Multi-coal seam mining optimization control method based on crossheading position stability

Abstract

The invention belongs to the technical field of coal bed gas mining, and particularly discloses a multi-coal-bed mining optimization control method based on crossheading position stability, and the method comprises the steps: comprehensively obtaining parameters such as coal bed spacing, thickness, inclination angle, roof and floor lithology and fault distribution, and generating an accurate three-dimensional geologic model through three-dimensional modeling software; and the condition of incomplete geological analysis of a traditional method is changed. And on the aspect of crossheading arrangement, empirical operation is replaced with finite element analysis, the anchor rod distance and the support density are accurately determined, and safety and economical efficiency are improved. The mining sequence planning is changed from dependence on a fixed rule to establishment of an identification model combined with a genetic algorithm, and dynamic optimization is realized. The mining height, the mining width and the advancing speed are flexibly regulated and controlled according to the mapping relation table in the mining process, and real-time dynamic adjustment is achieved. Meanwhile, the early warning mechanism calculates scores based on crossheading stress data, is more accurate and sensitive, and has higher pertinence and effectiveness.

Description

Technical Field

[0001] The invention belongs to the technical field of coalbed methane mining, and in particular relates to a multi-coal seam mining optimization control method based on drift position stability. Background Art

[0002] In coal resource mining, the occurrence of multiple coal seams is very common. The significant differences in spacing, thickness, inclination, geological structure, and other conditions between different coal seams make multi-seam mining more complex than single-seam mining. For example, when the spacing between coal seams is small, mining the upper coal seam may affect the stability of the lower coal seam, increasing mining difficulty and safety risks.

[0003] Existing technologies such as a coal underground gasification injection and production linkage control system and method disclosed in the Chinese invention patent application with application number 202111523795.0 and existing technologies such as a method for composite co-mining of deep coalbed methane and coal using multi-mode microwave radiation disclosed in the Chinese invention patent application with application number 202210086344.3. For the above two technical solutions, the multi-coal seam mining optimization control is aimed at improving the efficiency and stability of underground coal gasification mining, reducing manual intervention, and improving overall mining benefits.

[0004] However, the existing technology still has the following deficiencies: 1. In coal seam mining, real-time monitoring and feedback control are crucial for timely detection and resolution of problems such as drift instability and mutual influence between coal seams. The existing technical solutions do not involve the optimization of drift position stability and cannot effectively resolve the problem of drift instability in multi-coal seam mining; 2. In multi-coal seam mining, the optimization of mining sequence and parameters is crucial for improving mining efficiency and safety. The existing technical solutions do not involve the optimization of mining sequence and parameters for multi-coal seams, which may lead to low mining efficiency and increased safety risks. Summary of the Invention

[0005] In response to the problems in the prior art, the present invention provides a multi-coal seam mining optimization control method based on the stability of the drift position, which solves the problem that the prior art does not involve the optimization of the drift position stability and cannot effectively solve the problem of drift instability in multi-coal seam mining. At the same time, it solves the problem that the prior art does not involve the optimization of the multi-coal seam mining sequence and parameters, which may lead to low mining efficiency and increased safety risks.

[0006] The technical solution adopted in the present invention is as follows: In order to solve the problems raised in the above background technology, a multi-coal seam mining optimization control method based on the stability of the drift position is proposed.

[0007] The present invention provides a multi-coal seam mining optimization control method based on the position stability of the drift, comprising the following steps: S1. Obtain data on the inter-coal spacing, thickness, dip, roof and floor lithology, and fault distribution of multiple coal seams, and generate a three-dimensional geological model using three-dimensional modeling software; S2. Based on the generated 3D geological model, adjust the drift layout to obtain the target drift layout plan and determine the corresponding anchor spacing and support density; S3. Establish a coal mining sequence recognition model based on the 3D geological model. The coal mining sequence recognition model combines a genetic algorithm to generate a mining sequence. During the mining process, the trench stress data and mining position are monitored in real time. Based on a pre-set mapping relationship table between trench stress and coal mining parameters, the mining height, width, and advancement speed are dynamically controlled. S4. Calculate an early warning trigger score based on the chute stress data. When the early warning trigger score exceeds a set threshold, trigger a corresponding level of early warning signal.

[0008] Preferably, in step S2, the process of adjusting the chute arrangement is as follows: Based on the pre-set tunnel layout principle, the tunnel spacing, the relative position of the tunnel and the coal seam, and the tunnel shape are used as layout parameters. The specific values ​​of the layout parameters are set to establish a design space including the spacing parameter set, the position parameter set, and the shape parameter set. Generate a layout solution set containing N candidate solutions through a parameter combination generator, uniquely encode each candidate solution and record its geometric feature parameters; A parametric trench geometry model was established in the 3D geological model, the area surrounding the trench was calibrated, and adaptive meshing was used to refine the mesh in the area surrounding the trench. Set excavation simulation conditions, monitor the top and bottom plate movement, side convergence and plastic zone expansion data in real time, and construct a stress concentration factor , roof subsidence , support cost index and construction complexity evaluation indicator system; Calculate the comprehensive evaluation value of each candidate solution and rank them from largest to smallest according to the evaluation value. Perform secondary verification and parameter sensitivity analysis on the candidate solution ranked first. When the stress concentration factor exceeds the set critical value in the secondary verification, the candidate scheme is adjusted and the adjusted candidate scheme is recorded as the target drift layout scheme.

[0009] Preferably, the calibration of the peripheral area of ​​the drift includes: The coal seam spacing, coal seam thickness, coal seam dip, roof and floor lithology and fault distribution of multiple coal seams are combined into a coal seam parameter list; According to the mapping relationship table based on the pre-set coal seam parameter list and the encrypted area benchmark range, the encrypted area benchmark length is matched and encryption length threshold ; Set the encryption length expansion factor, denoted as , calculate the adjusted length of the encrypted area , ,in, To set the expansion factor threshold; Take the center point of the chute as the reference point, To extend the length, the area is expanded and the expanded area is marked as the area surrounding the chute.

[0010] Preferably, the setting of the encryption length extension coefficient includes: Extracting roof compressive strength from roof and floor lithology and the friction coefficient of the bottom plate , extract fault density from fault distribution and fault strike angle , and the coal seam spacing, coal seam thickness and coal seam inclination are respectively 、 and ; Will 、 、 、 、 and and Converted into stress superposition factor , thickness correction factor , inclination influence coefficient , roof stability index and fault disturbance factor ; Normalize the converted parameters to generate dimensionless stress gradient terms , plastic strain ratio and normalized fault factor ; Constructing an expansion coefficient based on the normalization parameter The calculation model outputs the encryption length expansion coefficient, where the calculation model is specifically expressed as: , Indicates the calculation of Sigmoid activation function, 、 and is the set constant term, Represents the rock mass integrity coefficient.

[0011] Preferably, the determining of the anchor spacing and the support density includes: The weights of coal seam inclination, coal seam thickness, fault strike angle and roof compressive strength are determined by the judgment matrix, which are respectively recorded as 、 、 and , and update the judgment matrix through the sliding window algorithm based on the pre-set window parameters; The coal seam inclination, coal seam thickness, fault strike angle and roof compressive strength are standardized by the range method and marked as 、 、 and , set the support adjustment coefficient ; Calculate anchor spacing , , The default anchor spacing is pre-set. The lower limit value of the interval for adjusting the anchor spacing; Calculation of scaffold density , , The default support density is preset.

[0012] Preferably, the specific construction process of the coal mining sequence recognition model includes: Calculate the seam spacing score based on seam spacing and seam thickness; The tunnel cross-section convergence rate is obtained in real time using a laser scanner. The tunnel stability correction factor is set based on the support stress monitored by the pressure sensor. The tunnel stability score is then calculated by combining the roof integrity index, support strength coefficient, mining stress coefficient, and tunnel stability correction factor. Establish a joint evaluation function of coal seam thickness and interbedded gangue distribution to output mining complexity score; Based on the pre-set entropy weight hybrid mechanism, the weights of coal seam spacing score, drift stability score, and mining complexity score are dynamically allocated; The linear regression function is selected as the basic model framework, the coal seam spacing score, the drift stability score, and the mining complexity score are taken as input variables, and the comprehensive mining adaptability score is taken as the output, thereby constructing a coal mining sequence identification model.

[0013] Preferably, the generating of the mining sequence includes: Construct a three-dimensional hybrid coding matrix consisting of a mining unit layer, a direction decision layer, and a time weight layer; Based on the output results of the coal mining sequence recognition model and the real-time monitored roof displacement value, the pre-set conflict detection based on constraint bundle perception is performed, the fitness value with conflict penalty is output, and the conflict-free optimized coding sequence is output through the genetic algorithm; Extract the mining sequence from the conflict-free optimized coding sequence, calculate the optimal travel path based on the Dijkstra algorithm, and generate a standardized mining instruction set with time window constraints; Check whether the time windows of adjacent tasks overlap. If so, trigger the time reordering operator immediately and update the mining order. If not, output the mining order based on the standardized mining instruction set.

[0014] Preferably, the mining unit layer uses one-hot coding to represent the coal seam number, and the coding dimension is equal to the total number of mineable coal seams; The direction decision layer stores enumeration values ​​of three-state mining direction vectors, including horizontal mining mode 0, upward mining mode 1, and downward mining mode 2; The temporal weight layer uses a floating-point value normalized by a Sigmoid function and mapped to the interval [0.1, 1.0].

[0015] Preferably, during the iterative process of the genetic algorithm, the interlayer displacement is detected in real time by means of millimeter wave radar, which is recorded as , and then dynamically adjust the mutation probability in the genetic algorithm satisfy: , , To set the first reference interlayer displacement, To set a constant; , , To set the second reference interlayer displacement, To set the scale; , , To set a constant.

[0016] Preferably, the calculation of the early warning trigger score includes: Extract the maximum stress value from the trench stress data; If the maximum stress value is lower than the set warning stress value, 0 is used as the warning trigger score. Otherwise, the stress concentration factor and stress change rate are extracted from the chute stress data. The maximum stress value, stress concentration factor, and stress change rate are normalized. The processing results are used as Sigmoid input variables, and the warning coefficient is output. The product of the warning coefficient and the preset warning score is used as the warning trigger score.

[0017] It can be seen from the above technical solutions that the advantages of the present invention are: (1) The present invention constructs a geological model by comprehensively collecting coal seam parameters, thus overcoming the shortcomings of traditional geological analysis. Finite element analysis replaces empirical judgment in the layout of the trench and accurately determines the anchor spacing and support density, thereby improving safety and economic benefits. At the same time, the mining sequence is planned with the help of recognition models and genetic algorithms, realizing the transition from fixed rules to dynamic optimization. Mining parameters are flexibly adjusted according to the mapping table, breaking the limitations of traditional real-time control. On the other hand, a targeted early warning mechanism is set up to ensure the effectiveness and reliability of the early warning.

[0018] (2) The present invention constructs a tunnel layout optimization system based on coal seam geological parameters through three-dimensional geological modeling and finite element dynamic analysis. This greatly improves the stability margin of the tunnel layout scheme, reducing safety risks while ensuring mining efficiency.

[0019] (3) The present invention establishes a dynamic calculation mechanism for the coordinated support parameters of anchor rods and supports, thereby achieving dynamic adaptation of support density to parameters such as coal seam inclination and thickness, ensuring the stability of the drift during coal seam mining, and effectively solving the problem of drift instability in multi-coal seam mining.

[0020] (4) The present invention realizes a leap-forward upgrade from experience-driven to data-driven decision-making for multi-coal seam mining sequence by setting up a three-dimensional hybrid coding genetic algorithm and taking into account constraints in multiple dimensions, thus providing core algorithm support for safe and efficient mining under complex geological conditions.

[0021] (5) The present invention constructs a scientific, dynamic and explainable encryption zone expansion control architecture when calibrating the surrounding area, which is significantly superior to the traditional single parameter or static model. It has breakthrough advantages in safety, economy and adaptability, and is particularly suitable for efficient mining of multiple coal seams under complex geological conditions.

[0022] (6) The present invention realizes closed-loop control of mining continuity constraints through deep coupling based on time reordering operators and dynamic coding parameters, thereby improving the rationality of mining timing. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solution of the present invention, the following is a brief introduction to the drawings required for the description. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0024] Figure 1 A schematic diagram of a process flow of a specific embodiment of the present invention; Figure 2This is a schematic diagram of the flow chart of the chute arrangement adjustment process according to a specific embodiment of the present invention; Figure 3 This is a schematic diagram of a flow chart of a chute peripheral area calibration process according to a specific embodiment of the present invention; Figure 4 A schematic diagram of a coal mining sequence recognition model construction process according to a specific embodiment of the present invention; Figure 5 This is a comparison chart of data acquisition performance in two scenarios according to a specific embodiment of the present invention; Figure 6 This is a comparison of data capture situations in an embodiment of the present invention, taking the 20m range on the right side of the drift as an example; Figure 7 This is a comparison chart of key indicators of an embodiment of the present invention; Figure 8 This is a mapping relationship diagram between the chute stress and coal mining parameters according to a specific embodiment of the present invention. DETAILED DESCRIPTION

[0025] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0026] See also Figure 1 As shown, the present invention provides a multi-coal seam mining optimization control method based on the stability of the drift position, the method includes: the method includes: S1, obtaining the coal seam spacing, coal seam thickness, coal seam inclination, roof and floor lithology and fault distribution of multiple coal seams, and generating a three-dimensional geological model through three-dimensional modeling software.

[0027] Understandably, geological radar, borehole detection and other technologies can be used to obtain geological data of multiple coal seams, including coal seam spacing, thickness, dip, roof and floor lithology and fault distribution, and based on the acquired geological data, a three-dimensional geological model can be generated using three-dimensional modeling software (such as Surpac or GOCAD).

[0028] S2. Based on the 3D geological model, adjust the drift layout to obtain the target drift layout plan, and determine the anchor spacing and support density.

[0029] Specifically, see Figure 2As shown, the adjustment process of the drift layout is as follows: B1. Based on the pre-set drift layout principle, the drift spacing, the relative position of the drift and the coal seam, and the drift shape are used as layout parameters, the specific values ​​of the layout parameters are set, and a design space including the spacing parameter set, the position parameter set, and the shape parameter set is established.

[0030] For example, a design space is established including a spacing parameter set {30m, 40m, 50m}, a location parameter set {top plate arrangement, bottom plate arrangement, middle arrangement}, and a shape parameter set {straight line type, broken line type, curved line type}.

[0031] B2. Generate a layout solution set containing N candidate solutions through a parameter combination generator, uniquely encode each candidate solution and record its geometric feature parameters.

[0032] It is understood that geometric features specifically include features such as size.

[0033] B3. Establish a parametric trench geometry model in the 3D geological model, calibrate the area surrounding the trench, and use adaptive meshing to implement mesh densification in the area surrounding the trench.

[0034] B4. Set up multi-step excavation simulation conditions, monitor the top and bottom plate movement, side convergence and plastic zone expansion data in real time, and construct a stress concentration factor , roof subsidence , support cost index and construction complexity evaluation indicator system.

[0035] Multi-step excavation simulation, as it's understood, involves employing a phased, progressive excavation strategy within numerical simulations, discretizing the complete mining process into multiple, time-sequentially linked construction phases. For example, excavation increments are performed in preset excavation units (e.g., 5m / step). Support sequence control involves aligning the installation of anchors / supports with the excavation sequence. Boundary condition updates dynamically adjust the surrounding rock stress field and displacement constraints.

[0036] It is also understandable that the evaluation index system can be used to calculate the stress concentration factor , roof subsidence , support cost index and construction complexity As the input variable, the linear regression function is selected as the evaluation function to construct the evaluation index system, and the output result is recorded as the evaluation value.

[0037] B5. Calculate the comprehensive evaluation value of each candidate solution and sort them from large to small according to the evaluation value. Perform secondary verification and parameter sensitivity analysis on the candidate solution ranked first.

[0038] Understandably, the calculation of the comprehensive evaluation value of each candidate solution is achieved through the evaluation and index system, and the secondary verification can build a verification hierarchy, such as multi-physics field coupling verification, extreme working condition verification, field data inversion verification, seismic dynamic response, fault activation simulation, borehole stress comparison, etc. The Sobol global sensitivity analysis method can be used to perform parameter sensitivity analysis, wherein the Sobol global sensitivity analysis method is an existing sensitivity analysis method, and its specific formula and specific analysis process will not be repeated here. And the candidate solution ranked first corresponds to Figure 2 The best solution.

[0039] B6. When the stress concentration factor exceeds the set critical value in the secondary verification, the candidate scheme is adjusted and the adjusted candidate scheme is recorded as the target drift layout scheme.

[0040] It is understandable that the critical value may be set accordingly in combination with coal seam mining experience.

[0041] It's important to note that the adjustment process for candidate solutions is as follows: Based on sensitivity analysis, the parameters with the greatest impact on the stress concentration factor are determined. Based on engineering feasibility and safety margins, the parameter adjustment range is set. Then, a parameter combination generator is used to generate new candidate solutions. Finite element analysis is then performed on these new candidate solutions to check whether the stress concentration factor meets the specified criteria. If it still does not, the adjustment process is repeated until the critical value is met. The sensitivity analysis and generator are both common existing technical methods and will not be detailed here.

[0042] The present invention uses 3D geological modeling and finite element dynamic analysis to construct a tunnel layout optimization system based on coal seam geological parameters. This significantly improves the stability margin of the tunnel layout scheme, reducing safety risks while ensuring mining efficiency.

[0043] Preferably, see Figure 3 As shown, in step B4, the area surrounding the drift is calibrated, including: forming a coal seam parameter list with the coal seam spacing, coal seam thickness, coal seam inclination, roof and floor lithology and fault distribution of multiple coal seams.

[0044] According to the mapping relationship table based on the pre-set coal seam parameter list and the encrypted area benchmark range, the encrypted area benchmark length is matched and encryption length threshold .

[0045] Set the encryption length expansion factor, denoted as , calculate the adjusted length of the encrypted area , ,in, To set the expansion factor threshold.

[0046] Take the center point of the chute as the reference point, To extend the length, the area is expanded and the expanded area is marked as the area surrounding the chute.

[0047] It is understandable that the expansion coefficient threshold may specifically be set to 0.6.

[0048] The embodiment of the present invention significantly outperforms traditional single-parameter or static models by constructing a scientific, dynamic, and explainable encryption zone expansion control architecture when calibrating surrounding areas. It has breakthrough advantages in safety, economy, and adaptability, and is particularly suitable for efficient mining of multiple coal seams under complex geological conditions.

[0049] In a specific embodiment, assuming that the geological conditions of a certain mining area are 3.5m mining height, 18m interlayer spacing, and 25GPa elastic modulus of roof sandstone, a 32-channel microseismic system + distributed optical fiber is arranged with a spacing of 5m. Two scenarios, fixed calibration peripheral area and dynamic calibration peripheral area, are set for comparative analysis. The fixed calibration area is centered on the center point of the drift and diffused with a diffusion length of 8m. The dynamic calibration peripheral area is centered on the center point of the drift and diffused with a diffusion length of 12m. A 6m dynamic encryption belt is added in the advancing direction for diffusion. The data acquisition performance is compared in the two scenarios. Figure 5 As shown in the figure, the data capture situation is compared with the 20m range on the right side of the chute. Figure 6 shown.

[0050] in Figure 6 Winning bid Indicates the stress value directly measured by field sensors such as strain gauges and optical fiber sensors, simulating It represents the stress value at the corresponding position calculated by the finite element model, and the error is the percentage of deviation between the measured value and the simulated value.

[0051] according to Figure 5 and Figure 6 It can be seen that the fixed calibration method has blind spots and large errors. In contrast, by expanding the encrypted area from a fixed 8m to a dynamic 12m and adding a 6m advance monitoring zone, the critical stress peak error was reduced from 14% to 3.8%, the plastic zone range prediction error was reduced by 65.9%, the number of missed stress inflection points was reduced from 3 (9m / 12m / 15m) to 0, the monitoring point coverage rate was increased from 68% to 94%, the monitoring blind area was reduced by 76.8% (185m²→43m²), and the interlayer stress transfer error was reduced from 31% to 9%. This verifies the engineering universality of this dynamic calibration peripheral area in stress anomaly warning, interlayer disturbance capture, and support dynamic optimization. Compared with the current fixed calibration peripheral area method, it has significantly improved the coverage and effectiveness of data collection, thereby ensuring the reliability of subsequent data use.

[0052] Furthermore, the densification length expansion coefficient is set, including: Step 1, extracting the top plate compressive strength from the top and bottom plate lithology and the friction coefficient of the bottom plate , extract fault density from fault distribution and fault strike angle , and the coal seam spacing, coal seam thickness and coal seam inclination are respectively 、 and .

[0053] Step 2: 、 、 、 、 and and Converted into stress superposition factor , thickness correction factor , inclination influence coefficient , roof stability index and fault disturbance factor .

[0054] Furthermore, , When the coal seam spacing D is less than 10m, the stress superposition compensation mechanism is automatically activated. As the coal seam spacing D decreases, the stress superposition effect increases. The value increases, , is the base thickness of the coal seam. The greater the coal seam thickness H is, the more obvious the amplification effect on the goaf area will be. , ,. With the inclination of coal seam As the value of The value increases, , the higher the compressive strength of the top plate, The larger the coefficient of friction, the stronger the anti-slip ability of the bottom plate, and the greater the coefficient of friction, the lower the expansion demand. , fault density The larger the fault strike angle, The closer it is to 90°, the greater the damage to stability. The larger the value.

[0055] It should be noted that the specific numerical part designed in the formula is obtained through specific experiments and theoretical derivation. The same principle applies to the subsequent calculation of stress, strain and other related parameters. For example, according to the theories of elasticity mechanics, material mechanics, etc., the formula is simplified and deformed to obtain the desired value. For example, the coefficient of 0.1 is derived mathematically through analysis of the principles of stress transfer and interaction between coal seams in the theoretical model considering the relationship between coal seam spacing and stress superposition, and is used to describe the degree of influence of coal seam spacing on the stress superposition effect.

[0056] Step 3: Normalize the converted parameters to generate dimensionless stress gradient terms , plastic strain ratio and normalized fault factor .

[0057] Furthermore, , is the vertical stress gradient, in MPa / m, which can be measured underground by geostress measuring instruments. This formula is obtained by dividing This unit conversion factor converts the product of stress gradient and coal seam thickness into a dimensionless quantity. , In order to monitor the plastic strain increment in real time, the plastic strain increment is converted into a dimensionless proportional value by calibrating it based on the elastic-plastic threshold of the rock mass of 0.15. , The maximum value is calibrated based on the historical data of the mining area.

[0058] Step 4: Construct an expansion coefficient based on the normalized parameters. The calculation model outputs the encryption length expansion coefficient, where the calculation model is specifically expressed as: , Indicates the calculation of Sigmoid activation function, 、 and is the set constant term, Represents the rock mass integrity coefficient.

[0059] in, represents the geological structure item, represents the deformation response term, represents the fault weakening term. , , , is the maximum theoretical value, which is used to normalize or scale the previous calculation results to ensure the rationality of subsequent values. is the rock mass integrity coefficient, which is assigned according to the RQD index classification. .

[0060] Specifically, determining the anchor spacing and support density includes: determining the weights of the coal seam inclination, coal seam thickness, fault strike angle, and roof compressive strength through a judgment matrix, which are respectively recorded as 、 、 and , and updates the judgment matrix through the sliding window algorithm based on the pre-set window parameters.

[0061] The coal seam inclination, coal seam thickness, fault strike angle and roof compressive strength are standardized by the range method and marked as 、 、 and , set the support adjustment coefficient .

[0062] Calculate anchor spacing , , The default anchor spacing is pre-set. The lower limit value of the interval for adjusting the anchor spacing.

[0063] Calculation of scaffold density , , The default support density is preset.

[0064] Understandably, the greater the coal seam inclination, the stronger the sliding tendency of the surrounding rock, and the need to reduce the anchor spacing to enhance the shear resistance. The increase in coal seam thickness leads to the expansion of the goaf area and the increase in roof subsidence, which requires the increase of anchor density, that is, the reduction of anchor spacing. When the fault strike angle approaches the axial direction of the roadway ( ), the anchor bolt arrangement needs to be denser. The lower the roof compressive strength, the worse the roof self-stabilization ability and the higher the support requirement. The coal seam inclination, coal seam thickness, fault strike angle, and roof compressive strength are selected as the main representative parameters for analysis. Among the selected parameters, the coal seam inclination and coal seam thickness have a negative impact on the anchor bolt spacing. That is, the larger the coal seam inclination and coal seam thickness, the smaller the anchor bolt spacing requirement. However, the fault strike angle and roof compressive strength have a positive impact on the anchor bolt spacing. That is, the larger the fault strike angle and roof compressive strength, the larger the anchor bolt spacing requirement.

[0065] It is also understandable that a large coal seam inclination will lead to uneven distribution of roof pressure, requiring a higher support density. An increase in coal seam thickness will bring greater roof pressure, also requiring more supports. A small fault angle may increase the instability of the surrounding rock, requiring dense support, and a low roof compressive strength may require more dense supports. The coal seam inclination and coal seam thickness have a positive effect on the support density. The greater the coal seam inclination and the greater the coal seam thickness, the greater the support density required. The fault strike angle and the roof compressive strength have a negative effect on the anchor bolt spacing. The greater the fault strike angle and the roof compressive strength, the lower the support density required.

[0066] The embodiment of the present invention establishes a dynamic calculation mechanism for the coordinated support parameters of anchor rods and supports, realizes the dynamic adaptation of support density to parameters such as coal seam inclination and thickness, ensures the stability of the drift during coal seam mining, and effectively solves the problem of drift instability in multi-coal seam mining. The specific comparison can be seen as follows Figure 7 shown.

[0067] Combine Figure 7 Data from the 2018 coal mine support design study show that the proposed model provides more accurate calculations of anchor spacing and support density compared to traditional methods. It also improves roof subsidence control accuracy by an average of 60%-67%, and enhances adaptability to extreme working conditions by 46%-59%. Furthermore, while maintaining effective support, the proposed model significantly reduces support costs (by 18%-21%) and accident rates (by 52%-78%), achieving a dual optimization of safety and economic efficiency. This allows support parameters to dynamically adapt to complex geological conditions, transcending the static limitations of traditional methods.

[0068] For example, the specific process of determining the weights of coal seam inclination, coal seam thickness, fault strike angle and roof compressive strength through judgment matrix is ​​as follows: According to the theory of hierarchical analysis method, experts compare the parameters pairwise to determine the relative importance, and construct a judgment matrix. The matrix elements are Representation parameters Relative to the parameter The importance of each factor was quantified using a 1-9 scale.

[0069] Construct a matrix based on the mean score of a mining area expert group and passed the consistency test (CR<0.1), Among them, a12=1 / 2 indicates that the importance of coal seam dip is 1 / 2 of coal seam thickness, that is, thickness has a more significant impact on stability; a13=3 indicates that the importance of coal seam dip is 3 times of fault strike angle, that is, faults are not developed in this mining area; a34=1 / 2 indicates that the importance of fault strike angle is 1 / 2 of roof compressive strength.

[0070] Calculate the matrix by entropy weight method or arithmetic mean method ,get 、 、 and The arithmetic mean method normalizes each column of the judgment matrix and then calculates the mean of each row. For example, each column element is divided by the column sum to obtain a normalized matrix. The mean of each row of the normalized matrix is ​​then calculated and used directly as the weight. The entropy weight method calculates the weight based on the degree of data dispersion (entropy value). The greater the data variation, the higher the weight.

[0071] Understandably, the specific setting formula for setting the support adjustment coefficient is: ,in, , , , is a natural constant, . 、 and Indicates that from 0 and or Or The maximum value is taken to ensure that the exponential part of the formula does not appear to be inconsistent with practical meaning due to negative numbers. At the same time, it ensures that the calculation results are in the same dimension, which is convenient for data analysis. The calculation result of the support adjustment coefficient is between 0 and 1, which is convenient for the quantitative determination of subsequent adjustments.

[0072] It should be noted that the impact of coal mine geological conditions, such as coal seam inclination, coal seam thickness, fault strike angle, and roof compressive strength, on support requirements is a gradual process. The exponential function, with its smooth and continuous variation, can well reflect this gradual relationship. For example, as the coal seam inclination gradually increases, the support requirement gradually increases. The exponential function can adjust the support coefficient in a continuous and realistic manner based on changes in geological parameters. Therefore, the exponential function is selected as the functional framework for the support adjustment coefficient.

[0073] S3. Establish a coal mining sequence recognition model, and then combine it with a genetic algorithm to generate a mining sequence. During the mining process, the stress data and mining position of the trench are monitored in real time. At the same time, the mining height, width and advancement speed are dynamically adjusted based on the pre-set mapping relationship table between the trench stress and the coal mining parameters.

[0074] Real-time monitoring of tunnel stress data can be achieved by installing stress sensors, including pressure sensors and strain sensors, at key locations within the tunnel, such as the tunnel walls, roof, and floor. The sensor density should be appropriately designed based on the tunnel length, geological conditions, and mining process to ensure comprehensive and accurate monitoring of tunnel stress distribution.

[0075] Specifically, see Figure 4 As shown in FIG, the specific construction process of the coal mining sequence identification model includes: calculating the coal seam spacing score based on the coal seam spacing and coal seam thickness.

[0076] The tunnel cross-section convergence rate is obtained in real time through a laser scanner, and the tunnel stability correction factor is set in combination with the support stress monitored by the pressure sensor. The tunnel stability score is then calculated by combining the roof integrity index, support strength coefficient, mining stress coefficient and tunnel stability correction factor.

[0077] A joint evaluation function of coal seam thickness and interbedded gangue distribution is established to output the mining complexity score.

[0078] Based on the pre-set entropy weight mixing mechanism, the weights of coal seam spacing score, drift stability score and mining complexity score are dynamically allocated.

[0079] The linear regression function is selected as the basic model framework, the coal seam spacing score, the drift stability score, and the mining complexity score are taken as input variables, and the comprehensive mining adaptability score is taken as the output, thereby constructing a coal mining sequence identification model.

[0080] Understandably, the specific formula for the seam spacing score is: , Score the seam spacing, For the The measured spacing between coal seams is accurately measured by geological exploration equipment. For the The thickness of the coal seam is determined by professional measurement means. is the structural influence factor, with a value of 0.8 in fault development areas and 1.2 in intact areas. It is the average value of the spacing between all coal seams, obtained by averaging the measured spacing between each layer. The cumulative number of months of mining is counted from the start of mining operations and is continuously updated as mining progresses. This partly reflects that the coal seam spacing score will be dynamically adjusted as the mining time changes. This part serves as an adjustment coefficient that changes with time and corrects the previous cumulative calculation results. The formula comprehensively considers the impact of coal seam characteristics, geological structure and mining time on the coal seam spacing score. Among them, when other conditions are fixed, the larger the coal seam spacing, the higher the score, the thicker the coal seam, the lower the score, the more complex the geological structure, the lower the score, and as the mining time goes by, the score related to the coal seam spacing will decrease to integrate the changes in the surrounding geological environment caused by long-term mining.

[0081] The specific formula of the dynamic stability correction factor is: , is the dynamic stability correction factor, It is the maximum value of the tunnel section convergence rate, which is determined according to the tunnel design standards and safety regulations. and They are the maximum and minimum values ​​of the support stress monitoring values, respectively, which are obtained through statistics of long-term monitoring data.

[0082] The specific formula for chute stability score is: , Score the stability of the chute. is the roof integrity index, which is obtained by evaluating the roof rock integrity, crack development, etc. The mining stress concentration factor reflects the degree of stress concentration caused by mining during the mining process and is related to factors such as mining technology and coal seam occurrence conditions. This assessment system integrates multiple real-time monitoring data and key influencing factors to accurately reflect the stability of the drift.

[0083] The specific formula of the joint evaluation function of coal seam thickness and interbedded gangue distribution is: ,in, represents the mining complexity score, is the standard deviation of coal seam thickness, reflecting the degree of change in coal seam thickness. is the average thickness of the coal seam. The number of interbedded gangue layers is determined by geological survey. is the ratio of the length of the interlayer to the characteristic length of the assessment area, , is the length of the interbedded gangue layer, which is also obtained from geological survey. To assess the area, the area is determined according to the mining planning range, and the present invention defaults to a rectangle. In order to measure the actual amount of water inflow, hydrological monitoring equipment is used for real-time measurement. , is the baseline water inflow, determined based on the hydrogeological conditions of the coal mine region and industry experience. This function comprehensively considers the impact of coal seam thickness variations, interbedded gangue distribution, and hydrogeological conditions on mining complexity.

[0084] Furthermore, the pre-set entropy weight hybrid mechanism is an entropy weight-expert hybrid mechanism, that is, the entropy weight-expert hybrid mechanism is used to dynamically allocate weights, and its specific formula is: .

[0085] This formula is used to calculate the index Dynamic weight , integrating the real-time entropy value and preset expert weight matrix , through a time interval The relevant coefficients are used to dynamically adjust the proportions of the two in the weight calculation, so that the weights can change dynamically according to real-time data and expert experience, and more comprehensively and accurately reflect the importance of each indicator in the comprehensive evaluation. It is a value between 0 and 1. The sum of the weights of all indicators is usually 1. It is used to measure indicators in comprehensive evaluation. Relative to other indicators, and These two parts are weight adjustment coefficients, Represents the time from a certain initial moment to the current moment The time interval is in seconds. 86400 is the number of seconds in a day. Indicates that with time interval Increase and gradually decrease, It means that as the time interval Increase and gradually increase. Reflection indicators The degree of dispersion or variation of the data, This represents the initial weighting assigned by experts to the importance of each indicator in the comprehensive evaluation based on their expertise and experience. This weighting is determined before considering real-time data and reflects the experts' prior judgment of each indicator.

[0086] in, This part is based on the weight part of the real-time entropy value calculation. If the change Smaller, that is, indicator The data is relatively stable and concentrated, and provides more effective information. is larger, then multiply by After that, this part has a significant impact on the final weight The contribution of this indicator is relatively greater, indicating that the importance of this indicator is relatively improved. This part is based on the expert weight part, when the time interval Smaller, Small, indicating that the influence of expert weight on the final weight is relatively small. The increase, As it increases gradually, the influence of expert weight on the final weight gradually increases.

[0087] In yet another specific embodiment, generating a mining sequence includes: I1, constructing a three-dimensional hybrid coding matrix consisting of a mining unit layer, a direction decision layer, and a time weight layer.

[0088] Among them, the mining unit layer uses one-hot coding to represent the coal seam number, and the coding dimension is equal to the total number of mineable coal seams.

[0089] Understandably, the dimensions of this encoding scheme precisely correspond to the total number of mineable coal seams. In the encoding vector, the corresponding coal seam number position is 1, and all other positions are 0. For example, if there are three mineable coal seams, the first one-hot encoding is [1,0,0], the second one-hot encoding is [0,1,0], and the third one-hot encoding is [0,0,1]. This encoding format clearly and uniquely identifies each mineable coal seam, providing basic data support for subsequent mining sequence planning.

[0090] The direction decision layer stores enumeration values ​​of three-state mining direction vectors, including horizontal mining mode 0, upward mining mode 1, and downward mining mode 2.

[0091] Understandably, the three-state mining direction vectors are stored in the direction decision layer. Through this vector representation, the mining directions adopted by different mining units can be clearly described, providing key direction decision information for comprehensive consideration of coal seam geological conditions and mining efficiency, and then clearly planning the mining direction of each unit to ensure the consistency and rationality of mining operations.

[0092] The temporal weight layer uses a floating-point value normalized by a Sigmoid function and mapped to the interval [0.1, 1.0].

[0093] Understandably, the time weight layer takes into account the varying mining times and importance of different mining units, assigning each unit a time weight. This time weight reflects the urgency of the unit's mining time or its impact on the overall mining schedule. For example, units near hazardous areas such as faults may be assigned a higher time weight to prioritize mining and avoid greater risks associated with geological changes. The code in this layer represents the time weight value corresponding to each mining unit. By constructing a three-dimensional hybrid coding matrix consisting of these three layers, a comprehensive and structured data foundation is provided for subsequent mining planning.

[0094] I2. Based on the output results of the coal mining sequence recognition model and the real-time monitored roof displacement value, the pre-set conflict detection based on constraint bundle perception is performed, the fitness value with conflict penalty is output, and the conflict-free optimized coding sequence is output through the genetic algorithm.

[0095] It is understood that conflict detection can be performed by a conflict detection algorithm, wherein the fitness function of the conflict detection algorithm is The specific formula is: It calculates the safety benefit and conflict penalty by weighting them to get a comprehensive evaluation value, which is used to measure the pros and cons of a mining plan or system operation status. The safety benefit part is expected to be as large as possible, while the conflict penalty part is expected to be as small as possible. , which is used to evaluate indicators related to safety benefits, represents the spacing parameter, Is a critical spacing value, here it can be taken as 50mm, is a constant term, The ratio of the actual advancing distance to the theoretical maximum distance is multiplied by the ratio of the actual mining height to the planned mining height. The result reflects the degree of compliance between the actual operating efficiency of the equipment in terms of advancement and mining height and the expected degree. The higher the value, the better the equipment efficiency. is the conflict penalty term, which is obtained by weighted summation of four different conflict detection values. is the critical spacing value, and for example, it can be 50 mm.

[0096] For example, outputting a conflict-free optimized coding sequence using a genetic algorithm includes optimizing the coding sequence using a genetic algorithm based on a fitness value with a conflict penalty. First, an initial population is generated, where the individuals in the population represent different mining sequence coding sequences. These sequences can be constructed by randomly generating some individuals, or by referencing historically successful mining sequence schemes to generate some individuals.

[0097] Genetic operations such as selection, crossover, and mutation are performed on the initial population. Selection is based on the fitness of individuals; individuals with higher fitness values ​​are more likely to be selected to form the next generation. Crossover swaps parts of the code of two selected individuals to create a new individual, simulating gene exchange in biological genetics. Mutation randomly changes certain codes within an individual to introduce new genetic features and prevent the algorithm from falling into local optimal solutions. For example, within the mining unit layer code, the numbering order of a mining unit is randomly changed.

[0098] By repeating these genetic operations over multiple generations, the individuals in the population gradually evolve towards a more optimal state, ultimately outputting a conflict-free, optimized coding sequence. In each iteration, the fitness value of each individual in the new population is calculated, and selection and evolution are performed based on the fitness value until a preset termination condition is met, such as reaching a certain number of iterations or no significant improvement in fitness.

[0099] I3. Extract the mining sequence from the conflict-free optimized coding sequence, calculate the optimal travel path based on the Dijkstra algorithm, and generate a standardized mining instruction set with time window constraints.

[0100] Understandably, a standardized mining instruction set with time window constraints is generated. Its data structure satisfies the following requirements: each mining step defines the mining height, width, advancement speed, and safety threshold. Instructions are transmitted using dual redundant channels of industrial Ethernet and 5G, and data packets are encrypted with AES-256 and signed with SHA-256.

[0101] It's also understandable that the time window constraint is determined based on factors such as the estimated mining time for each mining unit and the equipment preparation time. For example, the instruction set may specify that the first mining unit should mine between 9:00 and 11:00 AM, and that the equipment should arrive by 8:50 AM along a specific optimal path from its current location.

[0102] I4. Check whether the time windows of adjacent tasks overlap. If so, immediately trigger the time reordering operator and update the mining order. If not, output the mining order based on the standardized mining instruction set.

[0103] In one specific embodiment, outputting a mining order based on a standardized mining instruction set is specifically illustrated as follows: The mining order for each mining unit is directly extracted from the mining unit layer code of the conflict-free optimized code sequence. For example, if the code sequence is [3, 1, 5, 2, 4], the mining order is that the third mining unit is mined first, followed by the first, fifth, second, and fourth units.

[0104] Furthermore, during the iteration of the genetic algorithm, the inter-layer displacement is detected in real time with the help of millimeter-wave radar, which is recorded as , and then dynamically adjust the mutation probability in the genetic algorithm satisfy: , , To set the first reference interlayer displacement, To set a constant.

[0105] , , To set the second reference interlayer displacement, To set the scale.

[0106] , , To set a constant.

[0107] For example, 、 、 、 and Based on the experience of coal seam excavation engineering, the values ​​can be set to 5mm, 10mm, 0.05, 0.15 and 0.02 respectively.

[0108] When Δd≤5mm, the geological conditions are relatively stable and the probability of variation is relatively low. =0.05, which is conducive to the global search of the algorithm, maintains the diversity of the population, and avoids the algorithm from focusing on the local area search too early. When 5mm<Δd≤10mm, the geological conditions have changed, and the mutation probability is adjusted to =0.05+0.02(Δd−5). As Δd increases, the mutation probability increases, giving the algorithm more opportunities to mutate the current solution and explore better solutions in the local range, avoiding falling into the local optimum too early. When Δd>10mm, the geological conditions change greatly, and the mutation probability is fixed at =0.15, which enables the algorithm to quickly adjust the search direction, adapt to new geological conditions, and find a mining sequence that better meets the current geological conditions.

[0109] The embodiment of the present invention realizes closed-loop control of mining succession constraints through deep coupling of time reordering operators and dynamic coding parameters, improves the rationality of mining timing, and by setting up a three-dimensional hybrid coding genetic algorithm and considering constraints in multiple dimensions at the same time, realizes a leap-forward upgrade from experience-driven to data-driven decision-making in multi-coal seam mining sequence, providing core algorithm support for safe and efficient mining under complex geological conditions.

[0110] It should be added that the mapping relationship table between chute stress and coal mining parameters can be specifically as follows: Figure 8 shown.

[0111] Specifically, the specific adjustment value can be determined by comparing the actual monitored stress with the set stress critical value. If the stress critical value is exceeded, the minimum-maximum normalization process is performed, and the processing result is used as the input variable of the Sigmoid activation function. The output is the adjustment proportional coefficient and multiplied by the corresponding maximum adjustment value. If the stress critical value is not exceeded, no adjustment is performed.

[0112] S4. Calculate an early warning trigger score based on the chute stress data. If the early warning trigger score exceeds a set threshold, trigger a corresponding level of early warning signal.

[0113] Specifically, calculating the early warning trigger score includes extracting the maximum stress value from the chute stress data.

[0114] If the maximum stress value is lower than the set warning stress value, 0 is used as the warning trigger score. Otherwise, the stress concentration factor and stress change rate are extracted from the chute stress data. The maximum stress value, stress concentration factor, and stress change rate are normalized. The processing results are used as Sigmoid input variables, and the warning coefficient is output. The product of the warning coefficient and the preset warning score is used as the warning trigger score.

[0115] Understandably, the pre-set warning score can be based on a ten-point system or a hundred-point system. When the hundred-point system is used, the specific warning mechanism of the graded warning is as follows: when the warning trigger score is lower than 30, it is a low risk, no warning is triggered, and normal monitoring of the chute stress data continues.

[0116] When the warning trigger score is greater than or equal to 30 points and less than 60 points, it is a medium risk and a yellow warning is issued to remind relevant personnel to strengthen their attention and inspections of the chute area.

[0117] When the warning trigger score is greater than 60, it is a high risk and a red warning is issued. Appropriate safety measures must be taken immediately, such as stopping mining operations in related areas and carrying out support and reinforcement.

[0118] This invention constructs a geological model by comprehensively collecting coal seam parameters in real time, addressing the shortcomings of traditional geological analysis. Finite element analysis replaces empirical judgment in drift layout, accurately determining anchor spacing and support density, improving safety and economic efficiency. Furthermore, mining sequence planning utilizes recognition models and genetic algorithms, shifting from fixed rules to dynamic optimization. Mining parameters can be flexibly controlled based on mapping tables, breaking the limitations of traditional real-time control. Furthermore, a targeted early warning mechanism ensures both effectiveness and reliability.

[0119] The above contents are merely examples and explanations of the concept of the present invention. Those skilled in the art may make various modifications or additions to the described specific embodiments or replace them in a similar manner. As long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, they should all fall within the scope of protection of the present invention.

Claims

1. A multi-coal seam mining optimization control method based on the stability of the drift position, characterized in that: The method includes: S1. Obtain data on the inter-coal spacing, thickness, dip, roof and floor lithology, and fault distribution of multiple coal seams, and generate a three-dimensional geological model using three-dimensional modeling software; S2. Based on the generated 3D geological model, adjust the drift layout to obtain the target drift layout plan and determine the corresponding anchor spacing and support density; S3. Establish a coal mining sequence recognition model based on the 3D geological model. The coal mining sequence recognition model combines a genetic algorithm to generate a mining sequence. During the mining process, the trench stress data and mining position are monitored in real time. Based on a pre-set mapping relationship table between trench stress and coal mining parameters, the mining height, mining width, and advancement speed are dynamically controlled. S4. Calculate an early warning trigger score based on the chute stress data. When the early warning trigger score exceeds a set threshold, trigger a corresponding level of early warning signal.

2. The multi-coal seam mining optimization control method based on drift position stability according to claim 1 is characterized in that: In step S2, the process of adjusting the chute layout is as follows: Based on the pre-set tunnel layout principle, the tunnel spacing, the relative position of the tunnel and the coal seam, and the tunnel shape are used as layout parameters. The specific values ​​of the layout parameters are set to establish a design space including the spacing parameter set, the position parameter set, and the shape parameter set. Generate a layout solution set containing N candidate solutions through a parameter combination generator, uniquely encode each candidate solution and record its geometric feature parameters; A parametric trench geometry model was established in the 3D geological model, the area surrounding the trench was calibrated, and adaptive meshing was used to refine the mesh in the area surrounding the trench. Set excavation simulation conditions, monitor the top and bottom plate movement, side convergence and plastic zone expansion data in real time, and construct a stress concentration factor , roof subsidence , support cost index and construction complexity evaluation indicator system; Calculate the comprehensive evaluation value of each candidate solution and rank them from largest to smallest according to the evaluation value. Perform secondary verification and parameter sensitivity analysis on the candidate solution ranked first. When the stress concentration factor exceeds the set critical value in the secondary verification, the candidate scheme is adjusted and the adjusted candidate scheme is recorded as the target drift layout scheme.

3. The multi-coal seam mining optimization control method based on drift position stability according to claim 2 is characterized in that: The calibration of the surrounding area of ​​the drift includes: The coal seam spacing, coal seam thickness, coal seam dip, roof and floor lithology and fault distribution of multiple coal seams are combined into a coal seam parameter list; According to the mapping relationship table based on the pre-set coal seam parameter list and the encrypted area benchmark range, the encrypted area benchmark length is matched and encryption length threshold ; Set the encryption length expansion factor, denoted as , calculate the adjusted length of the encrypted area , ,in, To set the expansion factor threshold; Take the center point of the chute as the reference point, To extend the length, the area is expanded and the expanded area is marked as the area surrounding the chute.

4. The multi-coal seam mining optimization control method based on drift position stability according to claim 3 is characterized in that: The step of setting the encryption length extension factor includes: Extracting roof compressive strength from roof and floor lithology and the friction coefficient of the bottom plate , extract fault density from fault distribution and fault strike angle , and the coal seam spacing, coal seam thickness and coal seam inclination are respectively 、 and ; Will 、 、 、 、 and and Converted into stress superposition factor , thickness correction factor , inclination influence coefficient , roof stability index and fault disturbance factor ; Normalize the converted parameters to generate dimensionless stress gradient terms , plastic strain ratio and normalized fault factor ; Constructing an expansion coefficient based on the normalization parameter The calculation model outputs the encryption length expansion coefficient, where the calculation model is specifically expressed as: , Indicates the calculation of Sigmoid activation function, 、 and is the set constant term, Represents the rock mass integrity coefficient.

5. The multi-coal seam mining optimization control method based on drift position stability according to claim 3 is characterized in that: Determining the anchor rod spacing and support density includes: The weights of coal seam dip, coal seam thickness, fault strike angle and roof compressive strength are determined by the judgment matrix, which are respectively expressed as 、 、 and , and update the judgment matrix through the sliding window algorithm based on the pre-set window parameters; The coal seam inclination, coal seam thickness, fault strike angle and roof compressive strength are standardized by the range method and marked as 、 、 and , set the support adjustment coefficient ; Calculate anchor spacing , , The default anchor spacing is pre-set. The lower limit value of the interval for adjusting the anchor spacing; Calculation of scaffold density , , The default support density is preset.

6. The multi-coal seam mining optimization control method based on drift position stability according to claim 3 is characterized in that: The specific construction process of the coal mining sequence recognition model includes: Calculate the seam spacing score based on seam spacing and seam thickness; The tunnel cross-section convergence rate is obtained in real time using a laser scanner. The tunnel stability correction factor is set based on the support stress monitored by the pressure sensor. The tunnel stability score is then calculated by combining the roof integrity index, support strength coefficient, mining stress coefficient, and tunnel stability correction factor. Establish a joint evaluation function of coal seam thickness and interbedded gangue distribution to output mining complexity score; Based on the pre-set entropy weight hybrid mechanism, the weights of coal seam spacing score, drift stability score, and mining complexity score are dynamically allocated; The linear regression function is selected as the basic model framework, the coal seam spacing score, the drift stability score, and the mining complexity score are taken as input variables, and the comprehensive mining adaptability score is taken as the output, thereby constructing a coal mining sequence identification model.

7. The multi-coal seam mining optimization control method based on drift position stability according to claim 1 is characterized in that: The generating mining sequence includes: Construct a three-dimensional hybrid coding matrix consisting of a mining unit layer, a direction decision layer, and a time weight layer; Based on the output results of the coal mining sequence recognition model and the real-time monitored roof displacement value, the pre-set conflict detection based on constraint bundle perception is performed, the fitness value with conflict penalty is output, and the conflict-free optimized coding sequence is output through the genetic algorithm; Extract the mining sequence from the conflict-free optimized coding sequence, calculate the optimal travel path based on the Dijkstra algorithm, and generate a standardized mining instruction set with time window constraints; Check whether the time windows of adjacent tasks overlap. If so, trigger the time reordering operator immediately and update the mining order. If not, output the mining order based on the standardized mining instruction set.

8. The multi-coal seam mining optimization control method based on drift position stability according to claim 7, characterized in that: The mining unit layer uses one-hot coding to represent the coal seam number, and the coding dimension is equal to the total number of mineable coal seams; The direction decision layer stores enumeration values ​​of three-state mining direction vectors, including horizontal mining mode 0, upward mining mode 1, and downward mining mode 2; The temporal weight layer uses a floating-point value normalized by a Sigmoid function and mapped to the interval [0.1, 1.0].

9. The multi-coal seam mining optimization control method based on drift position stability according to claim 7, characterized in that: During the iteration of the genetic algorithm, the inter-layer displacement is detected in real time with the help of millimeter-wave radar, which is recorded as , and then dynamically adjust the mutation probability in the genetic algorithm satisfy: , , To set the first reference interlayer displacement, To set a constant; , , To set the second reference interlayer displacement, To set the scale; , , To set a constant.

10. The multi-coal seam mining optimization control method based on drift position stability according to claim 1, characterized in that: The calculation of the early warning trigger score includes: Extract the maximum stress value from the trench stress data; If the maximum stress value is lower than the set warning stress value, 0 is used as the warning trigger score. Otherwise, the stress concentration factor and stress change rate are extracted from the chute stress data. The maximum stress value, stress concentration factor, and stress change rate are normalized. The processing results are used as Sigmoid input variables, and the warning coefficient is output. The product of the warning coefficient and the preset warning score is used as the warning trigger score.

Images