Safety Evaluation and Simulation Research of Filling Mining Mine—A Case Study of Jisuo Coal Mine

Abstract

With the demand for green mining in coal mines, filling mining is becoming more and more popular, resulting in more complex production systems and more potential safety hazards. Therefore, it is very important to evaluate the safety of filling mining mines and propose improvement measures. Aiming at the safety evaluation method of filling mining mines, this paper innovatively proposes a safety evaluation method based on entropy weight–attribute mathematical theory, which enriches the theoretical research related to the safety evaluation model of filling mining mines. Five secondary indexes and twenty-two tertiary indexes were selected. The weights were determined via the entropy weight method, and then the attribute mathematical theory was used for safety evaluation. The evaluation results show that the safety level of Jisuo Coal Mine is “relatively safe”, and the evaluation results are in good agreement with the actual situation of Jisuo Coal Mine, which verifies the applicability of the attribute mathematical theory. Finally, from the perspective of safety input, the simulation study is carried out by using system dynamics, and the dynamic change rule is analyzed. Additionally, improvement measures for filling mining mine safety are proposed so as to realize the reasonable optimization of resource allocation.

1. Introduction

The coal industry has always been one of the pillar industries in China, but it is also a high-risk industry, with the accident severity being greater than other industries. Every coal mine safety production accident will bring economic loss and spiritual damage to the people and have a great negative impact on the country’s society. With the increasing attention being paid to the production safety of the coal industry, the government and the competent departments at all levels have made great progress in the safety production of China’s coal mining enterprises by taking a series of measures. Frequent underground coal mining will lead to problems such as overburden fracture, surface subsidence, groundwater damage and environmental pollution [1]. In 2020, China formally proposed the 2030 “carbon peak” and 2060 “carbon neutrality” targets, which introduced higher requirements for low-carbon production and clean utilization of coal. The 14th Five-Year Plan also proposed to reduce energy consumption and carbon dioxide emissions per unit of GDP by 13.5% and 18%, respectively, by 2035, which means that the timely adjustment of the development direction of coal resources will become the primary focus of China’s energy low-carbon development strategy in the future. Therefore, filling mining, as a green mining technology, is one of the important technical means to reduce the disturbance of mining to the overlying strata, the discharge of waste on the surface and environmental pollution. It can effectively solve the outstanding contradiction between ground environmental protection and coal resource mining damage. It meets the requirements of the new development concept of green intelligent mining for coal resource disaster reduction and loss reduction and has been widely used around the world.

At present, there are three basic filling mining methods in mining: dry, hydraulic and cement. Its primary tasks are waste management in underground mining and surface protection. In order to compare empty models with cribs filled with gangue, Skrzypkowski K. et al. introduced comparative laboratory strength tests of three-point and four-point wooden crib models. It was found that filling the three-point and four-point cribs with gangue increases their maximal load several times compared to the empty cribs [2]. Skrzypkowski K. et al. introduced a Backfill CAD model that relates to the determination of the backfilling time and determined the time of backfilling for the prospective deposits of zinc and lead ores in the Olkusz region in Poland [3]. Skrzypkowski K. et al. presented laboratory and spatial numerical modeling of cemented paste backfill. The main objective was to determine the changes in displacements around the haulage room and transportation roadway located in the immediate vicinity of the exploitation field [4]. Smolinski A. et al. studied the effect of leaving waste bottom rocks in the mined-out space of the longwall face without their drawing to the Earth’s surface via the geomechanical state of the rocks surrounding the longwall face [5]. Malashkevych D. et al. proposed an algorithm for predicting the ash and quality of coal by using the selective mining technology of waste rock deposits in goaf, which is of great significance to the technical and economic indicators of coal mines [6]. Based on the analysis and evaluation of the geological conditions, process parameters and support methods of a longwall, Vu T.T. et al. pointed out the causes and laws of the spalling and roof caving of the longwall working face. On this basis, the countermeasures to prevent this phenomenon are put forward to ensure the safety of the longwall [7].

Safety is the first objective in coal mine production. Coal resources play an important role in the sustainable development of the national economy, and coal mine safety evaluation also occupies an extremely important position in coal mine safety production. Green mining is the inevitable trend of mining development. In recent years, filling mining has become more and more popular in line with the demand for green mining. The use of filling mining can improve the recovery rate of minerals, make full use of effective resources and effectively control surface subsidence. However, it also makes the coal mine production system add more influencing factors, such as people, money and materials, resulting in more complex production systems. There are also more safety hazards. Therefore, it is very important to evaluate the safety of filling mining mines and put forward improvement measures.

Guo et al. proposed a safety evaluation model utilizing fusion weight and a cloud model with set pair analysis and verified it with coal mine examples to determine the safety status of coal mines. [8]. Hu et al. established a comprehensive hierarchical mine safety situation evaluation system taking into account four levels, which were hidden danger, location, function and system, and proposed an AHP-entropy weight evaluation model for the coal mine function weight based on hidden danger record data. The empirical results showed that the model could assist safety managers in understanding the safety threats of coal mines [9]. Yu et al. determined reasonable coal and gas outburst indexes through a cloud model and D-S theory, avoiding the influence of accidental factors and fuzzy uncertainty on the test results [10]. Gao et al. used the combination of an analytic hierarchy process and correlation analysis to optimize the index system, and, combined with the safety production practice of Haohua Energy Hongqingliang Coal Mine, the safety risk evaluation index system in line with the mine was obtained [11]. Jiang established the SEM model to analyze the actual effect rate of each influencing factor on the safety management level and the relationship between the indicators and then evaluated the safety management level of a mine through the analytic hierarchy process and fuzzy comprehensive evaluation method, subsequently putting forward countermeasures according to the evaluation results [12].

At present, many scholars have established different models to study the safety evaluation of coal mines, but few have considered the impact of filling mining factors on coal mine safety. The existing coal mine safety evaluation models and methods have the disadvantages of being too subjective and the indicators not being fully considered, affecting the judgment of the evaluation results. Li established a quantitative model of the geological strength index based on attribute mathematical theory, which provides a new method for the quantification of a GSI system [13]. Based on the theory of attribute mathematics, Zhang et al. established a risk assessment system for tunnel gas outbursts. The evaluation results are in good agreement with the actual situation, which verifies the applicability of the attribute evaluation system [14]. Li et al. constructed a high-rise building fire risk assessment model based on entropy weight–attribute mathematics, providing a new method for high-rise building fire risk assessment [15]. Cai et al. used gas, geology, coal and other factors as the evaluation index for the coal and gas outburst risk level. Additionally, based on the index critical value and the grading method, combined with the entropy weight method and the attribute mathematical theory, a comprehensive evaluation model for coal and gas outburst risk level was constructed [16]. Han used the analytic hierarchy process to determine the weight of the index and combined this with the attribute mathematical theory to form the final comprehensive evaluation model of the stability of the ventilation system, thus comprehensively evaluating the stability of the ventilation system in Changcun Coal Mine [17].

The index weight obtained using the entropy weight method is objective, and the at-tribute mathematical theory can efficiently solve the comprehensive evaluation problem of fuzzy attributes. The combination of the two makes the evaluation results even more reliable. Compared with subjective methods such as the analytic hierarchy process, the entropy weight method has certain accuracy. The weight obtained using the entropy weight method can reduce the shortcomings of subjectivity in previous evaluation methods and can also improve the calculation accuracy of attribute mathematics. The advantage of using attribute mathematical theory to evaluate is that it will not lose useful information when making judgments, and the classification of orderly segmented problems is more accurate and detailed. In the comprehensive evaluation model of attribute mathematical theory, the use of confidence recognition criteria makes the evaluation results more accurate, and there will be no unclear classification or unreasonable classification such as with the fuzzy comprehensive evaluation model. At the same time, the comprehensive evaluation model of attribute mathematical theory will not be debugged and trained repeatedly, as with the neural network model. Therefore, the use of attribute mathematical theory to evaluate can not only overcome the shortcomings of a fuzzy comprehensive evaluation but also overcome the shortcomings of the neural network model, in addition to the model being easy to implement. The entropy weight–attribute mathematical theory model has been used in high-rise building fire risk assessment, coal and gas outburst risk level comprehensive evaluation models and so on. However, it has not been applied to the safety evaluation of filling mining mines, and most of the current coal mine evaluation methods adopt subjective evaluation methods. Therefore, this study can provide an objective and accurate new method for the safety evaluation of filling mining mines from a quantitative point of view and enrich the theoretical research related to the safety evaluation model of filling mining mines.

Aiming at the risk perception ability of construction workers, Ma et al. constructed an SD model of the interaction between social factors and personal risk perception, simulated the evolution process of risk perception in different group environments and explained the reasons for the high accident rate [18]. Ma et al. analyzed the effectiveness of safety inspection between coal mining enterprises and government supervision departments by using system dynamics and dynamic game theory, putting forward suggestions to improve the performance of safety supervision [19]. Based on the coal mine construction project, Ma took the safety risk as the starting point and constructed the safety risk factor system. Additionally, the mechanism of action of factors was analyzed by using an association rule, decision-making and evaluation laboratory (DEMATEL) method and interpretative structural model (ISM). Additionally, the system dynamics method (SD) was used to construct the system dynamics model of the construction safety risk of a coal mine construction project, and the case analysis was carried out in combination with the actual project [20]. Guided by the methods and ideas of lean management, combined with the problem-solving ideas of system dynamics, Yang applied system dynamics to the lean safety management of coal mines, studied the basic theory and dynamic system of lean safety management based on system dynamics, constructed the system dynamic model of coal mine lean safety management and simulated the model [21].

As a theoretical method, the application of system dynamics includes urban economic development, engineering technology, energy planning and many other fields [22,23,24]. Using the system dynamics method to carry out the safety simulation of filling mining mines can systematically and comprehensively analyze the relationship between various factors, study the safety input distribution of each subsystem and put forward corresponding improvement measures and suggestions, so as to reasonably optimize the allocation of resources and continuously improve the safety of filling mining mines.

According to the factors affecting coal mine safety, this paper establishes the safety evaluation index system of filling mining mines from five aspects: personnel quality, filling mining equipment and facilities, safety management of filling mining mine, filling mining environment and filling mining technical ability. Then, based on the entropy weight–attribute mathematical theory, a comprehensive evaluation model of mine safety in filling mining is established, and then it is concluded that the safety level for a cable coal mine is “relative safety”. Finally, from the perspective of safety input, the system dynamics method is used for simulation research, and improvement measures and suggestions are put forward, with the rationality being verified via case analysis. Taking Jisuo Coal Mine as an example, the safety evaluation and simulation of filling mining mines can be used to know the safety level of a coal mine and put forward opinions and suggestions to avoid an unreasonable utilization of resources. It is beneficial to enrich the theoretical research related to the safety evaluation model of filling mining mines and provide a comprehensive and accurate new method for the safety evaluation of filling mining mines, which has guiding significance for the long-term safety development of filling mining mines.

2. Methods

According to the complexity of the factors affecting the safety of filling mining mines, the most suitable mathematical model for the safety evaluation of filling mining mines that needs to fully consider the influence of various factors and the results of single factor evaluation is achieved. Based on the opinions of experts in the field, this paper considers the fuzzy comprehensive evaluation model, neural network model and attribute mathematical theory as methods of evaluation, and considers the analytic hierarchy process, principal component analysis and entropy weight method to calculate the weight. However, the fuzzy comprehensive evaluation model will have unclear classification or unreasonable classification. The neural network model needs to be debugged and trained repeatedly. The use of attribute mathematical theory for evaluation can not only overcome the shortcomings of a fuzzy comprehensive evaluation but also overcome the shortcomings of the neural network model. Additionally, the model is easy to implement. Regarding the determination of weights, the analytic hierarchy process has the disadvantage of being too subjective. The data collected in this paper are not suitable for principal component analysis. The entropy weight method can not only reduce the subjectivity of the previous evaluation methods but is also suitable for the weight calculation of the data collected in this paper. Therefore, this paper chooses the comprehensive evaluation model of entropy weight–attribute mathematical theory to evaluate the safety level of filling mining mines.

2.1. Entropy Weight Method

The entropy weight method is a mathematical method used to calculate a comprehensive index based on the amount of information provided by various factors. Its main principle is to calculate the entropy and weight of the index by obtaining the score of the index and standardizing the score data [25]. The process of the entropy weight method is as follows.

The initial evaluation matrix was normalized to obtain matrix P, as shown in Equation (1).

$$P_{ij}=\frac{X_{ij}-\min \left(X_i\right)}{\max \left(X_i\right)-\min \left(X_i\right)},$$

(1)

Information entropy certainty Equation (2).

$$E_j=\frac{-{\displaystyle {\sum }_i^m\frac{P_{ij}}{{\displaystyle {\sum }_{i=1}^m{P}_{ij}}}\ln \frac{P_{ij}}{{\displaystyle {\sum }_{i=1}^m{P}_{ij}}}}}{\ln m},$$

(2)

For the determination of index weight, see Equation (3).

$$G_j=1-E_j,$$

(3)

Therefore, the weight of the JTH index can be expressed as Equation (4).

$$H_j=\frac{G_j}{{\displaystyle {\sum }_{j=1}^n{G}_j}},$$

(4)

2.2. Attribute Mathematical Theory

2.2.1. Single Index Attribute Measure Analysis

Consider a single index $I_j$, the measured value of the $jth$ index $I_j$ of sample $x_i$ is $x_{ij}$. $x_{ij}\in C_k$ denotes that $x_{ij}$ belongs to class $k$ ($C_K$ is the class $k$), and its attribute measure is $\mu \left(x_{ij}\in C_k\right),$ $i=1,2,\cdots ,n,k=1,2,\cdots ,K$. $F$ is some property on X, and $\left\{C_1,C_2,\cdots C_K\right\}$ is an ordered partition class of attribute space $F$ and satisfy the $C_1>C_2>\cdots >C_K$.The classification of the single index is shown in

Table 1. Single index classification table.

	Grade	${\mathit{C}}_1$	${\mathit{C}}_2$	…	${\mathit{C}}_{\mathit{K}}$
Index
$I_1$		$a_{10}-a_{11}$	$a_{11}-a_{12}$	…	$a_{1K-1}-a_{1K}$
$I_2$		$a_{20}-a_{21}$	$a_{21}-a_{22}$	…	$a_{2K-1}-a_{2K}$
…		…	…	…	…
$I_n$		$a_{n0}-a_{n1}$	$a_{n1}-a_{n2}$	…	$a_{nK-1}-a_{nK}$

Where $a_{ji}$ satisfies $a_{j0}<a_{j1}<\cdots <a_{jK}$ or $a_{j0}>a_{j1}>\cdots >a_{jK}$.

.

The $jth$ index value of sample $x_i$ is $t$, that is, $x_{ij}$. The single index attribute measure function ${\mu }_{x_{i}jk}\left(t\right)$, that is, ${\mu }_{x_{i}jk}\left(t\right)=\mu \left(x_{ij}\in C_k\right)$, should be determined using Table 1.

$$b_{jk}=\frac{a_{jk-1}+a_{jk}}{2}, k=1,2,\cdots , K,$$

(5)

$$d_{jk}=\min \left\{\left|b_{jk}-a_{jk}\right|,\left|b_{jk+1}-a_{jk}\right|\right\}, k=1,2,\cdots , K-1,$$

(6)

When $a_{j0}<a_{j1}<\cdots a_{jK}$, the single index attribute measure function ${\mu }_{x_{i}jk}\left(t\right)$ is determined as follows:

$${\mu }_{x_{i}j1}\left(t\right)=\{\begin{array}{ll}1\hfill & t<a_{j1}-d_{j1}\hfill \\ \frac{\left|t-a_{j1}-d_{j1}\right|}{2d_{j1}}\hfill & a_{j1}-d_{j1}\le t\le a_{j1}+d_{j1}\hfill \\ 0\hfill & a_{j1}+d_{j1}<t\hfill \end{array}$$

(7)

$${\mu }_{x_{i}jk}\left(t\right)=\{\begin{array}{ll}0\hfill & t<a_{jk-1}-d_{jk-1}\hfill \\ \frac{\left|t-a_{jk-1}+d_{jk-1}\right|}{2d_{jk-1}}\hfill & a_{jk-1}-d_{jk-1}\le t\le a_{jk-1}+d_{jk-1}\hfill \\ 1\hfill & a_{jk-1}+d_{jk-1}<t<a_{jk}-d_{jk}\hfill \\ \frac{\left|t-a_{jk}+d_{jk}\right|}{2d_{jk}}\hfill & a_{jk}-d_{jk}\le t\le a_{jk}+d_{jk}\hfill \\ 0\hfill & a_{jk}+d_{jk}<t\hfill \end{array}$$

(8)

$${\mu }_{x_{i}jK}\left(t\right)=\{\begin{array}{ll}1\hfill & a_{jK-1}+d_{jK-1}<t\hfill \\ \frac{\left|t-a_{jK-1}+d_{jK-1}\right|}{2d_{jK-1}}\hfill & a_{jK-1}-d_{jK-1}\le t\le a_{jK-1}+d_{jK-1}\hfill \\ 0\hfill & t<a_{jK-1}-d_{jK-1}\hfill \end{array}$$

(9)

When $a_{j0}>a_{j1}>\cdots a_{jK}$, the single index attribute measure function ${\mu }_{x_{i}jk}\left(t\right)$ is determined as follows:

$${\mu }_{x_{i}j1}\left(t\right)=\{\begin{array}{ll}1\hfill & a_{j1}+d_{j1}<t\hfill \\ \frac{\left|t-a_{j1}+d_{j1}\right|}{2d_{j1}}\hfill & a_{j1}-d_{j1}\le t\le a_{j1}+d_{j1}\hfill \\ 0\hfill & t<a_{j1}-d_{j1}\hfill \end{array}$$

(10)

$${\mu }_{x_{i}jk}\left(t\right)=\{\begin{array}{ll}0\hfill & a_{jk-1}+d_{jk-1}<t\hfill \\ \frac{\left|t-a_{jk-1}-d_{jk-1}\right|}{2d_{jk-1}}\hfill & a_{jk-1}-d_{jk-1}\le t\le a_{jk-1}+d_{jk-1}\hfill \\ 1\hfill & a_{jk}+d_{jk}<t<a_{jk-1}-d_{jk-1}\hfill \\ \frac{\left|t-a_{jk}+d_{jk}\right|}{2d_{jk}}\hfill & a_{jk}-d_{jk}\le t\le a_{jk}+d_{jk}\hfill \\ 0\hfill & t<a_{jk}-d_{jk}\hfill \end{array}$$

(11)

$${\mu }_{x_{i}jK}\left(t\right)=\{\begin{array}{ll}1\hfill & t<a_{jK-1}-d_{jK-1}\hfill \\ \frac{\left|t-a_{jK-1}+d_{jK-1}\right|}{2d_{jK-1}}\hfill & a_{jK-1}-d_{jK-1}\le t\le a_{jK-1}+d_{jK-1}\hfill \\ 0\hfill & a_{jK-1}+d_{jK-1}<t\hfill \end{array}$$

(12)

2.2.2. Multi-Index Comprehensive Attribute Measure Analysis

The multi-index comprehensive attribute measure can be obtained by using index weight $w_j$ and single index attribute measure ${\mu }_{x_{i}jk}\left(t\right)=\mu \left(x_{ij}\in C_k\right)$:

$${\mu }_{x_{i}k}=\mu \left(x_i\in C_k\right)={\displaystyle \sum _{j=1}^m{w}_j{\mu }_{x_{i}jk}},$$

(13)

2.2.3. Identification Based on Confidence Attribute

$\left\{C_1,C_2,\cdots ,C_K\right\}$ is an ordered segmentation class of attribute space $F$, $C_1>C_2>\cdots >C_K$ (or $C_1<C_2<\cdots <C_K$), and $\lambda$ is confidence, $0.5<\lambda \le 1$. If the following formula is satisfied, $x_i$ is considered to belong to the $C_k$ level:

$$k_0=\min \left\{k:{\displaystyle {\sum }_{i=1}^k{\mu }_{xi}\ge \lambda ,1\le k\le n}\right\},$$

(14)

or

$$k_0=n-\min \left\{k:{\displaystyle {\sum }_{i=1}^{k-1}{\mu }_{x\left(n-1\right)}\ge \lambda ,1\le k\le n-1}\right\},$$

(15)

2.3. System Dynamics

The safety of filling mining mines is affected by many factors, and the interaction between various factors constitutes a complex system. To establish the system dynamics simulation model, first of all, it is necessary to determine the system boundary, then carry out causality analysis, construct the SD stock flow diagram model, establish the system dynamics equation and carry out model debugging and simulation.

In this paper, after the comprehensive evaluation of the safety level of the filling mining mine, taking Jisuo Coal Mine as an example, from the perspective of safety input, the system dynamics simulation is carried in regard to the five aspects of personnel safety level, filling mining equipment and facilities, filling mining mine safety management, filling mining environment and filling mining technical ability, and the corresponding suggestions are put forward so as to provide theoretical guidance for follow-up filling mining mine safety improvements. By changing the proportion of safety investment in the subsystem and comparing the changes in the safety level of the coal mine, the optimal growth path can be found, the resources can be reasonably allocated, the waste of resources can be reduced, the stable development of the coal mine can be made and accidents can be prevented.

3. Case Study

3.1. General Situation of 16 Coal Seam in Jisuo Coal Mine

Jisuo Coal Mine is located in Tengzhou City, Shandong Province, China. It was built in March 1986 and put into operation in October 1991. The geographical coordinates of the mining area are 116°55′27″–116°58′22″ east longitude and 35°00′07″–35°02′48″ north latitude. The terrain in the field is flat, and the terrain gradually decreases from east to west. The slope of the ground is about one thousandth of that of the lakeside alluvial plain. The mine field weather belongs to the south of the Yellow River in North China, which is a monsoon transition climate. The mine production scale is 350,000 t/a, the mining area is 5.5797 ${\mathrm{k}\mathrm{m}}^2$ and the mining elevation is −50 m–−299 m. The main surface buildings in the mining area are mine industrial squares, industrial parks and villages, and the rest are farmland, forests, orchards, roads, etc. During coal mining, the working face is arranged in tension. Additionally, the longwall-retreating coal mining method is adopted. The high-grade general mining technology, continuous shearer cutting coal, scraper conveyor transporting coal, single hydraulic prop with wooden brick supporting roof and paste filling method are used to manage goaf. For Jisuo Coal Mine, the method of filling mining to mine coal is adopted, and after mining, the goaf is filled via pumping. Considering mining safety and technology, the selected filling material is a paste material with good fluidity composed of coal gangue, cement, fly ash and additives. The filling aggregate is broken coal gangue, construction waste, etc. The cementing material is ordinary Portland cement, and the pipeline transportation performance improver is fly ash.

3.2. The Evaluation Index System Construction

3.2.1. Safety Evaluation Index of Filling Mining Mine

Before the comprehensive evaluation, the evaluation index system is the basis of evaluation. A good and perfect evaluation index system can ensure the accuracy of the safety evaluation results of the filling mining mine, which helps to find the hidden dangers of the mine in time and take remedial measures to ensure the safety and stability of the mine. Our studying of the literature found that coal safety production operations are closely related to “man-machine-management-environment-technology”. It is difficult to prevent and control production safety accidents, mainly because employees’ safety literacy is not high, machinery and equipment are prone to accidents, personnel behavior is difficult to control, the natural environment is difficult to improve and safety technology is not advanced. In order to prevent production safety accidents, coal mining enterprises must improve the safety management level and reduce the incidence of accidents and the degree of accident loss by improving safety management and considering the five aspects of “human—machine—management—environment—technology” [26,27]. This paper therefore constructs as safety evaluation index system for filling mining mines by considering five aspects: personnel literacy, filling mining equipment and facilities, safety management of filling mining mine, filling mining environment and filling mining technical ability.

• Personnel literacy

On the one hand, it has great plasticity, and on the other hand, it is also difficult to account for. Therefore, the improvement of personnel literacy will often improve the overall safety level of the system. The personnel safety level is mainly evaluated from three aspects: education level, safety quality and work skills and experience.

2.

Filling mining equipment and facilities

Equipment and facilities are tools used by people to save time. However, the use of tools also introduces dangers, which may lead to accidents. Filling mining equipment and facilities is a factor mainly evaluated from four aspects: equipment maintenance, effectiveness of safety protection devices, equipment stability and mechanical automation level.

3.

Safety management of filling mining mines

Management is a means of combining multiple aspects through appropriate control and organization in order to achieve the expected goals. Through safety management, it can improve work efficiency, avoid economic losses caused by safety problems and ensure safety levels. The safety management of filling mining mines is mainly evaluated from four aspects: qualified rate of emergency drill, safety system and regulations, safety supervision and education and training level.

4.

Filling mining environment

Coal resources are buried under the ground, and coal mines in various regions also have different natural and working environments. The filling mining environment is mainly evaluated from seven aspects: dust concentration, absolute gas emission, Platts’ coefficient of coal, spontaneous combustion tendency of coal seam, geological structure, dip angle of coal seam and normal water inflow of the mine.

5.

Filling mining technical ability

There are many unsafe factors in coal mine production. With the development of modern science and technology, safety technology has become an important factor affecting coal mine safety production and an important indicator reflecting coal mine safety production levels. Filling mining technology ability is mainly evaluated from four aspects: mining depth, prevention and control technology, method and process innovation and filling body stability.

3.2.2. Index Grading Standards

Through access to information and on-site investigation, 22 indicators were divided into grade standards. In order to better evaluate the safety level of Jisuo Coal Mine, the relevant data of many coal mines were obtained through three months of online search and the inquiry of experts. Then, according to the purpose of evaluation and the principle of index selection, combined with the actual situation of Jisuo Coal Mine, the required evaluation index was preliminarily determined, and then the classification standard of evaluation index was determined. The evaluation index was divided into a quantitative qualitative index. Among them, the quantitative index determines the classification standard according to the “coal mine safety regulations”. Qualitative indicators are graded from 1 to 10 points after multiple consultations and feedback from relevant experts. Then, we compare and evaluate whether the content of the indicators is in line with the actual situation of Jisuo Coal Mine, quantify the qualitative indicators and give the grading standards of the qualitative indicators.

Among them, $Q_1$ indicates education level, $Q_2$ indicates safety quality, $Q_3$ indicates work skills and experience, $Q_4$ indicates equipment maintenance, $Q_5$ indicates effectiveness of safety protection device, $Q_6$ indicates equipment stability, $Q_7$ indicates mechanical automation level, $Q_8$ indicates qualified rate of emergency drill, $Q_9$ indicates safety system and regulations, $Q_{10}$ indicates safety supervision, $Q_{11}$ indicates education and training level, $Q_{12}$ indicates dust concentration, $Q_{13}$ indicates absolute gas emission, $Q_{14}$ indicates Platts’ coefficient of coal, $Q_{15}$ indicates spontaneous combustion tendency of coal seam, $Q_{16}$ indicates geological structure, $Q_{17}$ indicates inclination angle of coal seam, $Q_{18}$ indicates the normal water inflow of the mine, $Q_{19}$ indicates the mining depth, $Q_{20}$ indicates the prevention and control technology, $Q_{21}$ indicates the method and process innovation and $Q_{22}$ indicates the stability of the filling body.

The index classification criteria of $Q_1$~$Q_{11}$ are shown in

Table 2. Qualitative index grading standards.

Evaluation Index	${\mathit{C}}_1$ (Safety)	${\mathit{C}}_2$ (Relatively Safe)	${\mathit{C}}_3$ (General Safety)	${\mathit{C}}_4$ (Unsafe)
Qualitative index	8–10	6–8	4–6	1–4

. The classification criteria of $Q_{12}$~$Q_{18}$ are shown in

Table 3. Classification standard of the filling mining environment index.

Evaluation Index	${\mathit{C}}_1$ (Safety)	${\mathit{C}}_2$ (Relatively Safe)	${\mathit{C}}_3$ (General Safety)	${\mathit{C}}_4$ (Unsafe)
Mean dust concentration	≤1	1–2.5	2.5–5	≥5
Absolute gas emission	≤10	10–40	40–60	≥60
Platts’ coefficient of coal	≤2	2–4	4–8	≥8
Coal seam spontaneous combustion tendency	≥3	2–3	1–2	≤1
Geological structure	≤1	1–2	2–3	≥3
Coal seam dip angle	≤8	8–25	25–45	≥45
Normal discharge of mine water	≤180	180–600	600–2100	≥2100

. $Q_{19}$~$Q_{22}$ index grading standards are shown in

Table 4. Classification standard of filling mining technical ability index.

Evaluation Index	${\mathit{C}}_1$ (Safety)	${\mathit{C}}_2$ (Relatively Safe)	${\mathit{C}}_3$ (General Safety)	${\mathit{C}}_4$ (Unsafe)
Mining depth	≤300	300–600	600–800	≥800
Control technology	8–10	6–8	4–6	1–4
Method and process innovation	8–10	6–8	4–6	1–4
Stability of filling body	8–10	6–8	4–6	1–4

.

3.2.3. State Determination of Each Factor in Jisuo Coal Mine

Jisuo Coal Mine is a production mine. According to the occurrence conditions of 16 coal seams in the mine, based on the classification table of safety evaluation indexes of filling mining, the geological exploration data and quantitative data of 16 coal seams in Jisuo Coal Mine are collected, as shown in

Table 5. Jisuo Coal Mine 16 coal seams for each factor state judgment.

First-Order Index	Secondary Index	Three-Level Index	State Decision
Safety level of filling mining in Grade Suo Coal mine	Personnel literacy	Degree of education	5
		Safety quality	6
		Job skills and experience	7
	Filling mining equipment and facilities	Equipment maintenance	7
		Safety protection device effectiveness	6
		Equipment stability	7
		Mechanical automation level	7
	Safety management of filling mining	Pass rate of emergency drill	6
		Safety rules and regulations	7
		Safety supervision	6.5
		Education and training level	6
	Filling environment	Mean dust concentration	2.5
		Absolute gas emission	0.59
		Platts’ coefficient of coal	2.5
		Coal seam spontaneous combustion tendency	Spontaneous coal seam
		Geological structure	simple
		Coal seam dip angle	8
		Normal discharge of mine water	120
	Technical ability of filling mining	Mining depth	334
		Control technology	8
		Method and process innovation	8
		Stability of filling body	7

.

3.3. The Entropy Weight Method Determines the Weight of Each Index

Through MATLAB software, the entropy weight method is used to calculate the weight of each index. According to Equation (1), the standardized matrix is obtained as follows:

$$P=\left[\begin{array}{rrrrrrrrrrrrrrrrrrrrrr}\hfill 0.7500& \hfill 1.0000& \hfill 0.7500& \hfill 0.7500& \hfill 0.7500& \hfill 0.7500& \hfill 1.0000& \hfill 0.7500& \hfill 0.6667& \hfill 0.5556& \hfill 0.7500& \hfill 0.8039& \hfill 0.9298& \hfill 0.0098& \hfill 1.0000& \hfill 0.5000& \hfill 0.0833& \hfill 0.1429& \hfill 0.3031& \hfill 0.5000& \hfill 0.5000& \hfill 0.5000\\ \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0392& \hfill 0.4862& \hfill 0.3659& \hfill 1.0000& \hfill 0.5000& \hfill 0.1250& \hfill 0.2459& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000\\ \hfill 0.5000& \hfill 0.8333& \hfill 0.5000& \hfill 0.5000& \hfill 0.5000& \hfill 0.5000& \hfill 0.5000& \hfill 0.5000& \hfill 0.6667& \hfill 0.2778& \hfill 0.7500& \hfill 0.0000& \hfill 0.0640& \hfill 0.2195& \hfill 1.0000& \hfill 0.5000& \hfill 0.1667& \hfill 1.0000& \hfill 0.2734& \hfill 0.5000& \hfill 0.5000& \hfill 0.5000\\ \hfill 0.7500& \hfill 1.0000& \hfill 0.0000& \hfill 0.5000& \hfill 0.2500& \hfill 0.0000& \hfill 0.7500& \hfill 0.5000& \hfill 0.0000& \hfill 0.5556& \hfill 0.5000& \hfill 0.6078& \hfill 0.1066& \hfill 0.4195& \hfill 0.0000& \hfill 0.0000& \hfill 1.0000& \hfill 0.1224& \hfill 1.0000& \hfill 1.0000& \hfill 1.0000& \hfill 1.0000\\ \hfill 1.0000& \hfill 0.3333& \hfill 0.5000& \hfill 0.5000& \hfill 0.7500& \hfill 0.5000& \hfill 0.5000& \hfill 0.5000& \hfill 0.6667& \hfill 0.5556& \hfill 0.5000& \hfill 1.0000& \hfill 0.0364& \hfill 0.0244& \hfill 1.0000& \hfill 0.0000& \hfill 0.1250& \hfill 0.0000& \hfill 0.0579& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000\\ \hfill 0.1250& \hfill 0.6667& \hfill 0.5000& \hfill 0.0000& \hfill 0.5000& \hfill 0.2500& \hfill 0.5000& \hfill 0.0000& \hfill 0.6667& \hfill 0.0000& \hfill 0.7500& \hfill 0.0784& \hfill 0.8124& \hfill 0.8049& \hfill 0.5000& \hfill 0.0000& \hfill 0.0833& \hfill 0.3714& \hfill 0.1917& \hfill 0.5000& \hfill 0.0000& \hfill 0.0000\\ \hfill 0.5000& \hfill 0.3333& \hfill 0.2500& \hfill 0.5000& \hfill 0.5000& \hfill 0.7500& \hfill 0.7500& \hfill 0.5000& \hfill 0.6667& \hfill 0.5556& \hfill 0.5000& \hfill 0.0588& \hfill 0.0000& \hfill 1.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.7454& \hfill 0.9970& \hfill 1.0000& \hfill 1.0000& \hfill 1.0000\\ \hfill 0.5000& \hfill 0.7667& \hfill 1.0000& \hfill 1.0000& \hfill 1.0000& \hfill 1.0000& \hfill 1.0000& \hfill 1.0000& \hfill 1.0000& \hfill 1.0000& \hfill 1.0000& \hfill 0.0392& \hfill 1.0000& \hfill 0.0000& \hfill 0.0000& \hfill 1.0000& \hfill 0.3333& \hfill 0.5224& \hfill 0.7192& \hfill 1.0000& \hfill 1.0000& \hfill 0.5000\end{array}\right]$$

According to Equation (2), the P matrix is transformed into a normalized matrix as follows:

$$R=\left[\begin{array}{rrrrrrrrrrrrrrrrrrrrrr}\hfill 0.1818& \hfill 0.2027& \hfill 0.2143& \hfill 0.2000& \hfill 0.1765& \hfill 0.2000& \hfill 0.2000& \hfill 0.2000& \hfill 0.1538& \hfill 0.1587& \hfill 0.1579& \hfill 0.3060& \hfill 0.2706& \hfill 0.0034& \hfill 0.2222& \hfill 0.2000& \hfill 0.0435& \hfill 0.0453& \hfill 0.0856& \hfill 0.1111& \hfill 0.1250& \hfill 0.1429\\ \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0149& \hfill 0.1415& \hfill 0.1286& \hfill 0.2222& \hfill 0.2000& \hfill 0.0652& \hfill 0.0780& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000\\ \hfill 0.1212& \hfill 0.1689& \hfill 0.1429& \hfill 0.1333& \hfill 0.1176& \hfill 0.1333& \hfill 0.1000& \hfill 0.1333& \hfill 0.1538& \hfill 0.0794& \hfill 0.1579& \hfill 0.0000& \hfill 0.0186& \hfill 0.0772& \hfill 0.2222& \hfill 0.2000& \hfill 0.0870& \hfill 0.3174& \hfill 0.0772& \hfill 0.1111& \hfill 0.1250& \hfill 0.1429\\ \hfill 0.1818& \hfill 0.2027& \hfill 0.0000& \hfill 0.1333& \hfill 0.0588& \hfill 0.0000& \hfill 0.1500& \hfill 0.1333& \hfill 0.0000& \hfill 0.1587& \hfill 0.1053& \hfill 0.2313& \hfill 0.0310& \hfill 0.1475& \hfill 0.0000& \hfill 0.0000& \hfill 0.5217& \hfill 0.0389& \hfill 0.2823& \hfill 0.2222& \hfill 0.2500& \hfill 0.2857\\ \hfill 0.2424& \hfill 0.0676& \hfill 0.1429& \hfill 0.1333& \hfill 0.1765& \hfill 0.1333& \hfill 0.1000& \hfill 0.1333& \hfill 0.1538& \hfill 0.1587& \hfill 0.1053& \hfill 0.3806& \hfill 0.0106& \hfill 0.0086& \hfill 0.2222& \hfill 0.0000& \hfill 0.0652& \hfill 0.0000& \hfill 0.0164& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000\\ \hfill 0.0303& \hfill 0.1351& \hfill 0.1429& \hfill 0.0000& \hfill 0.1176& \hfill 0.0667& \hfill 0.1000& \hfill 0.0000& \hfill 0.1538& \hfill 0.0000& \hfill 0.1579& \hfill 0.0299& \hfill 0.2365& \hfill 0.2830& \hfill 0.1111& \hfill 0.0000& \hfill 0.0435& \hfill 0.1179& \hfill 0.0541& \hfill 0.1111& \hfill 0.0000& \hfill 0.0000\\ \hfill 0.1212& \hfill 0.0676& \hfill 0.0714& \hfill 0.1333& \hfill 0.1176& \hfill 0.2000& \hfill 0.1500& \hfill 0.1333& \hfill 0.1538& \hfill 0.1587& \hfill 0.1053& \hfill 0.0224& \hfill 0.0000& \hfill 0.3516& \hfill 0.0000& \hfill 0.0000& \hfill 0.0000& \hfill 0.2366& \hfill 0.2815& \hfill 0.2222& \hfill 0.2500& \hfill 0.2857\\ \hfill 0.1212& \hfill 0.1554& \hfill 0.2857& \hfill 0.2667& \hfill 0.2353& \hfill 0.2667& \hfill 0.2000& \hfill 0.2667& \hfill 0.2308& \hfill 0.2857& \hfill 0.2105& \hfill 0.0149& \hfill 0.2911& \hfill 0.0000& \hfill 0.0000& \hfill 0.4000& \hfill 0.1739& \hfill 0.1658& \hfill 0.2030& \hfill 0.2222& \hfill 0.2500& \hfill 0.1429\end{array}\right]$$

Finally, the weight of each index is shown in

Table 6. Weight of each index.

First-Order Index	Secondary Index	Three-Level Index	Weight
Safety level of filling mining in Grade Suo Coal mine	Personnel literacy	Degree of education	0.0275
		Safety quality	0.0236
		Job skills and experience	0.0419
	Filling mining equipment and facilities	Equipment maintenance	0.0375
		Safety protection device effectiveness	0.0232
		Equipment stability	0.0415
		Mechanical automation level	0.0199
	Safety management of filling mining	Pass rate of emergency drill	0.0375
		Safety rules and regulations	0.0342
		Safety supervision	0.0399
		Education and training level	0.0189
	Filling environment	Mean dust concentration	0.0789
		Absolute gas emission	0.0589
		Platts’ coefficient of coal	0.0625
		Coal seam spontaneous combustion tendency	0.0566
		Geological structure	0.0848
		Coal seam dip angle	0.0675
		Normal discharge of mine water	0.0407
	Technical ability of filling mining	Mining depth	0.0464
		Control technology	0.0391
		Method and process innovation	0.0590
		Stability of filling body	0.0601

.

3.4. Evaluation of Attribute Mathematics Theory

3.4.1. Single Index Attribute Measure Function

According to Equations (5)–(12), the single index attribute measure function is obtained. The single index attribute measure function of the qualitative index is shown in

Table 7. Single index attribute measure function of the qualitative index.

Evaluation Index	${\mathit{C}}_1$	${\mathit{C}}_2$	${\mathit{C}}_3$	${\mathit{C}}_4$
$Q_i$	${\mu }_{j1}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<7\hfill \\ \frac{t-7}{2},\hfill & 7\le t\le 9\hfill \\ 1,\hfill & t>9\hfill \end{array}$	${\mu }_{j2}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<5\hfill \\ \frac{t-5}{2},\hfill & 5\le t<7\hfill \\ \frac{9-t}{2},\hfill & 7\le t\le 9\hfill \\ 0,\hfill & t>9\hfill \end{array}$	${\mu }_{j3}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<3\hfill \\ \frac{t-3}{2},\hfill & 3\le t<5\hfill \\ \frac{7-t}{2},\hfill & 5\le t\le 7\hfill \\ 0,\hfill & t>7\hfill \end{array}$	${\mu }_{j4}\left(t\right)=\{\begin{array}{ll}1,\hfill & t<3\hfill \\ \frac{5-t}{2},\hfill & 3\le t\le 5\hfill \\ 0,\hfill & t>5\hfill \end{array}$

, and the single index attribute measure function of the quantitative index is shown in

Table 8. Single index attribute measure function of the quantitative index.

Evaluation Index	${\mathit{C}}_1$	${\mathit{C}}_2$	${\mathit{C}}_3$	${\mathit{C}}_4$
$Q_{12}$	${\mu }_{121}\left(t\right)=\{\begin{array}{ll}1,\hfill & t<\frac{1}{2}\hfill \\ \frac{3-2t}{2},\hfill & \frac{1}{2}\le t\le \frac{3}{2}\hfill \\ 0,\hfill & t>\frac{3}{2}\hfill \end{array}$	${\mu }_{122}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<\frac{1}{2}\hfill \\ \frac{2t-1}{2},\hfill & \frac{1}{2}\le t\le \frac{3}{2}\hfill \\ 1,\hfill & \frac{3}{2}<t<\frac{7}{4}\hfill \\ \frac{13-4t}{6},\hfill & \frac{7}{4}\le t\le \frac{13}{4}\hfill \\ 0,\hfill & t>\frac{13}{4}\hfill \end{array}$	${\mu }_{123}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<\frac{7}{4}\hfill \\ \frac{4t-7}{6},\hfill & \frac{7}{4}\le t\le \frac{13}{4}\hfill \\ 1,\hfill & \frac{13}{4}<t<\frac{15}{4}\hfill \\ \frac{25-4t}{10},\hfill & \frac{15}{4}\le t\le \frac{25}{4}\hfill \\ 0,\hfill & t>\frac{25}{4}\hfill \end{array}$	${\mu }_{124}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<\frac{15}{4}\hfill \\ \frac{4t-15}{10},\hfill & \frac{15}{4}\le t\le \frac{25}{4}\hfill \\ 1,\hfill & t>\frac{25}{4}\hfill \end{array}$
$Q_{13}$	${\mu }_{131}\left(t\right)=\{\begin{array}{ll}1,\hfill & t<5\hfill \\ \frac{15-t}{10},\hfill & 5\le t\le 15\hfill \\ 0,\hfill & t>15\hfill \end{array}$	${\mu }_{132}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<5\hfill \\ \frac{t-5}{10},\hfill & 5\le t\le 15\hfill \\ 1,\hfill & 15<t<30\hfill \\ \frac{50-t}{20},\hfill & 30\le t\le 50\hfill \\ 0,\hfill & t>50\hfill \end{array}$	${\mu }_{133}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<30\hfill \\ \frac{t-30}{20},\hfill & 30\le t\le 50\hfill \\ \frac{70-t}{20},\hfill & 50<t\le 70\hfill \\ 0,\hfill & t>70\hfill \end{array}$	${\mu }_{134}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<50\hfill \\ \frac{t-50}{20},\hfill & 50\le t\le 70\hfill \\ 1,\hfill & t>70\hfill \end{array}$
$Q_{14}$	${\mu }_{141}\left(t\right)=\{\begin{array}{ll}1,\hfill & t<\frac{23}{20}\hfill \\ \frac{57-20t}{34},\hfill & \frac{23}{20}\le t\le \frac{57}{20}\hfill \\ 0,\hfill & t>\frac{57}{20}\hfill \end{array}$	${\mu }_{142}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<\frac{23}{20}\hfill \\ \frac{20t-23}{34},\hfill & \frac{23}{20}\le t\le \frac{57}{20}\hfill \\ 1,\hfill & \frac{57}{20}<t<3\hfill \\ \frac{5-t}{2},\hfill & 3\le t\le 5\hfill \\ 0,\hfill & t>5\hfill \end{array}$	${\mu }_{143}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<3\hfill \\ \frac{t-3}{2},\hfill & 3\le t\le 5\hfill \\ 1,\hfill & 5<t<6\hfill \\ \frac{10-t}{4},\hfill & 6\le t\le 10\hfill \\ 0,\hfill & t>10\hfill \end{array}$	${\mu }_{144}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<6\hfill \\ \frac{t-6}{4},\hfill & 6\le t\le 10\hfill \\ 1,\hfill & t>10\hfill \end{array}$
$Q_{15}$	${\mu }_{151}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<\frac{5}{2}\hfill \\ \frac{2t-5}{2},\hfill & \frac{5}{2}\le t\le \frac{7}{2}\hfill \\ 1,\hfill & t>\frac{7}{2}\hfill \end{array}$	${\mu }_{152}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<\frac{3}{2}\hfill \\ \frac{2t-3}{2},\hfill & \frac{3}{2}\le t<\frac{5}{2}\hfill \\ \frac{7-2t}{2},\hfill & \frac{5}{2}\le t\le \frac{7}{2}\hfill \\ 0,\hfill & t>\frac{7}{2}\hfill \end{array}$	${\mu }_{153}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<\frac{1}{2}\hfill \\ \frac{2t-1}{2},\hfill & \frac{1}{2}\le t<\frac{3}{2}\hfill \\ \frac{5-2t}{2},\hfill & \frac{3}{2}\le t\le \frac{5}{2}\hfill \\ 0,\hfill & t>\frac{5}{2}\hfill \end{array}$	${\mu }_{154}\left(t\right)=\{\begin{array}{ll}1,\hfill & t<\frac{1}{2}\hfill \\ \frac{3-2t}{2},\hfill & \frac{1}{2}\le t\le \frac{3}{2}\hfill \\ 0,\hfill & t>\frac{3}{2}\hfill \end{array}$
$Q_{16}$	${\mu }_{161}\left(t\right)=\{\begin{array}{ll}1,\hfill & t<\frac{1}{2}\hfill \\ \frac{3-2t}{2},\hfill & \frac{1}{2}\le t\le \frac{3}{2}\hfill \\ 0,\hfill & t>\frac{3}{2}\hfill \end{array}$	${\mu }_{162}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<\frac{1}{2}\hfill \\ \frac{2t-1}{2},\hfill & \frac{1}{2}\le t\le \frac{3}{2}\hfill \\ \frac{5-2t}{2},\hfill & \frac{3}{2}<t\le \frac{5}{2}\hfill \\ 0,\hfill & t>\frac{5}{2}\hfill \end{array}$	${\mu }_{163}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<\frac{3}{2}\hfill \\ \frac{2t-3}{2},\hfill & \frac{3}{2}\le t\le \frac{5}{2}\hfill \\ \frac{7-2t}{2},\hfill & \frac{5}{2}<t\le \frac{7}{2}\hfill \\ 0,\hfill & t>\frac{7}{2}\hfill \end{array}$	${\mu }_{164}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<\frac{5}{2}\hfill \\ \frac{2t-5}{2},\hfill & \frac{5}{2}\le t\le \frac{7}{2}\hfill \\ 1,\hfill & t>\frac{7}{2}\hfill \end{array}$
$Q_{17}$	${\mu }_{171}\left(t\right)=\{\begin{array}{ll}1,\hfill & t<4\hfill \\ \frac{t-4}{8},\hfill & 4\le t\le 12\hfill \\ 0,\hfill & t>12\hfill \end{array}$	${\mu }_{172}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<4\hfill \\ \frac{t-4}{8},\hfill & 4\le t\le 12\hfill \\ 1,\hfill & 12<t<\frac{33}{2}\hfill \\ \frac{67-2t}{34},\hfill & \frac{33}{2}\le t\le \frac{67}{2}\hfill \\ 0,\hfill & t>\frac{67}{2}\hfill \end{array}$	${\mu }_{173}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<\frac{33}{2}\hfill \\ \frac{2t-33}{34},\hfill & \frac{33}{2}\le t\le \frac{67}{2}\hfill \\ 1,\hfill & \frac{67}{2}<t<35\hfill \\ \frac{55-t}{20},\hfill & 35\le t\le 55\hfill \\ 0,\hfill & t>55\hfill \end{array}$	${\mu }_{174}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<35\hfill \\ \frac{t-35}{20},\hfill & 35\le t\le 55\hfill \\ 0,\hfill & t>55\hfill \end{array}$
$Q_{18}$	${\mu }_{181}\left(t\right)=\{\begin{array}{ll}1,\hfill & t<90\hfill \\ \frac{270-t}{180},\hfill & 90\le t\le 270\hfill \\ 0,\hfill & t>270\hfill \end{array}$	${\mu }_{182}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<90\hfill \\ \frac{t-90}{180},\hfill & 90\le t\le 270\hfill \\ 1,\hfill & 270<t<390\hfill \\ \frac{810-t}{420},\hfill & 390\le t\le 810\hfill \\ 0,\hfill & t>810\hfill \end{array}$	${\mu }_{183}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<390\hfill \\ \frac{t-390}{420},\hfill & 390\le t\le 810\hfill \\ 1,\hfill & 810<t<1350\hfill \\ \frac{2850-t}{1500},\hfill & 1350\le t\le 2850\hfill \\ 0,\hfill & t>2850\hfill \end{array}$	${\mu }_{184}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<1350\hfill \\ \frac{t-1350}{1500},\hfill & 1350\le t\le 2850\hfill \\ 1,\hfill & t>2850\hfill \end{array}$
$Q_{19}$	${\mu }_{191}\left(t\right)=\{\begin{array}{ll}1,\hfill & t<150\hfill \\ \frac{450-t}{300},\hfill & 150\le t\le 450\hfill \\ 0,\hfill & t>450\hfill \end{array}$	${\mu }_{192}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<150\hfill \\ \frac{t-150}{300},\hfill & 150\le t\le 450\hfill \\ 1,\hfill & 450<t<500\hfill \\ \frac{700-t}{200},\hfill & 500\le t\le 700\hfill \\ 0,\hfill & t>700\hfill \end{array}$	${\mu }_{193}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<500\hfill \\ \frac{t-500}{200},\hfill & 500\le t\le 700\hfill \\ \frac{900-t}{200},\hfill & 700<t\le 900\hfill \\ 0,\hfill & t>900\hfill \end{array}$	${\mu }_{194}\left(t\right)=\{\begin{array}{ll}0,\hfill & t<700\hfill \\ \frac{t-700}{200},\hfill & 700\le t\le 900\hfill \\ 1,\hfill & t>900\hfill \end{array}$

.

3.4.2. Single Index Attribute Measure Calculation

According to the field measured values and the measure functions in Table 7 and Table 8, the single index attribute measure matrix is as follows:

$$F=\left[\begin{array}{cccccccccccccccccccccc}0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 1& 0.2059& 0& 0.5& 0.5& 0.8333& 0.3867& 0.5& 0.5& 0\\ 0& 0.5& 1& 1& 0.5& 1& 1& 0.5& 1& 0.75& 0.5& 0.5& 0& 0.7941& 0& 0.5& 0.5& 0.1667& 0.6133& 0.5& 0.5& 1\\ 1& 0.5& 0& 0& 0.5& 0& 0& 0.5& 0& 0.25& 0.5& 0.5& 0& 0& 0.5& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0.5& 0& 0& 0& 0& 0& 0& 0\end{array}\right]$$

3.4.3. Multi-Index Attribute Measure Calculation and Comprehensive Attribute Recognition

According to the test and collected data of each index, the value of each index is obtained after sorting out. According to the corresponding calculation formula of the single index attribute measure function, the single index attribute measure of each index is obtained. Then, according to Equation (13), the evaluation vector of the multi-index attribute measure is calculated as $\alpha =(0.2488,0.5661,0.1568,0.0283)$. The value range of confidence $\lambda$ is $0.5<\lambda <1$. After consulting the literature, it is found that $\lambda =0.6$ is basically taken in practical application, and the safety level can be determined more accurately when $\lambda =0.6$. Therefore, $\lambda =0.6$ is adopted in this paper. According to the attribute recognition of Equations (14) and (15), the safety grade of filling mining in 16 coal seams in the working face of Jisuo Coal Mine is “relatively safe”, which is consistent with the actual situation. The feasibility and effectiveness of this method for the safety evaluation of the filling mining mine are verified, which provides a new method for the safety evaluation of filling mining mines and enriches the theoretical research related to the evaluation model.

3.5. SD Simulation Model Construction

3.5.1. Determining System Boundaries

According to the SD modeling principle and the safety evaluation index system proposed above, the system dynamics model boundary of the safety level of the mine is constructed, and the simulation model of the safety level of the mine is established from the perspective of the safety input. The time boundary of the system simulation is set to 48 months, and the simulation step is 1 month.

3.5.2. Simulation Variable Determination

In order to quantitatively analyze the interaction between variables and better establish the SD simulation model of the safety level of the filling mining mine, it is necessary to accurately define the variables. The variables can be divided into four categories: level variables, rate variables, auxiliary variables and constants. The variables and their meanings for the model are shown in

Table 9. SD model variable meaning table.

Variable Type	Symbol	Variable Meaning
Level variable	L1	Personnel safety level
	L2	Safety level of equipment and facilities
	L3	Safety management level
	L4	Environmental safety level
	L5	Safety technical competence level
Rate variable	R1	The increment in personnel safety level per unit time
	R2	The increment in safety level of equipment and facilities per unit time
	R3	The increment in security management level per unit time
	R4	The increment in environmental safety level per unit time
	R5	The increment in security technical capability level per unit time
Auxiliary variable	L	Safety level of filling mining
	A11	The variation in personnel educational degree factor in unit time
	A12	The variation in personnel safety quality factor in unit time
	A13	The change in the work skill and experience factor per unit of time
	A21	The change in the equipment maintenance factor per unit time
	A22	The variation in the effectiveness factor of the safety protection device per unit time
	A23	The change in the stability factor of the device per unit time
	A24	The change in the mechanical automation level factor per unit time
	A31	The variation in the pass rate factor of emergency drill in unit time
	A32	The change in the educational training level factor per unit time
	A41	The change in the dust concentration factor per unit time
	A42	The change in the absolute gas emission factor in unit time
	A51	The variation in the mining depth factor per unit time
	A52	The variation in the control technology factor per unit time
	A53	The variation in the method and process innovation factor per unit time
	A54	The variation in the stability factor of the filling body per unit time
	ZHL11	The conversion rate of personnel safety input to personnel education factor
	ZHL12	The conversion rate of personnel safety input to personnel safety quality factor
	ZHL13	The conversion rate of personnel safety input to job skill and experience factor
	ZHL21	The conversion rate of equipment and facility safety input to equipment maintenance factor
	ZHL22	The conversion rate of equipment and facility safety input to mechanical automation level factor
	ZHL31	The conversion rate of safety management input to qualification factor of emergency drill
	ZHL32	The conversion rate of the safety management input to education and training level factor
	ZHL41	The conversion rate of environmental safety inputs to dust concentration factors
	ZHL42	The conversion rate of environmental safety input to absolute gas emission factor
	ZHL51	The conversion rate of safety technology input to mining depth factor
	ZHL52	The conversion rate of safety technology input to prevention technology factor
	ZHL53	The conversion rate of safety technology input to method and process innovation factor
	ZHL54	The conversion rate of safety technology input to stability factor of backfill
	B1	Percentage of personnel safety investment
	B2	Percentage of safety investment in equipment and facilities
	B3	Percentage of investment in security management
	B4	Environmental safety investment percentage
	B5	Percentage of investment in safety technology
	B11	Personnel education factor input percentage
	B12	Percentage of personnel safety quality factor input
	B13	Job skills and experience factor input percentage
	B21	Equipment maintenance factor input percentage
	B22	Mechanical automation level factor input percentage
	B31	Emergency drill-qualified rate factor input percentage
	B32	Education and training level factor input percentage
	B41	Dust concentration factor input percentage
	B42	Absolute gas emission factor input percentage
	B51	Mining depth factor input percentage
	B52	Percentage of investment in control technology factor
	B53	Method and process innovation factor input percentage
	B54	Filling body stability factor input percentage
Constant	YX21	The influence coefficient of equipment maintenance on equipment stability
	YX22	The influence factor of equipment maintenance on the effectiveness of safety protection devices
	YX31	The influence coefficient of safety system and regulations on safety management level
	YX32	The influence coefficient of safety supervision on safety management level
	YX41	Influence coefficient of Pratts’ coefficient of coal on environmental safety level
	YX42	Influence coefficient of coal seam spontaneous combustion tendency on environmental safety level
	YX43	Influence coefficient of geological structure on environmental safety level
	YX44	Influence coefficient of coal seam dip angle on environmental safety level
	YX45	Influence coefficient of mine normal water inflow on environmental safety level

. In addition, if the dimension is inconsistent, it will lead to obstacles in the calculation process. Therefore, after comprehensively analyzing the applicability of the model, it is unified into a dimensionless state.

3.5.3. SD Model Construction

In this paper, Vensim simulation software is selected for the study, and the system dynamics model of the safety level of the filling mining mine is constructed in the software [28], as shown in Figure 1.

The system dynamics equation of the personnel safety level subsystem is as follows:

$$L1{\left(\mathrm{t}\right)}_K=L1{(t)}_J+(DT)\times R1(t),$$

(16)

$$R1(t)=A11\times W11+A12\times W12+A13\times W13,$$

(17)

$$SI1=SI\times B1,$$

(18)

$$A11=SI1\times B11\times ZHL11,$$

(19)

$$A12=SI1\times B12\times ZHL12,$$

(20)

$$A13=SI1\times B13\times ZHL13,$$

(21)

The system dynamics equations of the equipment and facilities safety level subsystem are as follows:

$$L2{\left(\mathrm{t}\right)}_K=L2{(t)}_J+(DT)\times R2(t),$$

(22)

$$R2(t)=A21\times W21+A22\times W22+A23\times W23+A24\times W24,$$

(23)

$$SI2=SI\times B2,$$

(24)

$$A21=SI2\times B21\times ZHL21,$$

(25)

$$A22=A21\times YX22,$$

(26)

$$A23=A21\times YX21,$$

(27)

$$A24=SI2\times B22\times ZHL22,$$

(28)

The system dynamics equation of the safety management level subsystem is as follows:

$$L3{\left(\mathrm{t}\right)}_K=L3{(t)}_J+(DT)\times R3(t),$$

(29)

$$R3(t)=(A31\times W31+A32\times W32)\times YX31\times YX32,$$

(30)

$$SI3=SI\times B3,$$

(31)

$$A31=SI3\times B31\times ZHL31,$$

(32)

$$A32=SI3\times B32\times ZHL32,$$

(33)

The system dynamics equation of the environmental safety level subsystem is as follows:

$$L4{\left(\mathrm{t}\right)}_K=L4{(t)}_J+(DT)\times R4(t),$$

(34)

$$R4(t)=(A41\times W41+A42\times W42)\times YX41\times YX42\times YX43\times YX44\times YX45,$$

(35)

$$SI4=SI\times B4,$$

(36)

$$A41=SI4\times B41\times ZHL41,$$

(37)

$$A42=SI4\times B42\times ZHL42,$$

(38)

The system dynamics equation of the safety technology capability level subsystem is as follows:

$$L5{\left(\mathrm{t}\right)}_K=L5{(t)}_J+(DT)\times R5(t),$$

(39)

$$R5(t)=A51\times W51+A52\times W52+A53\times W53+A54\times W54,$$

(40)

$$SI5=SI\times B5,$$

(41)

$$A51=SI5\times B51\times ZHL51,$$

(42)

$$A52=SI5\times B52\times ZHL52,$$

(43)

$$A53=SI5\times B53\times ZHL53,$$

(44)

$$A54=SI5\times B54\times ZHL54,$$

(45)

The system dynamics equation of the safety level of the filling mining mine is as follows:

$$L_K=L{1}_K\times W1+L{2}_K\times W2+L{3}_K\times W3+L{4}_K\times W4+L{5}_K\times W5$$

(46)

The system dynamics equation of the safety input subsystem is as follows:

$$SI=SI1+SI2+SI3+SI4+SI5,$$

(47)

In the formula, $K$ is the present moment, $J$ is the past moment, $DT$ is the time step and the unit is month.

3.5.4. Model Checking and Simulation Results Analysis

Model checking is one of the key parts of the whole modeling process, which is related to whether the situation reflected by the model is true or not and determines the accuracy and effectiveness of the whole simulation analysis results. In this paper, two test methods are selected: the dimensional test and extreme value test. All aspects of the model are effectively tested via these two methods, and the simulation results are basically consistent with the actual situation of Jisuo Coal Mine. Therefore, the model has passed the test.

The coal mine safety input parameter value is set to 10, and the SD model is used for simulation. The trend chart and simulation data of the personnel safety level, equipment and facility safety level, safety management level, environmental safety level, safety technical ability level and filling mining mine safety level can be obtained, as shown in Figure 2, Figure 3, Figure 4, Figure 5, Figure 6 and Figure 7. In the figure, “Dmnl” is the abbreviation of dimensionless, which means there is no unit.

From Figure 2, Figure 3, Figure 4, Figure 5, Figure 6 and Figure 7, it can be seen that the safety level of the filling mining mine and the five level variables all increase with time. In the early stage, the growth rate of the five level variables and the safety level of the filling mining mine is relatively slow, and the growth of the safety level lags behind. This is because the system has a delay, and the safety input has a lag. After the safety investment, it takes time to make the corresponding decision, and the change of the system state also takes time. Therefore, the initial overall growth rate is relatively slow. In the medium term, the growth rate of the five level variables and the safety level of the filling mining mine is faster. This is due to the accumulation of safety investment over time, effective implementation of various safety procedures and the improvement of all aspects of the safety level. Among them, according to the simulation data of level variables, the safety level of equipment and facilities and the environmental safety level reached more than 90 in the ninth month and then gradually approached the limit value. The safety level of personnel, safety management level and safety technical ability level reached more than 90 in the 21st, 22nd and 25ths month, respectively, and then gradually approached the limit value. It can be seen that the improvement effect of the environment and equipment is faster, and the improvement of personnel, management and technology can only achieve initial results after a period of implementation. In the later period, the growth rate of the five level variables and the safety level of the filling mining mine is basically unchanged. At this time, the safety level has reached the expected value, and the safety investment can be appropriately reduced to save resources.

Through the model simulation, it is concluded that the change trend of the safety level of each variable is consistent with the reality under the stable safety input. Therefore, the constructed system dynamics model has high reliability and can be used to compare the safety input effects and action rules under different safety input schemes so as to rationally optimize resource allocation and maximize resource benefits.

3.5.5. Comparison of Safety Input Schemes

In order to analyze the influence of different safety input factors on the total safety input effect, five different input schemes (see

Table 10. Subsystem scale adjustment comparison scheme table.

	Factor	Personnel Safety Investment Ratio	Proportion of Investment in Equipment and Facilities Safety	Security Management Investment Ratio	Environmental Safety Investment Ratio	Proportion of Investment in Safety Technology
Scheme
Scheme 1		0.6	0.1	0.1	0.1	0.1
Scheme 2		0.1	0.6	0.1	0.1	0.1
Scheme 3		0.1	0.1	0.6	0.1	0.1
Scheme 4		0.1	0.1	0.1	0.6	0.1
Scheme 5		0.1	0.1	0.1	0.1	0.6

) were designed for simulation [29], and the changes in the safety level of the filling mining mines under different schemes were analyzed, so as to provide theoretical basis for the adjustment of the safety input schemes of coal enterprises.

The parameters are modified according to the schemes in the table, and the simulation is carried out in Vensim software to obtain the change trend diagram of the safety level of the filling mining mine of each scheme, as shown in Figure 8.

From Figure 8, it can be seen that in the early stage, for the improvement of the safety level of filling mining mines, the improvement rate of Scheme 2 and Scheme 4 is greater than that of other schemes; in terms of the size of the safety level of filling mining mines, the order is Scheme 1 > Scheme 2 > Scheme 5 > Scheme 3 > Scheme 4, but the final safety level of the filling mining mine is not much different.

By comparing the simulation results, it can be concluded that because of the delay of the system, the five schemes have a slow increase for a period of time in the early stage of improvement. Increasing equipment safety investment and environmental safety investment can quickly improve the safety level because the maintenance of equipment can improve the stability and safety of equipment rapidly, in addition to the dust concentration and gas emission also being improved efficiently by taking certain engineering measures. Therefore, the safety level of equipment and environment improves faster than that of personnel, management and technical ability. However, because the influencing factors of the environmental subsystem contain many natural factors, and some natural factors cannot be improved, the increase in environmental safety investment makes the improvement of the safety level of the filling mining mine limited and less than the improvement effect of the other four subsystems. Increasing the safety input of the personnel subsystem can maximize the safety level of the filling mining mine, followed by the equipment subsystem, the technical subsystem third, the management subsystem fourth and the environmental subsystem fifth. This shows that Jisuo Coal Mine should focus its resources on four aspects: staff literacy, equipment and facilities, technical ability and safety management. Jisuo Coal Mine can strengthen the quality of employees by increasing the proportion of “soft power” input so as to maximize the effect of resource utilization and realize the rational allocation of resources.

When the level of Jisuo Coal Mine safety is low, the proportion of equipment safety investment and environmental safety investment can be increased to quickly improve the level of coal mine safety and avoid accidents. When the level of Jisuo Coal Mine safety is at a sufficient level, the proportion of personnel safety investment can be increased to keep coal mine safety at its highest possible level.

4. Countermeasures and Suggestions

The production system of a coal mine is extremely complex, composed of many subsystems such as the coal mining system, filling system, tunneling system, drainage system, transportation system and so on. Each system is composed of different subsystems, and each system involves humans, machines, the environment and other aspects, and is affected by external social, economic and human factors. Therefore, reasonable organization and arrangement are conducive to improving mine production efficiency and reducing the occurrence of disasters [30]. In the process of filling mining, the geological conditions of different mines are different, and the performance requirements of filling materials are also different. Under the condition of a hard roof, the instantaneous compression deformation performance standard of filling materials can be reduced accordingly. Additionally, the higher the protection level of surface buildings in a mining area is, the more the creep compression deformation performance of filling materials needs to be adjusted accordingly [31]. With the continuous development and innovation of mining science and technology, the mining industry is developing in a green and intelligent manner, and the improvement of technology is of great benefit to the realization of precise, safe and intelligent mining [32]. In view of the safety production of filling mining mines, from the perspective of the long-term development of safety production of coal enterprises it is necessary to improve the safety level of filling mining mines and rationally allocate coal mine safety investment so as to further improve the safety production status of mines, avoid waste of resources and reduce accidents. The specific measures are as follows.

(1)

Personnel literacy

In the production and operation activities of coal mining enterprises, personnel safety has always been the focus of safety work in coal mining enterprises. “People are the product of the environment”, and a good safe production environment is a favorable guarantee for coal mine safety production. Coal mining enterprises should pay attention to personnel education and training and technical exchange and learning, and constantly optimize the “soft environment” of the mining area. Managers should take the lead in demonstration and create a culture of good safety so that employees can constantly check their bad behaviors and standardize their own operations so as to continuously improve their safety literacy.

(2)

Filling mining equipment and facilities

Optimizing equipment and facilities can quickly improve the safety level of filling mining mines. Enterprise managers should pay attention to improve the completeness, intact rate, update rate and maintenance level of protective equipment, production equipment and facilities, standardize the management of tools and equipment and constantly improve the performance of equipment and facilities so as to reduce the safety production accidents and casualties caused by equipment and facilities.

(3)

Safety management of filling mining mines

The safety management of filling mining mines is the key to improve the safety ability of the coal mine system. Enterprises should constantly improve the safety supervision system and establish a targeted, scientific and effective incentive mechanism to help mobilize the enthusiasm of employees for safe production. Additionally, the construction of safety information systems and regularly carrying out all-round, multi-level safety management assessment work should be observed. Additionally, mines should aim to strengthen safety education and training, reduce the occurrence of safety accidents from the source and continuously improve the safety production efficiency of coal mine enterprises.

(4)

Filling mining environment

The coal mine industry is a high-risk industry. The environmental problems in the mining area involve mining, processing, filling, storage and transportation. Green mining is an inevitable trend within mining development. Coal mining enterprises should establish a mining area environmental monitoring system, systematically understand and analyze the mining area environment and formulate targeted mining area environmental governance measures according to the different environmental conditions of coal mines and local conditions. In addition, mines should aim to improve the overall process; control dust, gas and water inflow; adopt the method of “prevention-treatment-management”; coordinate the development of various environmental themes and ensure the sustainable development and green development of coal mine production and environmental protection.

(5)

Filling mining technical ability

The technical ability of filling mining is an indispensable driving force for the development of coal mine production safety. Coal mining enterprises should make full use of filling mining technology to improve mining safety, oversee the development of new high-strength filling materials, pay attention to the improvement and innovation of production technology and build a professional and innovative technical personnel team through training and introduction. Solving the safety problems of coal mining enterprises provides help for the safety management, technical engineering and disaster management of the coal industry, ultimately consolidating the sustainable development of the coal industry.

5. Conclusions

Taking Jisuo Coal Mine as an example, this paper innovatively proposes a safety evaluation method based on entropy weight–attribute mathematical theory, making an objective and accurate safety evaluation of Jisuo Coal Mine. Finally, the simulation model is established by using system dynamics, and combined with the conclusion of safety evaluation, the corresponding countermeasures are given to improve the safety level of Jisuo Coal Mine. This study can provide an objective and accurate new method for the safety evaluation of filling mining mines from a quantitative point of view and enrich the theoretical research related to the safety evaluation model of filling mining mines. The conclusions are as follows.

(1)

Combined with the influencing factors of filling mining mine safety and the construction principle of the evaluation index system, the safety evaluation index system of filling mining mines is determined, and each index in the evaluation system is quantified.

(2)

The entropy weight method is used to calculate the weight of each index, and then the attribute measurement function is constructed by using the attribute mathematical method, which conforms to the characteristics of the safety evaluation of the filling mining mines. The comprehensive evaluation model of entropy weight–attribute mathematics is established to evaluate the safety level of Jisuo Coal Mine. The evaluation result demonstrates a state of “relative safety”. The evaluation results are in good agreement with the actual situation of Jisuo Coal Mine, which verifies the applicability of entropy weight–attribute mathematical theory.

(3)

The SD model of safety level of the filling mining mine is constructed. According to the obtained index weight value and various simulation parameters, the system dynamics simulation of Jisuo Coal Mine is carried out from the perspective of safety input. Additionally, the simulation comparison of different schemes is carried out, and the change trend of the safety level of the filling mining mine and the relationship between the subsystems are analyzed. Additionally, the improvement measures of the safety of the filling mining mine are investigated in terms of five aspects in order to realize the rational optimization of resource allocation and consolidate the sustainable development of the coal industry.