Site Selection of Multi-Level Material Bases in Coal Mine Considering the Hierarchical Reserve Mode and Demand Difference—A Case Study

As one of the important energy industries, the safe production of coal mines not only concerns the safety of employees, but also affects the sustainable development of society. Recent rescue cases have shown that the layout of material bases and the allocation of emergency materials need to be further optimized. For this problem, this paper adopts the material hierarchical reserve mode for the first time, combining the material hierarchical reserve mode, multi-level material bases selection model, and the Fuzzy Analytic Hierarchy Process (FAHP) method to determine the optimal location of the coal mine material bases in Liaoning Province. Determine the reserve level of emergency materials through evaluation and clustering analysis of various emergency materials. Improve the traditional gravity model and construct a multi-objective planning model for multi-level material bases site selection. Apply the reserve level of emergency materials and real regional data to the site selection model, and use the Non-dominated Sorting Genetic Algorithm III (NSGA-III) to solve the model. Then, the optimal site selection scheme was selected using FAHP, and sensitivity analysis was conducted. The research results indicate that: 1) 9 county-level, 6 city-level and 2 provincial-level material bases will be constructed; 2) location selection of material bases is not only affected by costs, but also by the density of road networks; 3) use of hierarchical material reserve mode can significantly reduce the construction cost of material bases. Compared to the traditional material reserve mode, it can reduce costs by 86.6%. Compared to a single-level material bases, it can still reduce costs by 74.7%. The research results of this paper can be used for the selection of emergency material bases and the allocation of emergency material reserves, which can reduce costs and promote the improvement of regional emergency rescue capabilities.


I. INTRODUCTION A. BACKGROUND
Coal mining is an essential component of China's energy industry, with one of the highest risk levels [1], [2].Recent years' data demonstrate a decline in coal mining fatalities; nonetheless, accidents continue to occur frequently [3].To enhance rescue efficiency and minimize losses in the The associate editor coordinating the review of this manuscript and approving it for publication was Emanuele Crisostomi .event of accidents, the Chinese government has implemented laws and regulations mandating that coal mining enterprises construct material bases to store necessary equipment and materials for rescue operations.Furthermore, coal mine rescue teams are equipped with essential equipment and materials for executing coal mine accident rescues.These materials play a crucial role in accurate rescue operations, contributing to the reduction of casualties, economic losses, and the maintenance of social stability and sustainable development [4].
According to a report from the Ministry of Emergency Management of China, the provision of emergency materials during the rescue process of several coal mine accidents has improved the effectiveness of rescue and reduced casualties and economic losses [5].Nevertheless, there exist some problems such as outdated rescue equipment and lack of maintenance.During the accident rescue operation, there is a requirement to borrow rescue materials and equipment from distant sources, which significantly impacts the rescue duration [6].Therefore, how to optimize the configuration of emergency materials is particularly important to improve the rescue efficiency [7].

B. MOTIVATION
Based on the lessons learned from various emergencies in recent years, the Chinese government attaches significant importance to the emergency rescue system and continuously strengthens the construction of the emergency material reserve system.In 2022, the Chinese government issued the 14th Five-Year Plan for the Emergency Management System, which points out that it is necessary to improve the system of the emergency material reserve, optimize the layout of the material bases, and build comprehensive emergency material bases.In the future, the ''county-city-provincial'' three-level emergency material reserve network will be perfected.At the same time, it is becoming a trend that can change the existing material reserve mode in the future and implement the various types of emergency materials to hierarchical responsibility, hierarchical reserve [8].The release of these policies is important to improve the emergency reserve system.Coal mine emergency material reserve should be incorporated into the emergency material reserve system, as it can reduce pressure on coal mine enterprises in emergency material storage and maintenance while also facilitating integration of emergency rescue materials and equipment from industries, ultimately improving the utilization rate of materials and equipment [9].
Although the importance of material bases in emergency rescue has been confirmed, research on optimizing the location of coal mine emergency material bases is still limited.Therefore, this paper aims to determine the reasonable layout of coal mine emergency material bases, while studying the allocation and optimization of material reserves.Integrate the coal mine emergency material bases as a comprehensive material base into the national emergency reserve system.On the one hand, emergency materials can be utilized not only for coal mine rescue, but also for other rescue activities.On the other hand, the government can coordinate the allocation of emergency materials, unify the maintenance, and guarantee the integrity rate of emergency materials.
Liaoning Province has a large number of coal mines, complex geological conditions, and serious accidents and disasters.In recent years, multiple major and particularly serious accidents have occurred [10].The construction of the coal mine emergency material base is focused on coal mining enterprises, with insufficient storage of rescue materials, outdated rescue equipment, and a lack of maintenance.At present, Liaoning Province has not yet established a comprehensive material reserve base [11].However, in order to strengthen emergency material support capabilities, the government plans to build a comprehensive material bases in the future.Therefore, choosing Liaoning Province as the research area is of great significance for building multi-level and comprehensive material bases, which helps emergency management departments better plan the distribution of material bases and improve material support capabilities.The research results can also serve as a reference for other regions in the construction of material bases, helping them better achieve the goal of improving emergency capabilities.

C. OBJECTIVE AND INNOVATION
This paper establishes an emergency material evaluation model from the perspective of coal mines, and conducts cluster analysis of emergency materials based on the evaluation values.Determine the reserve level of various emergency materials by analyzing the clustering results.On the basis of improving the traditional gravity model, a multi-objective planning site selection model for multi-level material bases in the ''county-city-province'' region was established, taking into account the different factors of material demand points such as risk value, accident rate, and material demand.Use the Non-dominated Sorting genetic algorithm III (NSGA-III) to solve the site selection model considering the hierarchical reserve mode of materials.Furthermore, the Fuzzy Analytic Hierarchy Process (FAHP) was utilized to screen the optimal layout plan.ArcGIS software was used for data processing and the site selection results was visualized.
This paper seeks to address the existing gap in the literature regarding the selection of coal mine material base locations.More specifically, a case study of Liaoning was conducted, taking into account the disparities among various coal mines.The layout of material bases was optimized, resulting in detailed configuration for each base.To the best of the author's knowledge, there is no research on the above investigation in Liaoning.Methodologically, for the first time, the combination of material hierarchical reserve mode, site selection model and FAHP were used to solve the problem.Cluster analysis is a classification method, which is used to divide the classification objects according to the different structures and characteristics.By analyzing the characteristics of each type of emergency materials, the appropriate reserve location can be determined.According to the classification results of materials, it is applied in the multi-objective site selection model, which can comprehensively measure the degree of association between multiple objectives and give a series of feasible location schemes.On the other hand, FAHP has received extensive recognition in dealing with the objectives of the analysis scheme and making the final scheme selection [12], [13], [14].FAHP is used to measure the objectives of each 36480 VOLUME 12, 2024 Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.
site selection scheme and screen the site selection scheme according to the professional knowledge and experience of decision-makers.Three methods are comprehensively adopted to consider the objective results and the subjective choice of decision-makers in the layout optimization of material bases.Overall, the proposed model is a more detailed multi-stage decision-making method, which can be used in the location of material base, and provide a reference for the location of coal mine emergency material bases.

D. PAPER ORGANIZATION
The rest of this paper was organized as follows.In Section II, the literature review was presented.In Section III, the material hierarchical reserve mode of materials was described, a site selection model was established, and the process of FAHP was explained.In Section IV, The case study and solutions were demonstrated.Results and discussion were presented in section V. Finally, the conclusion and future works of this research were detailed in section VI.

II. LITERATURE REVIEW
The location selection of emergency material bases has been the focus point of research by scholars in recent years.Reasonable selection of the location of the emergency material bases can reduce casualties and economic losses, which is of great significance to the sustainable development of society.A number of researchers have chosen different objectives to model the optimization of emergency material bases [15], [16].Among them, Cheng considered factors such as the cost and capacity of material bases and established a model for siting urban emergency material bases [17].Linhua and Jin proposed the concept of emergency efficiency satisfaction and established a robust optimization model for siting emergency material bases [18].Wang and Xu used a safety function to describe the distance-perceived cost, constructed emergency bases siting model based on reliability and safety-perceived level, and also solved the model using an improved simulated annealing [19].These studies have been validated and proved the feasibility of the model.However, these studies mainly focus on cost, transportation distance, and reliability as the optimization objectives and did not consider other factors that may affect the results of site selection [20].The siting of emergency material bases is a complex process.In order to optimize the layout reasonably, it is necessary to fully consider various influencing factors.With the deepening of research, some scholars have considered the factors affecting the siting results and studied the siting model of the emergency material bases.Yu considered the randomness of disaster time occurrence and the uncertainty of material demand, and developed a two-stage model for site selection and storage [21].Alam et al. identified and evaluated the location of emergency bases considering factors such as flood risk and traffic evacuation [22].Xiang et al. assessed the possibility of road network damage under disaster conditions and calculated the road damage rate.With the objectives of minimizing the construction area of emergency bases and minimizing the total evacuation distance, the emergency material bases siting model was constructed [23].Wenjing and Wuyangconsidered the typhoon path information and distribution characteristics, and constructed a siting model for emergency material bases in a typhoon environment.The model was solved using GA [24].Ailing et al. applied fuzzy logic to historical disaster data at the material demand point and constructed an emergency material bases siting model which aimed at ensuring timely transportation of materials under disaster circumstances [25].In the above study, researchers mainly focused on single-level material bases and optimizing the location of material bases.Due to economic considerations, the number of material bases has been limited.However, due to the limitations of the scale of the material bases, the quantity of stored materials and equipment is difficult to meet the rescue needs.
Some researchers have recognized the benefits of multilevel emergency material bases siting model, which included a simple management process, high management efficiency, clear responsibility, and professionalism [26].This model is widely applicable to emergency and public facility siting [27], [28], [29].How to determine the location of material bases at all levels to improve rescue efficiency while reducing construction costs has been the subject of research [30].Ma et al. introduced the concept of ''supplementary siting'' and established a multilevel emergency material bases siting-distribution model.This model's validity was verified through empirical analysis [31].Shavarani constructed multilevel facility siting for determining the number and layout of emergency facilities.It was found that multilevel emergency facilities could improve the efficiency and response capacity of emergency response [32].Bo and Xiaomin proposed the concept of joint storage of materials by government and enterprises and established a siting model of emergency material bases with multilevel coverage [33].Dezhi et al. constructed a dual-objective planning siting model for multilevel emergency material bases by combining elements of multiple coverage.Through empirical analysis, it was discovered that this model could enhance supply efficiency and stability of materials [27].Shi et al. constructed a multilevel emergency material bases siting model by considering the capacity constraints of the emergency material bases, which contributed to a reduction in construction cost [34].These studies indicated that the multi-level material bases model not only reduced the construction cost but also enhanced rescue efficiency compared to traditional model [35], [36].The importance of multi-level material bases has been widely recognized in the above research.However, compared to single-level material bases, multi-level bases can improve rescue efficiency, which lead to storing more emergency materials, thereby increasing construction costs.In addition, excessive emergency materials will increase the difficulty of maintaining the material bases.In summary, most of the existing research results have focused on the analysis of location selection methods for material bases, while there is relatively little research that combines the actual situation and needs of the region.In addition, under the constraints of construction costs, the layout of material bases also has certain limitations.In view of this, this study adopts a hierarchical reserve mode for materials in the multi-level material bases site selection model, ensuring a reasonable layout of the material bases while further reducing construction costs.Applying the hierarchical reserve mode of materials in multi-level material bases can not only meet rescue needs, but also reduce the storage of materials and equipment.On the one hand, minimize the storage of materials and equipment to the greatest extent possible.On the other hand, a reasonable quantity of materials and equipment can avoid excessive maintenance pressure, thereby improving the integrity rate of materials and equipment, and ensuring the smooth progress of emergency rescue work.From a macro perspective, the layout of the material bases and the reasonable quantity of material equipment can meet the requirements of efficient, low-cost emergency rescue and economic operation of the material bases.To achieve the goal of timely rescue and disaster reduction in the face of major accidents, this is a very important research topic.

III. MATERIALS AND METHODS
Emergency materials refer to the necessary material support during the entire process of responding to sudden public events such as severe natural disasters, accident disasters, public health events, and social security events [37].This includes not only medical materials, but also various equipment required for emergency rescue.At present, the material bases usually store all the materials required for accident rescue to ensure that each base can meet the rescue demand.Due to the variety of rescue materials, low short-term usage and high storage conditions and other characteristics [38].
The traditional material reserve mode will duplicate the configuration of rescue materials, increase the construction cost of the reserve system, and increase the difficulty of daily maintenance.Therefore, this paper adopts a hierarchical reserve mode for emergency materials, which takes the cost, frequency of use, and difficulty of maintenance of rescue materials into account.Emergency materials are stored in separate emergency material bases at all levels.The material reserve system can reduce the reserves of materials while meeting the rescue needs within the region.At the same time, the hierarchical reserve mode of materials can form resource complementarity among multi-level bases forming a complete material reserve system.
Firstly, this study establishes an evaluation index system for emergency materials and evaluate materials using the entropy weight method.We perform cluster analysis on the evaluation values of emergency materials.Based on the clustering results, we summarize the storage principles for various emergency materials.Then, improve the traditional gravity model and establish a multi-objective planning site selection model.Use the NSGA-III to solve the model and obtain candidate schemes.Finally, the FAHP method is used to evaluate candidate schemes and obtain the optimal site selection scheme.The framework of this study is shown in the Figure 1.

A. MATERIAL HIERARCHICAL RESERVE MODE
The emergency materials stored in the material bases are used in rescue work to quickly eliminate the impact of accidents, rescue disaster victims, and reduce losses.The goal of storing emergency materials is to transport them to the coal mine where the accident occurred in the shortest possible time, so that rescue work can be carried out as soon as possible.There are differences in the functions, uses, and stages of use of different types of emergency materials in accident rescue, so their importance also varies [39].In addition, the difficulty of obtaining different emergency materials and equipment is crucial for emergency rescue work.If the difficulty of obtaining emergency materials is high, then they cannot be obtained in a timely manner.This will increase the rescue time.Therefore, such materials and equipment should be stored in advance.When storing emergency materials, the capacity of each level of material bases and the number of professional personnel may also vary [40].A suitable storage environment is necessary for materials and equipment.In view of this, this study adopted three indicators that affect the storage location of materials and equipment: Importance (Z1), Scarcity (Z2), and Storage (Z3).
(1) The importance of emergency materials refers to the degree to which they are essential for emergency rescue work in coal mine accidents, including four sub-indicators: Urgency of demand (Z11), Rescue effectiveness (Z12), Lack of cost (Z13), and Amount used (Z14).''Urgency of demand (Z11)'' refers to the urgency of a certain emergency material needing to participate in accident rescue, and also measures the importance of timely and accurate transportation of a certain emergency material to the accident mine after the accident for the effectiveness of emergency rescue.''Rescue effectiveness (Z12)'' refers to the rescue effect of a certain type of emergency material on accidents, the better the rescue effect, the higher the evaluation value.''Lack of cost (Z13)'' refers to the additional losses caused by the lack of certain emergency materials.The greater the loss, the higher the evaluation value of the shortage cost of such emergency materials.''Amount used (Z14)'' refers to the amount of emergency materials used in rescue works, the less they are used, the more important they are in rescue work.
(2) The scarcity of emergency materials refers to the importance of the characteristics of emergency materials to the accident mine, including four sub-indicators: Irreplaceability (Z21), Uncertainty (Z22), Number of providers (Z23), and Hysteresis (Z24).''Irreplaceability (Z21)'' refers to the degree to which a certain emergency material is irreplaceable for a certain accident.If there are many substitutes for this type of material equipment in accident rescue, the lower the irreplaceable evaluation value of this material.''Uncertainty (Z22)'' refers to the inability to determine the quantity of materials needed after an accident occurs.''Number of providers (Z23)'' refers to the number of providers who can provide certain emergency materials.The number of providers represents the difficulty of obtaining emergency materials.''Hysteresis (Z24)'' A refers to the use stage of emergency materials in rescue work.
(3) The storage refers to the difficulty of storing emergency materials, including four sub-indicators: Capacity (Z31), Usage frequency (Z32), Maintenance difficulty (Z33), Storage conditions (Z34).''Capacity (Z31)'' refers to the requirement for storage space for a certain type of emergency materials due to volume or other reasons.''Usage frequency (Z32)'' refers to the times of emergency materials used in the rescue process of coal mine accidents.If a certain emergency material needs to be used in accidents of different severity, the frequency of use of that type of emergency material is relatively high.''Maintenance difficulty(Z33)'' refers to the difficulty of maintaining a certain emergency material in order to ensure its normal use.''Storage conditions(Z34)'' refer to the requirements for the storage environment of emergency materials.
In summary, this study selected 3 indicators and 12 subindicators to establish emergency material evaluation model, as shown in Table 1.
From Table 1, indicators can be divided into two categories: qualitative and quantitative.Before conducting an evaluation, quantify qualitative indicators using expert scoring.For quantitative indicators, their specific values are determined based on real data.Due to the different dimensions of quantitative indicators, it is necessary to normalize them.
The process of emergency material evaluation and clustering is as follows: Step 1: Establish an evaluation dataset.Invite experts in the field of emergency rescue to evaluate various emergency materials and construct an indicator dataset X a = {x a1 , x a2 , . . ., x aB }, a = 1, 2, . . ., A. Where A represents the number of datasets and B represents the total number of indicators.
Step 2: Calculate the objective weights of the indicators.Use the entropy weight method to calculate the objective weight of indicators.In order to eliminate the differences in various indicators, it is necessary to standardize the data.
Standardize the positive and negative indicators according to (1) and ( 2) respectively.The y ab represents sub-indicator b in the standardized dataset a.
Step 3: Calculate the subjective weights of the subindicator.Calculate the subjective weights of sub-indicator using the order relationship analysis method (G1 method).The order relationship analysis method is an improvement on the basis of the analytic hierarchy process (AHP), which does not require the construction of a judgment matrix and does not require consistency checks, and has strong operability.
Sort based on the importance of the sub-indicators.
Invite experts to determine the relative importance level r λ of adjacent sub-indicators x λ−1 and x λ , and the judgment criteria are shown in Table 2.
Calculate the subjective weight w sub b of sub-indicators based on their relative importance r λ . Step Optimize ( 9) by using the Lagrange multiplier method to obtain the calculation formula for the comprehensive weight.
Step 5: Calculate the evaluation value of emergency materials.Calculate the evaluation values of the three types of emergency materials using the comprehensive weights of the sub-indicators.
Step 6: Cluster analysis of emergency materials.Cluster analysis was conducted using SPSS software, using average evaluation values to distinguish three levels of emergency materials, and the characteristics of each level of emergency materials were analyzed.

B. MATERIAL BASE LOCATION MODEL 1) GRAVITY MODEL
The gravity model is one of the measurement models that utilize Newton's law of gravity to measure spatial interactions.It has been widely used in different scientific fields [41], [42], [43].With the deepening of the research, the theoretical basis of the gravity model has become more mature, and it has gradually become an important measurement method to study the interrelationship between two things in the socio-economy.
It often requires comprehensively considering the mutual attraction relationship between the emergency rescue force and the demand point in accident rescue, i.e., the more adequate materials equipped in the material bases and the shorter transportation distance, the better its rescue effect on the mining accident.Therefore, the gravity model is introduced into the coal mine emergency material bases siting model, and this model is improved by comprehensively considering the material matching and spatial distance between the material base and the coal mine.The formulas for this improvement are as follows: where T ij represents the gravity between the material base j and material demand point i. d ij is the distance between the coal mine i and the material base j.Z jl is the amount of category l accident disposal material stored in the material base j.Z jg is the amount of category g accident protection material stored in the base j.Q il is the probability of a category l accident occurring at the coal mine i.

2) MODEL PARAMETER
The parameters and decision variables involved in the model are described below: The parameters and decision variables involved in the model are described below: 1) Parameters i = {1, 2, . . ., I } is set of demand points.C 1j is the construction cost of the candidate base.C 2j is the expansion cost of the material base, i.e., the construction cost of the base due to the additional reserve of materials.
S j is expansion area of the material base j.q, t, h is the cost of purchasing, storing, and transporting the emergency materials.
Z jk is the amount of category k emergency material stored in the material base j.
M ijk is the amount of category k emergency material transported from the material base j to the demand point i.
W ik is the amount of category k material demanded by the demand point i.
d ij is the distance between the demand point i and the material base j.d jj ′ is the distance between the adjacent material bases.R initial is the rescue distance from the material base.Q il is the probability of a category l accident in demand point i.
U is the material stacking factor.

2) Decision Variables
A j is a 0-1 decision variable, i.e., when emergency the material base j is selected, it is 1, otherwise it is 0.
B ij is a 0-1 decision variable, i.e., when the coal mine i is within the coverage area of the base j, it is 1, otherwise it is 0.
Y ijk is a 0-1 decision variable, i.e., when the material base j is transporting categories k materiel to the coal mine i, it is 1, otherwise it is 0.

3) MODEL BUILDING
The purpose of building an emergency material base is to improve the efficiency of emergency rescue.After an accident occurs, it is necessary to transport emergency materials to the accident site in the shortest possible time in order to carry out rescue work in the first time around.In addition, due to the universality of emergency materials, scalability needs to be considered when selecting the location of the material base.Therefore, the objective function of the model includes: the total cost of the material base, the gravitational value between the material base and the coal mine, the transportation and transfer distance of material, and the distance between the adjacent bases.
(1) The total cost of the material base In order to make the model more realistic.We consider that the total cost of the material base includes five parts, which are fixed construction cost, expansion cost, transportation cost, transshipment cost and emergency materials purchase cost.
When base is selected for construction, there would be a construction cost, which is a one-time fixed cost related to the location and may vary depending on the location.The calculation formula is as follows: After the completion of the material base, it is necessary to determine the reserve of materials in the base based on the different demand for coal mines in the rescue area of the base.Different reserves will result in different expansion costs.The calculation formula is as follows: The transportation of emergency materials from the base to the demand point will incur certain costs, which mainly depend on the quantity and distance of the transported materials.The calculation formula is as follows: During the transportation of materials, there may be situations of material transfer due to factors such as transportation capacity and road conditions.Therefore, this cost also needs to be considered in the model.The calculation formula is as follows: The purchase cost of emergency materials is not only limited to the purchase cost, but also needs to consider the maintenance cost.The calculation formula is as follows: (2) The gravitational value between the material base and the coal mine For demand points.If there is sufficient material in the base and the distance to the demand point is close, the gravity value is greater.
(3) Transportation and transfer distance of material In the process of site selection, it is necessary to consider the risk value and distance factors of the coal mine.The higher the risk value of a coal mine, the closer it should be to the material base.Therefore, risk weighted distance needs to be considered in the model.The calculation formula is as follows: (4) Distance between the adjacent bases In order to improve the rescue area of the base and the expansion of the base, the need between the adjacent bases should be maximized as much as possible.The formula is as follows: Based on the above analysis, the structure of the multi-level emergency material base site selection model for coal mines established in this paper.It is as follows: min A j 1 = P (27) A j = {0, 1}, ∀j ∈ J (28) The objective function F 1 represents the minimum total cost of the material base; the objective function F 2 represents the maximum gravity value between the material base and the coal mine; the objective function F 3 represents the shortest transportation distance for emergency materials; The objective function F 4 represents the maximum distance between adjacent material bases.Formula ( 21) ensures that at least one material base provides rescue to the coal mine at each level; formula (22) ensures that coal mines are only rescued by selected material bases; formula (22) ensures that the emergency materials stored in the base can meet the needs of the coal mine; formula (24) indicates that the emergency materials transported from the material base cannot exceed the storage quantity in the base; formulas ( 25) and ( 26) require the number of city-level and provincial-level bases to be N and P, respectively.Formulas ( 27), (28), and ( 29) are decision variables.

C. NSGA-III ALGORITHM
This model is a non-deterministic polynomial (NP) problem, which refers to a problem with non-deterministic polynomial complexity.Genetic algorithms are a type of search algorithms constructed by mathematical simulation based on Darwin's proposed laws of evolution (survival of the fittest genetic mechanism) in the biological world, and are very effective for solving the NP problem in this type of optimization combination.
Our goal is to choose a low cost (F 1 ), high gravity (F 2 ), minimal risk weighted distance (F 3 ), and high adjacent distance (F 4 ) material base layout plan.However, these four goals are contradictory to each other.Therefore, it is necessary to find a solution that can balance the four objectives, rather than a single optimal solution.NSGA-III is a highly advanced algorithm that can handle large-scale multi-objective optimization problems.NSGA-III is widely used in multi-objective solving methods.This algorithm can determine the optimal trade-off between multiple objectives, find the solution set that best balances multiple objectives, and help decision-makers make choices when considering multiple objectives.Therefore, this study used NSGA-III to solve the model.The NSGA-III algorithm firstly randomly generates an initial population, and processes the initial population in the solution space through operations such as crossover and mutation to obtain the corresponding offspring population.Secondly, it combines the parent and offspring populations together.Thirdly, the crowding distance comparison operator and non-dominated sorting method are used to evaluate individuals.An elite strategy is adopted to filter the population, thereby restoring the population size and obtaining a new generation of population.Finally, compare the NSGA-III algorithm with the termination condition to determine whether to stop the operation.

D. METHOD OF DECISION MAKING
A set of optimal Pareto solutions can be obtained after using NSGA-III, but the solutions in the solution set cannot make the four objective functions optimal at the same time.Therefore, after obtaining the Pareto solution set, a set of relatively optimal solutions should be selected according to the needs of decision-makers [44].In this paper, FAHP is used to select the solution that best meets the decision-makers preference.The specific process is as follows: Step 1: Construct a judgment matrix ς * based on the model's objective function.
Step 2: The fuzzy logarithmic least squares method is used to obtain the target weight vector W .
Step 3: For each solution x * s , s = 1, 2, . . ., S in the Pareto solution set X * , the corresponding target vector where f max s and f min s are the maximum and minimum values of the corresponding objective vectors in each dimension for all solutions in X * , respectively.
Step 4: Subsequently, the satisfaction ξ s corresponding to each solution can be obtained by weighted summation of each normalized objective value F s .
Step 5: Finally, the solution corresponding to the maximum satisfaction value is the most satisfactory solution x best .
The whole decision-making process starts with obtaining the decision-makers preference for each objective, i.e., the weight between each objective, using 0∼1 to describe the importance of pairwise relationships between objectives [45].
Through expert interviews, we obtained the judgment matrices as: In order to avoid situations where the consistency of indicators is difficult to distinguish, the fuzzy logarithmic least squares method is used to obtain the weight vector:

IV. CASE STUDY
In this section, we evaluated and analyzed various clustering of materials.By analyzing the clustering results, the types of emergency materials that should be reserved in countylevel, city-level, and provincial-level material bases could be determined.We applied the established site selection model to the optimization of the site selection of emergency material bases in coal mines in Liaoning Province.

A. EMERGENCY MATERIAL EVALUATION AND CLUSTER ANALYSIS
We selected 13 commonly used emergency materials as the evaluation objects for emergency materials evaluation and cluster analysis, as shown in Table 3.After obtaining a dataset of 13 emergency materials through expert interviews and surveys, the weights of the sub-indicators were calculated using entropy weight method and ordinal relationship analysis method, as shown in Table 4.
According to the weight of the sub-indicators, the evaluation values of 13 emergency materials were obtained.Cluster analysis of the evaluation values of emergency materials using SPSS software.When analyzing, use emergency materials Z1, Z2 and Z3 indicators.The current classification of emergency materials is to determine the reserve location of various types of emergency materials.Therefore, it is divided into three clustering levels.The evaluation values and clustering results of 13 emergency materials are shown in Table 5.
In order to compare and analyze the characteristics of emergency materials at all levels, the average comprehensive evaluation value was used to distinguish three levels of emergency materials and determine their characteristics.The comparison of evaluation values for various emergency materials is shown in Figure 2.
It can be seen from Table 5 and Figure 2. The first level emergency materials are ''all terrain crane'' and ''large caliper drilling rig'', with evaluation values of all three indicators above 0.8.The first level emergency materials are mainly large rescue equipment, which have the characteristics of high cost, large volume, difficult maintenance, and low frequency of use.The excessive quantity of emergency materials at this level will result in significant economic and maintenance pressure.Therefore, for the first level emergency materials, they can be stored in the provinciallevel material base, with dedicated personnel responsible  for regular maintenance.By optimizing the location of the provincial-level bases, the first level emergency materials can be transported quickly, ensuring the timeliness of emergency rescue.The second level emergency materials are ''hydraulic tunnel drilling machine'' and ''large drainage equipment'', with evaluation values slightly lower than the first level emergency materials, mainly specialized emergency materials.Compared with the first level emergency materials, the volume, cost, and maintenance difficulty of the second level emergency materials are slightly lower, and the frequency of use is slightly higher.This category of emergency materials can be stored in city-level material bases.The evaluation values of various indicators for the third level of emergency materials are the lowest.These emergency materials are mostly small rescue equipment and rescue auxiliary materials, which have the characteristics of small size and large reserves.This type of emergency materials can be reserved in county-level bases.
Through analysis, the reserve principles for various emergency materials are determined as follows: (1) large equipment is stored in provincial-level material bases; (2) medium sized equipment is stored in city-level material bases; (3) Small equipment and other materials are stored in county-level material bases.

B. MODEL DATA
According to the documents issued by the Development and Reform Commission of Liaoning Province, there are 30 existing coal mines in Liaoning.We used ArcGIS software to randomly generate the locations of candidate bases, and after preliminary screening, we determined that the number of the candidate bases at the county, city, and provincial levels were 40, 50, and 100, respectively.The distribution of candidate material bases at all levels is shown in Figure 3.
We obtained geographic information, risk value, and accident data of 30 coal mines from Liaoning Mining Safety Supervision Bureau in previous years, as show in Table 6.Since there are no documented historical records available for the demand for coal mine emergency materials and the use of specific materials during the accident rescue process, we referred to literature [46] and used the Python random generation program to simulate the demand for coal mine emergency materials.Due to the variety of coal mine rescue materials and equipment, and the parameters of various models of equipment are different, in order to facilitate the calculation, this paper uses the volume as the specific quantity and unifies the price, as shown in Table 7.
Based on the previous analysis, the storage locations for various emergency materials have been determined.(1) Large equipment is stored in provincial-level material bases; (2) Medium sized equipment is stored in municipal material bases; (3) Small equipment, medical, subsidiary, and accident disposal materials are stored in county-level material bases.
In addition, we obtained the road network information of Liaoning from the OSM (OpenStreetMap) website, and imported it into the ArcGIS software for the road network analysis, and finally obtained the road network data between the coal mine and the candidate bases.Coordinate System Selection GCS_WGS_1984.
According to the Construction Standards for Disaster Relief Material Reserve Base, in addition to the emergency material bases also needs to be equipped with assistance buildings and management rooms.Therefore, the areas of these buildings are used as fixed construction area for the material bases, and the construction area of the bases are regarded as the expansion area of the material bases.According to the requirements of the Planning and Design Parameter of General Warehouse and Warehouse Area, we set the material stacking height to 9 m, and we multiplied the stacking height by 0.7 to determine the expansion area of the base.The calculation formula is as follows: According to the Notice of the Ministry of Land and Resources on Adjustment of Land Levels in Certain Areas and the National Minimum Land Offering Standardissued by the Ministry of Land and Resources, the construction cost of all different level candidate bases is determined by taking the location of the candidate bases into account.According to the development plan of Liaoning Province, in the future, there will be 2 provincial-level comprehensive material bases, 14 city-level comprehensive material bases and 100 countylevel comprehensive material bases.Considering that the 30 coal mines are distributed in 6 cities, we selected 2 provincial-level material bases and 6 city-level material bases.The number of county-level material bases is not limited, but mainly to meet the needs of coal mine rescue.
The rescue radius of candidate material bases at all levels is determined based on the rescue distance of mine rescue teams of different levels.The model parameters are set as shown in Table 8.

A. RESULT OF SITE SELECTION
Use NSGA-III to solve the model and obtain multiple feasible location options.Subsequently, FAHP screening was used to select the optima site selection.The visualization result of ArcGIS was shown in Figure 4.
The best solution was to select 17 emergency material bases.The total construction cost is 431.003M CNY.Among them, the fixed cost is 4.319 M CNY.The cost of material procurement and maintenance is 422.486M CNY, the expansion cost is 0.908 M CNY, the transportation cost is 2.975 M CNY, and the transshipment cost of materials between different levels of bases is 0.315 M CNY.The specific results are shown in Table 9.
In the final site selection plan, the numbers of the selected provincial-level bases are respectively 27 and 37, and are located in Shenyang and Jinzhou.The numbers of the selected city-level bases are 2, 17, 21, 24, 33, and 45, respectively.The city-level bases are located in Changtu Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.The rescue areas of various levels of material bases are shown in Figure 5. Due to the different extreme rescue distances of various levels of material bases, there are significant differences in the rescue areas of each level of material bases.According to Figure 5(a) and Table 10, the rescue areas of the county-level bases are 2889.4km 2 , 3258.6 km 2 , 2377.1 km 2 , 2394.7 km 2 , 533.08 km 2 , 2305 km 2 , 1328.1 km 2 , 1742.5 km 2 , and 2017.2 km 2 , respectively.Excluding overlapping areas, the entire countylevel material bases can cover 16172.7 km 2 .Among them, the rescue area of base No.42 is the smallest, at 533.1 km 2 ; The rescue area of base No.7 is the largest, at 3258.6 km 2 .According to Figure 5(b), the rescue areas of the citylevel bases are 13344.5km 2 , 12724.4 km 2 , 15825.6 km 2 , 15769 km 2 , 15242.8 km 2 , and 7063.5 km 2 , respectively.Excluding overlapping areas, the entire city-level material bases can cover 57931.2km 2 .Among them, the rescue area of base No.45 is the smallest, at 7063.5 km 2 ; The rescue area of base No.21 is the largest, at 15825.6 km 2 .According to Figure 5(c), the rescue areas of the provincial-level bases are 50475.9km 2 and 57383.1 km 2 , respectively.Excluding overlapping areas, the rescue area of the entire provinciallevel bases is 89820.5 km 2 .The rescue area of the No.42 county-level and No.45 city-level bases is smaller than that of other bases at the same level.Due to significant differences in road network density among different regions in Liaoning province, the road network density is higher in the central region and coastal cities, while the road network density is the smallest in the eastern region.Overall, material bases are often located in the suburbs of cities.On the one hand, the location of coal mine demand points is far from the urban area.Setting up material bases in the suburbs can shorten the transportation distance of emergency materials.On the other hand, the lower land transfer prices in suburban areas can reduce the construction costs of material bases.In addition, material bases located in the suburbs of cities can use road network to cover the urban area, while also being able to rescue remote areas far from the city.Although building material bases in urban areas can increase rescue areas through developed road network, there is a possibility of traffic congestion, which is not conducive to the rapid transportation of emergency materials and thus increases emergency response time.

B. SENSITIVITY ANALYSIS 1) SENSITIVITY ANALYSIS OF MATERIAL BASES
Set the rescue limit distance (R initial ) of the material base to 40 km to calculate the multiple coverage and high-risk coverage rate of the coal mine, as shown in Figure 7. Multiple coverage refers to the proportion of coal mines covered multiple times by the base, while high-risk coverage refers to the proportion of coal mines with risk values of 3 and 4 being covered multiple times.Both of these indicators and the number of county-level base constructed show a trend of first increasing and then slowing down, that is, as the number of bases increases, the increase in both indicators requires more resources to be consumed.
In addition, it can be seen from Figure 6 that when the number of county-level bases is 9, a turning point in the growth rate of gravity value occurs, and the growth rate of gravity value at this time is about 56%.The inflection point of the multiple coverage curve in Figure 7 appears around 14, at which point the multiple coverage rate is around 70%.The turning point of the high-risk coverage curve appears at 13, at which point the high-risk coverage rate is 68.2%.This result indicates that when providing rescue services for coal mines based on coverage indicators, more bases need to be built and more emergency materials need to be stored, while providing rescue services based on gravity value indicators, various resources are consumed less.This is because in the gravity value indicator, only the rescue distance and the degree of material matching need to be considered.

2) SENSITIVITY ANALYSIS OF RESCUE DISTANCE
Set the rescue distances (R initial ) of county-level material bases to 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, and 90 km, respectively.Keep the other parameters of the model unchanged and obtain the relationship between rescue distance and the costs of county-level material bases.As shown in Figures 8 and 9.The cost of the base consists of fixed costs, expansion costs, transportation costs, and material purchase and maintenance costs.From Figure 9, it can be seen that in the construction of material base, the cost of purchasing and maintaining materials accounts for about 90% of the total cost.From Figure 8, it can be seen that as the rescue distance of county-level material bases increases, the number of bases constructed also decreases.When distance is 35 km, the optimal plan is to build 10 bases; When distance is 40 km, the optimal plan is to build 9 bases.When distance is 45, the optimal plan is to build 8 bases; When distance is 50 km, the optimal plan is to select the location and build 7 bases; When distance is 70 km.Based on the construction quantity of the reserve, optimal solution is to build 6 bases.it can be found that the fixed construction cost and expansion cost of the base will decrease with the increase of rescue distance.The largest decrease in costs between 35-50 km and 65-70 km is due to a decrease in the number of bases.The cost of material transportation will increase with the increase of rescue distance.The significant increase in transportation costs between 35-50 km and 65-70 km is due to the decrease in the number of bases, which not only increases transportation distance but also increases the transportation route between the material bases and the coal mine.

3) THE IMPACT OF MATERIAL HIERARCHICAL RESERVE MODE ON SITE SELECTION RESULTS
In order to analyze the impact of the material hierarchical reserve mode on the results, the location results under the traditional material reserve mode were solved under the same model parameters.The comparison results are shown in Table 11.
Under the traditional material reserve mode, each base will reserve all types of materials required for the material demand points within its rescue area.The advantage of this storage mode is that a single material base can meet the needs of coal mine rescue.However, there are two shortcomings of this model.Firstly, the construction cost of material bases is usually one of the goals for selecting material bases.In order to improve the economy of the entire reserve system, the number of bases will be reduced.As a result, a base need to rescue more coal mine demand points, which will increase the reserve of materials in the base.In other words, the reserve of materials will also increase accordingly.Secondly, during the rescue process, it is necessary to transport a large amount and multiple types of rescue materials from a single base, which will increase the busyness of the base and reduce its transportation efficiency.
According to the Table 11 the emergency material bases reserves 3746 m 3 , 10895 m 3 , 16675 m 3 of disposal materials and 9582 m 3 , 27936 m 3 , 42615 m 3 of protection materials, respectively.While under the hierarchical reserve mode, the provincial and city-level bases only need to reserve 6000 m 3 , 8634 m 3 of protection materials, and the countylevel bases store 15382 m 3 of disposal materials and 2422 m 3 of protection materials, which reduce 15934 m 3 of disposal materials and 63077 m 3 of protection materials.Therefore, the total cost can be reduced by 86.6%.When selecting the site for a single level material base, the optimal program is the site selection program for county-level material bases under the traditional material reserve mode.Compared with county-level material bases, our program was able to reduce the cost by 74.7%.The location of multilevel material bases may result in an increase in the number of bases compared to single-level material bases, but this can also help in minimizing the duplication of emergency materials and subsequently reduce the costs associated with them.The results show that considering the hierarchical reserve of emergency materials and rationally planning the storage location of different levels of material can reduce the cost of the reserve system.

C. RESULT ADAPTABILITY ANALYSIS
According to the final result, the two provincial-level bases are located in Jinzhou City and Shenyang City respectively, which is consistent with the planning of the Liaoning Provincial Government.The shortest rescue time between 30 coal mines and various levels of material bases is shown in the Table 12.
After an accident, the county-level material bases can cover 100% of the coal mines within 0.5 half an hour.Within 0.4 hours, it can cover 73% of the coal mines.The longest transportation time for city-level bases is 1.2 hours.It can cover 83.3% of coal mines within 1 hour.It can cover 66.6% of coal mines in 0.5 hours.The longest transportation time for provincial-level bases is 1.7 hours, and the shortest transportation time is 0.4 hours.Among them, 93.3% of coal mines can be covered within 1.5 hours, and 43.4% of coal mines can be covered within 1 hour.If an accident occurs in no.1 Coal Mine, after receiving a rescue notice, the No.42 county-level material base can transport rescue materials to the coal mine within 0.1 hours.The No.25 citylevel material base can transport materials and equipment within 0.3 hours.The No.27 provincial material base will transport the materials to the coal mine within 1 hour.Overall, the location of the material bases meets the requirements of government planning and shortens rescue time.Therefore, the model proposed in this study can effectively optimize the site selection of material bases in the region, reduce the construction cost of material bases, and improve emergency response time.This plays an important role in improving the overall emergency level of the region.

VI. CONCLUSION AND FUTURE WORKS
In this paper, for the problem of hierarchical reserve of emergency materials and the location and layout of material bases at all levels, a site selection model for emergency materials bases based on hierarchical reserve of materials was established.
First, by evaluating and clustering various emergency materials, and based on the clustering results, analyzed the characteristics of emergency materials at all levels.The principle of hierarchical reserve of emergency materials been determined.The traditional gravity model was improved by using the quantity of material reserve and the accidents rate which measure the gravitational force between emergency material bases and demand points.After comprehensively considering the rescue characteristics and rescue needs of all levels of bases, a multi-level emergency material base location model was constructed.Subsequently, the data required for the model were obtained from the official, and the real road network distance was obtained using ArcGIS software.Finally, the NSGA-III was used to solve the model, Finally, and the FAHP method was used to select the optimal layout from multiple schemes, and the scheme was visualized.
This study takes Liaoning Province as an example to verify the effectiveness of the model.The results show that within Liaoning Province, 9 county-level material bases, 6 city-level material bases, and 2 provincial-level material bases have been constructed, with a total cost of 431.003 M CNY.A total of 32580m 3 of emergency materials were stored in the material bases, including 15382m 3 of disposal materials and 17198m 3 of accident protection materials.Analyzing the construction cost of the material bases, it can be found that the purchase and maintenance cost of emergency materials accounts for the highest proportion.After adopting the hierarchical reserve mode of materials, it can significantly reduce the repeated configuration of materials, thereby reducing costs.Compared with traditional material reserve methods, it can reduce construction costs by 86.6%.Although more bases will be built, compared with single-level emergency material bases, reducing the quantity of materials through hierarchical storage can still reduce costs by 74.7%.This means that the hierarchical reserve mode of materials and the multi-level material base location selection model have better applicability.In addition, this study also indicates that in addition to land transfer costs, road network density can also affect the selection of material bases.For emergency management departments, the selection of multi-level material bases remains a major issue, and a reasonable selection of material base locations is conducive to improving the level of emergency rescue.In addition, this study enriches the literature on the location selection of multi-level material bases.This also provides a basis for the application of the hierarchical storage mode of materials in multi-level material bases.This model can not only be used in the site selection planning of emergency material bases in Liaoning Province, but also in the site selection of other demand points besides coal mines, and it can be applied to any region.
Although this study has made some contributions to related fields, there are still some limitations.If emergency rescue materials from other industries are taken into account, the results will be more comprehensive and accurate.However, due to the difficulties of data collection, it was not considered in this study.This study focuses on the site selection of coal mine emergency material bases, aiming to provide reference for emergency management departments to choose the construction location of material bases.In the future, we will further refine the classification and grading of emergency materials, and consider factors such as road network damage and congestion that affect transportation time.Meanwhile, incorporating relevant data on emergency materials from other industries into the model makes the multi-level material bases selection model more realistic.

4 :,
Calculate the comprehensive weight of the subindicators.Adopting the minimum relative information entropy principle, the comprehensive weight w com b is determined by objective w obj b and subjective weight w sub b .b = 1, 2, . . ., B

FIGURE 2 .
FIGURE 2. Comparison of evaluation indicators for emergency materials.

FIGURE 3 .
FIGURE 3. Distribution of demand points and candidate material bases.

FIGURE 4 .
FIGURE 4. Distribution of material bases at all levels.

FIGURE 5 .TABLE 10 .
FIGURE 5. Rescue areas of various levels emergency material bases.(a) County-level material bases rescue area.(b) City-level material bases rescue area.(c) Provincial-level material bases rescue area.

FIGURE 6 .
FIGURE 6.The relationship between the growth rate of various indicators and the number of material bases.

FIGURE 7 .
FIGURE 7. The relationship between the coverage rate of various indicators and the number of material bases.

FIGURE 8 .
FIGURE 8.The impact of different rescue distances on various costs.

FIGURE 9 .
FIGURE 9.The proportion of various costs under different rescue distances.

TABLE 1 .
Emergency material evaluation form.

TABLE 3 .
List of emergency materials.

TABLE 4 .
Weights of each sub-indicator.

TABLE 5 .
Evaluation values and clustering results of emergency materials.

TABLE 6 .
Risk value and accident's rate in coal mines.

TABLE 7 .
Demands for coal mining materials.

TABLE 9 .
Calculation results of site selection plan.

TABLE 11 .
Comparison of results of different material reserve modes.

TABLE 12 .
The shortest transportation time and material bases number for each coal mine.