks,” improving on-site risk perception capabilities; SST addresses the organizational issue of “how to continuously manage risks,” achieving closed-loop accountability control; the HAS addresses the technical issue of “how to respond rapidly to risks,” enabling dynamic information feedback. Together, these four modules form a safety governance chain of “spatial perception—behavioral drive—organizational coordination—intelligent feedback,” driving the transformation of coal mine safety management from static inspections to dynamic prevention and control, from one-way supervision to collective governance by all personnel, and from experience-based management to data-driven decision-making. This provides a theoretically sound and practically applicable framework and pathway for the precise control of safety risks under complex coal mine production conditions.

4.2. An Analysis of the Implementation Effectiveness of the “Four-in-One” Safety Management Method

Figure 7 illustrates the evolutionary trends in various types of safety hazards from 2017 to the first half of 2020. Overall, since the implementation of the “Four-in-One” safety management approach in the second half of 2018, the number of potential safety hazards in coal mines has exhibited a dynamic pattern of “short-term increase—rapid decline—stable control.” The temporary increase in the number of hazards during the initial implementation phase did not indicate a deterioration in safety conditions. Rather, it resulted from the introduction of the TWR hazard identification mechanism, which enhanced frontline employees’ ability to identify potential risks and hidden issues, thereby fully exposing risk factors that had previously gone undetected. This phenomenon reflects a shift in the safety management system from “passive hazard exposure” to “active identification and control” and is also a significant manifestation of improved safety governance capabilities. With the continued operation of the “Four-in-One” management system, the APN method has improved the efficiency of safety inspection resource allocation by implementing precise control over high-risk areas and critical locations; the SST method has strengthened accountability and closed-loop management during the hazard remediation process; and the HAS has further accelerated the speed of hazard information feedback and corrective action responses. Consequently, exposed potential hazards can be promptly addressed, and the total number of hazards has gradually decreased and stabilized, indicating that this management system has effectively enhanced the coal mine’s ability to control safety risks.

Based on the evolution patterns of different types of hidden risks, there are certain differences in how various types of hidden risks respond to the “Four-in-One” management approach. As shown in Figure 7a, there was a short-term increase following the implementation of the new management model, after which the numbers gradually stabilized and began to decline. This was primarily due to the SST closed-loop management mechanism, which strengthened the identification of organizational responsibilities, bringing previously hidden management loopholes to light and effectively controlling them through accountability tracking and continuous improvement mechanisms. The trends in the two categories of hazards—unsafe behaviors and unsafe conditions—exhibited a high degree of consistency (Figure 7b,c). Following the implementation of the “Four-in-One” approach in 2018, the TWR mechanism encouraged employees to actively participate in hazard identification, leading to the timely detection of a large number of potential behavioral risks and equipment status anomalies, which caused a short-term increase in the reported numbers. Subsequently, as employee safety awareness improved, on-site risk control was strengthened, and the closed-loop rectification mechanism was refined, the numbers of both hazard categories gradually stabilized and declined between 2019 and 2020. This indicates that the “Four-in-One” approach not only reduced risk exposure levels but also enhanced the proactive defense capabilities of the on-site safety management system. The number of safety hazards caused by environmental factors generally showed a fluctuating downward trend (Figure 7d). Since environmental factors are significantly influenced by production conditions, geological conditions, and external disturbances—and are characterized by high uncertainty and recurrence—their management process exhibits more pronounced fluctuations compared to those related to human behavior and management factors. However, the overall downward trend still demonstrates that the “Four-in-One” approach exerts a certain degree of control over environmental risks. The number of hazards in other categories is relatively small, accounting for only 4% of the total potential safety hazards. Their trends exhibit significant fluctuations and relatively weak statistical regularity; therefore, the effectiveness of their mitigation is difficult to fully evaluate based on changes in quantity alone. Further analysis reveals that the trend in the total number of safety hazards closely aligns with that in the three main categories of hazards—management deficiencies, unsafe behaviors, and unsafe conditions (Figure 7f). This indicates that these three factors are the primary determinants influencing coal mine safety conditions and are also the key targets for the “Four-in-One” management system to take effect. Overall, by integrating precise risk identification, active employee participation, closed-loop accountability, and dynamic information feedback, this method has transformed coal mine safety management from a passive “identify problems—rectify problems” model to a dynamic governance model of “identify risks—proactively intervene—continuously optimize,” effectively enhancing the safety management system’s perception capabilities, responsiveness, and resilience.

5. Conclusions

Addressing issues in traditional coal mine safety management—such as delayed risk identification, a fragmented chain of responsibility, and insufficient information feedback—this paper proposes a multi-scale, collaborative “four-in-one” safety management framework oriented toward the lifecycle of safety hazards, based on the spatiotemporal evolution patterns of such hazards. This framework breaks away from the traditional linear “inspection–correction” governance model. It treats safety hazards as an evolutionary process shaped by the combined effects of spatial risk exposure, human behavior, organizational management, and dynamic state changes, thereby establishing a comprehensive safety governance system that integrates precise risk identification, proactive behavioral incentives, closed-loop accountability governance, and dynamic information feedback. Practical validation has demonstrated that this “four-in-one” control method achieves significant results in managing unsafe human behavior, unsafe conditions, and management deficiencies. The main conclusions are as follows:

(1)

A multidimensional collaborative safety management framework was established, integrating APN risk classification and control, the TWR active participation mechanism, the SST closed-loop governance model, and the HAS dynamic early warning system. This framework achieves coordinated control over spatial risks, human behavior, organizational responsibilities, and digital information, driving the transformation of coal mine safety management from single-factor governance to systematic governance.

(2)

A dynamic closed-loop mechanism covering the entire lifecycle of hidden hazards was established, effectively linking risk identification, hazard remediation, and feedback on outcomes. This facilitated a shift in safety management from post-incident rectification to proactive prevention and from experience-driven to data-assisted decision-making, thereby enhancing the dynamic adaptability of the safety management system.

(3)

A proactive governance mechanism targeting human factor risks was proposed. By incentivizing frontline employees to participate in risk identification and hazard detection, this mechanism strengthened employees’ sense of responsibility for safety, integrated external oversight with internal self-regulation, and effectively enhanced the ability to detect latent risks on-site.

(4)

A smart mine safety management model based on “digital technology empowerment + organizational collaborative governance” was developed. This model overcomes the limitations of traditional intelligent monitoring, which tends to focus on equipment status perception, and provides a new theoretical framework and practical pathway for the coordinated control of human factor risks, management risks, and equipment risks in the context of intelligent mine construction.