Close-distance coal seam dynamic coal-pillar-free mining method based on intelligent monitoring

Abstract

The invention provides a close-distance coal seam dynamic coal-pillar-free mining method based on intelligent monitoring, which comprises the following steps: deploying a multi-parameter sensor network in an upper coal seam working surface, a lower coal seam working surface and overlying strata, monitoring the stress, displacement and fracture development state of the overlying strata in real time, and dynamically optimizing roof cutting parameters and a mining sequence in combination with a hydraulic fracturing intelligent regulation and control technology; in the mining process, monitoring data are analyzed through an MATLAB algorithm, an overlying strata caving prediction model is generated, and the overlying strata caving time and range are guided; and meanwhile, the carbon fiber reinforced polymer nano composite material is adopted for reinforcing the roadway side in the lower coal seam gob-retaining roadway, so that the structural stability is improved. Through intelligent monitoring and dynamic regulation and control, accurate control over overlying strata collapse is achieved, the stress concentration risk is reduced, and the resource recovery rate and mining safety are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of safe coal mining, and in particular to a dynamic pillarless mining method for close-range coal seams based on intelligent monitoring, which is suitable for efficient and safe mining of close-range coal seams with inter-layer spacing less than 50 meters. Background Art

[0002] In the field of coal mining, although the close-range coal seam pillar-free technology can reduce resource waste, the traditional method has the following significant defects. First, the control of overburden collapse relies too much on experience. By fixing blasting parameters or inducing overburden collapse by natural collapse, it is impossible to obtain key parameters such as overburden strain and cracks in real time, and it is difficult to adapt to complex geological conditions. For example, when the inclination of the coal seam changes or faults develop, empirical operations can easily lead to uneven collapse, and the local stress concentration coefficient can reach more than 2.8, inducing the risk of rock burst. Secondly, the existing technology lacks the ability to monitor the dynamic response of the overburden in real time. The expansion of overburden cracks and stress evolution have temporal and spatial differences. It is difficult to accurately predict the distribution of low-stress areas and characterize the nonlinear process of overburden collapse with static models alone. As a result, the mining of the lower coal seam still needs to face residual stress interference, and the resource recovery rate has stagnated at around 75% for a long time. Furthermore, traditional gob-retained tunnel support relies on anchor bolts and masonry structures, which lack the ability to resist deformation under dynamic loads. This makes it difficult to withstand the dynamic impact of overburden collapse, leading to tunnel deformation rates as high as 15% to 20%, and maintenance costs exceeding 1.2 million yuan per kilometer, severely restricting mining economics. Thus, the existing technical bottleneck manifests itself in the triple contradictions of "lagging monitoring, extensive control, and fragile support." Summary of the Invention

[0003] In response to the technical problems of existing pillarless mining of close-range coal seams, such as delayed monitoring, extensive control, and fragile support, the present invention provides a dynamic pillarless mining method for close-range coal seams based on intelligent monitoring. By integrating multi-parameter real-time monitoring, data-driven modeling, precise control of hydraulic fracturing, and new material reinforcement technology, dynamic optimization of overburden collapse and efficient resource recovery are achieved.

[0004] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0005] A dynamic pillar-free mining method for close-range coal seams based on intelligent monitoring comprises the following steps:

[0006] S1. Multi-parameter sensor network deployment and data collection:

[0007] S11. Deploy a multi-parameter sensor network in the working face of the upper and lower coal seams in close proximity to the coal seam and in the overburden, wherein the sensors include a fiber Bragg grating sensor for stress monitoring, a laser displacement sensor for displacement monitoring, and a three-dimensional acoustic emission sensor for crack detection;

[0008] S12, upload the sensor data to the cloud platform in real time through the LoRa wireless transmission module, filter and de-noise the sensor data, and normalize it, and extract key characteristic parameters such as stress gradient, displacement rate, and crack density;

[0009] S2. Construction of overburden collapse prediction model and optimization of top cutting parameters:

[0010] S21. Establishment of theoretical model: Based on Griffith's crack propagation theory in rock mechanics, the following critical conditions for overburden collapse are established:

[0011]

[0012] Among them, σ1 and σ3 are the maximum and minimum principal stresses, K IC is the rock fracture toughness, α is the crack half-length;

[0013] S22. Machine Learning Model Training: A fusion model of the random forest algorithm and long short-term memory network is constructed based on MATLAB. The input characteristic parameters include real-time stress gradient, displacement rate, and crack density. The output is the prediction of the time and range of overburden collapse.

[0014] S23. Dynamic visualization and early warning: MATLAB is used to generate three-dimensional overburden stress cloud maps and crack expansion dynamics, displaying the overburden status in real time. When the predicted collapse time deviation is greater than 10%, an early warning is triggered.

[0015] S24, dynamic adjustment of cutting parameters:

[0016] The drilling spacing L is dynamically calculated based on the collapse range R:

[0017] L=max(0.6R,2m)

[0018] The blasting energy E is determined by the following formula:

[0019] E=α·σ eff ·V rock

[0020] Where m represents the unit meter, V rock Indicates the volume of rock mass affected by blasting, in m 3 ;

[0021] S3. Precision control of hydraulic fracturing and optimization of overburden collapse:

[0022] S31. Drilling design in the top-cut area: Based on the model prediction results, radial drill holes are designed above the goaf of the upper coal seam, with the drilling depth being 80% of the overburden thickness;

[0023] S32, Fracturing fluid formula: Use water-based fracturing fluid, add 0.5% nano-montmorillonite to enhance sand carrying capacity, and the sand ratio is 15%;

[0024] S33. Dynamic control strategy of hydraulic fracturing parameters:

[0025] Inject high-pressure water into the borehole at a pressure of 10-25 MPa and a flow rate of 20-50 L / min. Adjust the injection pressure according to the real-time fracture expansion speed v:

[0026]

[0027] Wherein, P0 is the initial water injection pressure;

[0028] Fracturing termination conditions: overburden collapse height H ≥ 1.2 times the mining height, and collapse zone stress σ ≤ 3 MPa;

[0029] S4. Dynamic planning of lower coal seam mining and stress field optimization:

[0030] S41. Finite element modeling: A 3D geomechanical model was constructed based on COMSOL Multiphysics. The stress distribution data after mining of the upper coal seam was input, and the stress concentration coefficient of the lower coal seam was solved as follows:

[0031]

[0032] Among them, σ max represents the maximum stress, σ avg represents the mean stress;

[0033] S42, mining priority rule: if K σ >1.5, marked as high-risk area, priority mining; if 1.2≤K σ <1.5, marked as medium-risk area, with secondary priority for mining; if K<K σ <1.2, marked as a safe area and mined last; where K is the lower limit of the threshold;

[0034] S5. Nanocomposite reinforcement and intelligent support of gob-retained tunnels:

[0035] S51. Material preparation: 2% graphene nanosheets are added to carbon fiber reinforced polymer to increase the tensile strength to 800 MPa;

[0036] S52. Spraying process: Use high-pressure airless spraying equipment to spray carbon fiber reinforced polymer composite materials on the surface of the lower coal seam empty roadway. The spraying pressure is 20MPa, the nozzle diameter is 0.5mm, and the coating thickness t is adjusted according to the roadway curvature radius r as follows:

[0037]

[0038] Among them, mm represents the unit millimeter;

[0039] S53, Smart Wall Design: The wall is composed of a shape memory alloy skeleton and ultra-high performance concrete, with the shape memory alloy accounting for 10%. A built-in fiber grating sensor monitors the strain ε. When ε ≥ 0.15%, the shape memory alloy is heated to 60°C, restoring the pre-deformation.

[0040] Retractable hydraulic supports are installed on the side of the tunnel, the support force is automatically adjusted based on stress monitoring data, and the support layout is optimized in real time using MATLAB.

[0041] Furthermore, in step S11, fiber grating sensors are buried in key overburden layers, with one group arranged every 5 meters along the coal seam and arranged in layers every 2 meters in the vertical direction; laser displacement sensors are installed on the tunnel roof and both sides, with the center of the working surface as the reference point and a radial spacing of 10 meters; three-dimensional acoustic emission sensors are arranged around the goaf to capture rock formation crack expansion signals in real time.

[0042] Furthermore, in step S12, the sensor data is subjected to noise suppression and data fusion using a Kalman filter algorithm.

[0043] Furthermore, in step S22, the feature importance ranking of the random forest algorithm is: stress gradient > crack density > displacement rate.

[0044] Furthermore, in step S32, the viscosity μ of the fracturing fluid is adjusted to 50 mPa·s by adding hydroxypropyl guar gum, and the shear rate is 100 s -1 The stability is ≥90%.

[0045] Furthermore, the step S3 further includes optimizing the morphology of the fracturing cracks:

[0046] Directional perforation technology is used to arrange perforations in a spiral pattern with a spacing of 30 cm and a diameter of 8 mm in the borehole to guide the cracks to expand in the preset direction;

[0047] The microseismic event location data is analyzed using MATLAB image processing technology to generate a three-dimensional fracture morphology map in real time, and the perforation density is adjusted based on feedback.

[0048] Furthermore, in step S53, the ultra-high performance concrete compressive strength of the smart wall is ≥150 MPa, and the shape memory alloy restoring force is ≥200 kN / m.

[0049] Furthermore, the step S5 further includes adaptive nano-coating thickness control:

[0050] Before spraying the carbon fiber reinforced polymer layer, the curvature of the roadway surface was scanned using LiDAR to generate a 3D point cloud model;

[0051] Based on the curvature radius r and stress gradient The coating thickness t is dynamically calculated according to the following formula:

[0052]

[0053] A six-axis robotic arm equipped with a spray gun is used to achieve sub-millimeter precision coating construction; the smaller the curvature radius, the more significant the surface stress concentration, and the coating thickness needs to be increased to disperse the load; the elastic modulus E of carbon fiber reinforced polymer c And the coating thickness d satisfies:

[0054] E c =E m +k·d

[0055] Among them, E m represents the initial elastic modulus of carbon fiber reinforced polymer; k represents the modulus enhancement coefficient of carbon fiber reinforced polymer per unit thickness, which is positively correlated with the fiber volume fraction.

[0056] Furthermore, the method further comprises the following steps:

[0057] S6. Real-time assessment of overburden stability driven by multi-source data:

[0058] S61: Based on the sensor data of step S1, construct the overburden stability index SI:

[0059]

[0060] Among them, σ avg is the mean stress; σ peak is the peak stress; δ max is the maximum displacement; L crit is the critical displacement threshold;

[0061] When S62 and SI≤0.6, the early warning system is triggered, mining is suspended and reinforcement measures are initiated.

[0062] Compared with the existing technology, the dynamic pillarless mining method for close-range coal seams based on intelligent monitoring provided by the present invention has the following beneficial effects:

[0063] 1. Precise control of overburden collapse: Real-time monitoring of overburden stress, displacement, and crack data through a multi-parameter sensor network is combined with a machine learning model to dynamically predict the time and range of overburden collapse, achieving an error rate of less than 5%. This significantly reduces the risk of local stress concentration caused by uneven overburden collapse and avoids the probability of strong mining pressure disasters by more than 70%.

[0064] 2. Significantly improved resource recovery rate: Dynamically optimize hydraulic fracturing parameters and mining sequence to ensure that overburden collapse completely covers the goaf, reducing coal pillars left behind. Resource recovery rate is increased to over 90%, an increase of 15% to 20% compared to traditional methods.

[0065] 3. Enhanced tunnel stability: Carbon fiber reinforced polymer (CFRP) nanocomposite materials are sprayed to reinforce the lower coal seam empty tunnels, increasing the compressive strength by 40%. Combined with the intelligent support wall, the support force can be adaptively adjusted to effectively suppress tunnel deformation, extend the service life, and reduce maintenance costs by more than 30%.

[0066] 4. Strong technical scalability: The system supports collaborative monitoring and control of multiple coal seams and is suitable for complex geological conditions (such as high-gas and soft rock formations). Through adaptive updates of model parameters, it can quickly adapt to different mine scenarios.

[0067] 5. The technical system of the present invention overcomes the data blind spots, control lag and structural instability problems of traditional methods, increases the resource recovery rate to more than 90%, and reduces the stress concentration factor to 1.2, providing a new paradigm for safe and efficient mining of close-range coal seams. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 It is a flow chart of the dynamic pillarless mining method for close-range coal seams based on intelligent monitoring provided by the present invention.

[0069] Figure 2 The present invention provides a MATLAB-based three-dimensional overburden stress cloud map and a dynamic visualization diagram of crack expansion.

[0070] Figure 3 This is a thermal map of the distribution of low stress areas in the lower coal seam provided by the present invention.

[0071] Figure 4 This is a three-dimensional graph comparing the stresses in the tunnel before and after reinforcement with the nanocomposite material provided by the present invention. DETAILED DESCRIPTION

[0072] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the present invention is further described below with reference to specific illustrations.

[0073] Please refer to Figure 1 As shown, the present invention provides a dynamic pillar-free mining method for close-range coal seams based on intelligent monitoring, comprising the following steps:

[0074] S1. Multi-parameter sensor network deployment and data collection:

[0075] S11. Deploy a multi-parameter sensor network in the working faces of the upper and lower coal seams of the proximal coal seams and in the overburden. The sensors include fiber Bragg grating sensors for stress monitoring, laser displacement sensors for displacement monitoring, and three-dimensional acoustic emission sensors for crack detection. As a specific embodiment, the fiber Bragg grating sensors are buried in key overburden layers, with one group arranged every 5 meters along the coal seam strike and layered every 2 meters vertically, with a measurement range of 0-50 MPa and an accuracy of ±0.1 MPa. Laser displacement sensors are installed on the roadway roof and sides, with the center of the working face as the reference point, at radial intervals of 10 meters, and with a measurement accuracy of ±1 mm. Three-dimensional acoustic emission sensors are deployed around the goaf to capture rock crack expansion signals in real time, with a monitoring frequency of 1-100 kHz and a positioning error of ≤0.5 m.

[0076] S12. Upload the sensor data to the cloud platform in real time through the LoRa wireless transmission module, filter and de-noise the sensor data, and normalize it, and extract key characteristic parameters such as stress gradient, displacement rate, and crack density. As a specific embodiment, the Kalman filter algorithm is used to suppress noise and fuse data on the sensor data, and its normalization processing and key feature extraction can be achieved using existing technologies.

[0077] S2. Construction of overburden collapse prediction model and optimization of top cutting parameters:

[0078] S21. Establishment of theoretical model: Based on Griffith's crack propagation theory in rock mechanics, the following critical conditions for overburden collapse are established:

[0079]

[0080] Among them, σ1 and σ3 are the maximum and minimum principal stresses, K IC is the rock fracture toughness, α is the crack half-length;

[0081] S22. Machine Learning Model Training: A fusion model of the random forest algorithm and the long short-term memory network (LSTM) was constructed based on MATLAB. The input feature parameters included real-time stress gradient, displacement rate, and crack density, and the output was the prediction of the time and range of overburden collapse.

[0082] S23, dynamic visualization and early warning: MATLAB is used to generate three-dimensional overburden stress cloud map and crack expansion dynamics (such as Figure 2 It displays the overburden status in real time and triggers an early warning when the predicted collapse time deviation is greater than 10%;

[0083] S24, dynamic adjustment of cutting parameters:

[0084] The drilling spacing L is dynamically calculated based on the collapse range R:

[0085] L=max(0.6R,2m)

[0086] The blasting energy E is determined by the following formula:

[0087] E=α·σ eff ·V rock

[0088] Where m represents the unit meter, V rock Indicates the volume of rock mass affected by blasting, in m 3 ;α=0.05.

[0089] S3. Precision control of hydraulic fracturing and optimization of overburden collapse:

[0090] S31. Drilling design in the top-cut area: Based on the model prediction results, radial drill holes are designed above the goaf of the upper coal seam, with the drilling depth being 80% of the overburden thickness;

[0091] S32, Fracturing fluid formula: Use water-based fracturing fluid, add 0.5% nano-montmorillonite to enhance sand carrying capacity, and the sand ratio is 15%;

[0092] S33. Dynamic control strategy of hydraulic fracturing parameters:

[0093] Inject high-pressure water into the borehole at a pressure of 10-25 MPa and a flow rate of 20-50 L / min. Adjust the injection pressure according to the real-time fracture expansion speed v:

[0094]

[0095] Among them, P0 is the initial water injection pressure, and P0 is set to 12 MPa;

[0096] Fracturing termination conditions: overburden collapse height H ≥ 1.2 times the mining height, and collapse zone stress σ ≤ 3 MPa.

[0097] S4. Dynamic planning of lower coal seam mining and stress field optimization:

[0098] S41. Finite element modeling: A 3D geomechanical model was constructed based on COMSOL Multiphysics. The stress distribution data after mining of the upper coal seam was input, and the stress concentration coefficient of the lower coal seam was solved as follows:

[0099]

[0100] Among them, σ max represents the maximum stress, σ avg represents the mean stress;

[0101] Finite element analysis combined with monitoring data is used to generate a low stress distribution thermal map of the lower coal seam (such as Figure 3 As shown), priority is given to mining areas with stress values ​​less than 5MPa;

[0102] S42, mining priority rule: if K σ >1.5, marked as high-risk area, priority mining; if 1.2≤K σ <1.5, marked as medium-risk area, with secondary priority for mining; if K <K σ <1.2, marked as a safe area and mined last; where K is the lower limit of the threshold.

[0103] S5. Nanocomposite reinforcement and intelligent support of gob-retained tunnels:

[0104] S51. Material preparation: 2% graphene nanosheets are added to carbon fiber reinforced polymer (CFRP) to increase the tensile strength to 800 MPa;

[0105] S52. Spraying process: Use high-pressure airless spraying equipment to spray carbon fiber reinforced polymer composite materials on the surface of the lower coal seam empty roadway. The spraying pressure is 20MPa, the nozzle diameter is 0.5mm, and the coating thickness t is adjusted according to the roadway curvature radius r as follows:

[0106]

[0107] Among them, mm represents the unit millimeter;

[0108] Construction process: Clean the tunnel surface → spray primer → spray CFRP layer by layer (each layer is 1mm thick) → cure for 24 hours. The three-dimensional stress comparison of the tunnel before and after nanocomposite reinforcement is shown in the figure below. Figure 4 As shown;

[0109] S53. Smart wall design: The wall is composed of a shape memory alloy (SMA) skeleton and ultra-high performance concrete (UHPC). The SMA accounts for 10% of the total material, with a UHPC compressive strength of ≥150 MPa and an SMA restoring force of ≥200 kN / m. A built-in fiber grating sensor monitors strain ε. When ε ≥ 0.15%, the shape memory alloy is heated to 60°C, restoring the pre-deformation.

[0110] Retractable hydraulic supports are installed on the side of the tunnel, and the support force (range 50-200kN) is automatically adjusted based on stress monitoring data. The support layout is optimized in real time through MATLAB.

[0111] As a specific embodiment, the feature importance ranking of the random forest algorithm in step S22 is: stress gradient (weight 0.45) > crack density (weight 0.35) > displacement rate (weight 0.20).

[0112] As a specific example, in step S32, the viscosity μ of the fracturing fluid is adjusted to 50 mPa·s by adding hydroxypropyl guar gum, and the shear rate is 100 s -1The stability is ≥90%. It can be concluded that adding an appropriate amount of hydroxypropyl guar gum can not only effectively adjust the viscosity of the fracturing fluid to achieve the required rheological properties, but also effectively adjust the viscosity of the fracturing fluid to achieve the desired rheological properties at a shear rate of 100s -1 When the fracturing fluid is heated, it can maintain a high degree of stability, ensuring that it can still maintain good operating performance under high flow rate conditions. This stability enables the fracturing fluid to play a better role in the formation, thereby improving the efficiency and safety of oil and gas production.

[0113] As a specific embodiment, the step S3 further includes optimizing the morphology of the fracturing cracks:

[0114] Directional perforation technology is used to arrange perforations in a spiral pattern with a spacing of 30 cm and a diameter of 8 mm in the borehole to guide the cracks to expand in the preset direction;

[0115] The microseismic event location data is analyzed using MATLAB image processing technology to generate a three-dimensional fracture morphology map in real time, and the perforation density is adjusted based on feedback.

[0116] As a specific embodiment, the step S5 further includes adaptive nanocoating thickness control:

[0117] Before spraying the carbon fiber reinforced polymer (CFRP) layer, a lidar was used to scan the curvature of the roadway surface and generate a 3D point cloud model;

[0118] Based on the curvature radius r and stress gradient The coating thickness t is dynamically calculated according to the following formula:

[0119]

[0120] A six-axis robotic arm equipped with a spray gun is used to achieve sub-millimeter precision coating construction; the smaller the curvature radius, the more significant the surface stress concentration, and the coating thickness needs to be increased to disperse the load; the elastic modulus E of carbon fiber reinforced polymer c And the coating thickness d satisfies:

[0121] E c =E m +k·d

[0122] Among them, E m represents the initial elastic modulus of the carbon fiber reinforced polymer; k represents the modulus enhancement factor per unit thickness of the carbon fiber reinforced polymer, which is positively correlated with the fiber volume fraction. As a specific example, k = 2 GPa / mm. By adjusting the thickness, the roadway's deformation resistance can be improved by 50%-80%.

[0123] As a specific embodiment, the method further includes the following steps:

[0124] S6. Real-time assessment of overburden stability driven by multi-source data:

[0125] S61: Based on the sensor data of step S1, construct the overburden stability index SI:

[0126]

[0127] Among them, σ avg is the mean stress; σ peak is the peak stress; δ max is the maximum displacement; L crit is the critical displacement threshold, which is calibrated through triaxial rock tests;

[0128] When S62 and SI≤0.6, the early warning system is triggered, mining is suspended and reinforcement measures are initiated. The overburden stability index SI comprehensively reflects the stress concentration and displacement accumulation effects. Through standardization, SI can effectively quantify the collapse risk. Critical displacement threshold L crit The calibration is based on the Mohr-Coulomb criterion and is determined in combination with laboratory rock sample compression tests.

[0129] Compared with the existing technology, the dynamic pillarless mining method for close-range coal seams based on intelligent monitoring provided by the present invention has the following beneficial effects:

[0130] 1. Precise control of overburden collapse: Real-time monitoring of overburden stress, displacement, and crack data through a multi-parameter sensor network is combined with a machine learning model to dynamically predict the time and range of overburden collapse, achieving an error rate of less than 5%. This significantly reduces the risk of local stress concentration caused by uneven overburden collapse and avoids the probability of strong mining pressure disasters by more than 70%.

[0131] 2. Significantly improved resource recovery rate: Dynamically optimize hydraulic fracturing parameters and mining sequence to ensure that overburden collapse completely covers the goaf, reducing coal pillars left behind. Resource recovery rate is increased to over 90%, an increase of 15% to 20% compared to traditional methods.

[0132] 3. Enhanced tunnel stability: Carbon fiber reinforced polymer (CFRP) nanocomposite materials are sprayed to reinforce the lower coal seam empty tunnels, increasing the compressive strength by 40%. Combined with the intelligent support wall, the support force can be adaptively adjusted to effectively suppress tunnel deformation, extend the service life, and reduce maintenance costs by more than 30%.

[0133] 4. Strong technical scalability: The system supports collaborative monitoring and control of multiple coal seams and is suitable for complex geological conditions (such as high-gas and soft rock formations). Through adaptive updates of model parameters, it can quickly adapt to different mine scenarios.

[0134] 5. The technical system of the present invention overcomes the data blind spots, control lag and structural instability problems of traditional methods, increases the resource recovery rate to more than 90%, and reduces the stress concentration factor to 1.2, providing a new paradigm for safe and efficient mining of close-range coal seams.

[0135] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the purpose and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A dynamic pillar-free mining method for close-range coal seams based on intelligent monitoring, characterized in that: The following steps are involved: S1. Multi-parameter sensor network deployment and data collection: S11. Deploy a multi-parameter sensor network in the working face of the upper and lower coal seams in close proximity to the coal seam and in the overburden, wherein the sensors include a fiber Bragg grating sensor for stress monitoring, a laser displacement sensor for displacement monitoring, and a three-dimensional acoustic emission sensor for crack detection; S12, upload the sensor data to the cloud platform in real time through the LoRa wireless transmission module, filter and de-noise the sensor data, and normalize it, and extract key characteristic parameters such as stress gradient, displacement rate, and crack density; S2. Construction of overburden collapse prediction model and optimization of top cutting parameters: S21. Establishment of theoretical model: Based on Griffith's crack propagation theory in rock mechanics, the following critical conditions for overburden collapse are established: Among them, σ1 and σ3 are the maximum and minimum principal stresses, K IC is the rock fracture toughness, α is the crack half-length; S22. Machine Learning Model Training: A fusion model of the random forest algorithm and long short-term memory network is constructed based on MATLAB. The input characteristic parameters include real-time stress gradient, displacement rate, and crack density. The output is the prediction of the time and range of overburden collapse. S23. Dynamic visualization and early warning: MATLAB is used to generate three-dimensional overburden stress cloud maps and crack expansion dynamics, displaying the overburden status in real time. When the predicted collapse time deviation is greater than 10%, an early warning is triggered. S24, dynamic adjustment of cutting parameters: The drilling spacing L is dynamically calculated based on the collapse range R: L=max(0.6R,2m) The blasting energy E is determined by the following formula: E=a·s eff ·V rock Where m represents the unit meter, V rock Indicates the volume of rock mass affected by blasting, in m 3 ; S3. Precision control of hydraulic fracturing and optimization of overburden collapse: S31. Drilling design in the top-cut area: Based on the model prediction results, radial drill holes are designed above the goaf of the upper coal seam, with the drilling depth being 80% of the overburden thickness; S32, Fracturing fluid formula: Use water-based fracturing fluid, add 0.5% nano-montmorillonite to enhance sand carrying capacity, and the sand ratio is 15%; S33. Dynamic control strategy of hydraulic fracturing parameters: Inject high-pressure water into the borehole at a pressure of 10-25 MPa and a flow rate of 20-50 L / min. Adjust the injection pressure according to the real-time fracture expansion speed v: Wherein, P0 is the initial water injection pressure; Fracturing termination conditions: overburden collapse height H ≥ 1.2 times the mining height, and collapse zone stress σ ≤ 3 MPa; S4. Dynamic planning of lower coal seam mining and stress field optimization: S41. Finite element modeling: A 3D geomechanical model was constructed based on COMSOL Multiphysics. The stress distribution data after mining of the upper coal seam was input, and the stress concentration coefficient of the lower coal seam was solved as follows: Among them, σ max represents the maximum stress, σ avg represents the mean stress; S42, mining priority rule: if K σ >1.5, marked as high-risk area, priority mining; if 1.2≤K σ <1.5, marked as medium-risk area, with secondary priority for mining; if K < K σ <1.2, marked as a safe area and mined last; where K is the lower limit of the threshold; S5. Nanocomposite reinforcement and intelligent support of gob-retained tunnels: S51. Material preparation: 2% graphene nanosheets are added to carbon fiber reinforced polymer to increase the tensile strength to 800 MPa; S52. Spraying process: Use high-pressure airless spraying equipment to spray carbon fiber reinforced polymer composite materials on the surface of the lower coal seam empty roadway. The spraying pressure is 20MPa, the nozzle diameter is 0.5mm, and the coating thickness t is adjusted according to the roadway curvature radius r as follows: Among them, mm represents the unit millimeter; S53, Smart Wall Design: The wall is composed of a shape memory alloy skeleton and ultra-high performance concrete, with the shape memory alloy accounting for 10%. A built-in fiber grating sensor monitors the strain ε. When ε ≥ 0.15%, the shape memory alloy is heated to 60°C, restoring the pre-deformation. Retractable hydraulic supports are installed on the side of the tunnel, the support force is automatically adjusted based on stress monitoring data, and the support layout is optimized in real time using MATLAB.

2. The method for dynamic pillar-free mining of close-range coal seams based on intelligent monitoring according to claim 1 is characterized in that: In step S11, fiber Bragg grating sensors are buried in key overburden layers, with one group arranged every 5 meters along the coal seam and arranged in layers every 2 meters in the vertical direction; laser displacement sensors are installed on the tunnel roof and both sides, with the center of the working surface as the reference point and a radial spacing of 10 meters; three-dimensional acoustic emission sensors are arranged around the goaf to capture rock formation crack expansion signals in real time.

3. The method for dynamic non-pillar mining of close-range coal seams based on intelligent monitoring according to claim 1 is characterized in that: In step S12, the sensor data is subjected to noise suppression and data fusion using a Kalman filter algorithm.

4. The method for dynamic pillar-free mining of close-range coal seams based on intelligent monitoring according to claim 1 is characterized in that: The order of importance of the features of the random forest algorithm in step S22 is: stress gradient > crack density > displacement rate.

5. The method for dynamic non-pillar mining of close-range coal seams based on intelligent monitoring according to claim 1 is characterized in that: In step S32, the viscosity μ of the fracturing fluid is adjusted to 50 mPa·s by adding hydroxypropyl guar gum, and the shear rate is 100 s -1 The stability is ≥90%.

6. The method for dynamic non-pillar mining of close-range coal seams based on intelligent monitoring according to claim 1 is characterized in that: The step S3 further includes optimizing the morphology of the fracturing cracks: Directional perforation technology is used to arrange perforations in a spiral pattern with a spacing of 30 cm and a diameter of 8 mm in the borehole to guide the cracks to expand in the preset direction; The microseismic event location data is analyzed using MATLAB image processing technology to generate a three-dimensional fracture morphology map in real time, and the perforation density is adjusted based on feedback.

7. The method for dynamic non-pillar mining of close-range coal seams based on intelligent monitoring according to claim 1 is characterized in that: In step S53, the ultra-high performance concrete compressive strength of the smart wall is ≥150 MPa, and the shape memory alloy restoring force is ≥200 kN / m.

8. The method for dynamic non-pillar mining of close-range coal seams based on intelligent monitoring according to claim 1 is characterized in that: The step S5 further includes adaptive nanocoating thickness control: Before spraying the carbon fiber reinforced polymer layer, the curvature of the roadway surface was scanned using LiDAR to generate a 3D point cloud model; Based on the curvature radius r and stress gradient The coating thickness t is dynamically calculated according to the following formula: A six-axis robotic arm equipped with a spray gun is used to achieve sub-millimeter precision coating construction; the smaller the curvature radius, the more significant the surface stress concentration, and the coating thickness needs to be increased to disperse the load; the elastic modulus E of carbon fiber reinforced polymer c And the coating thickness d satisfies: E c =E m +k·d Among them, E m represents the initial elastic modulus of carbon fiber reinforced polymer; k represents the modulus enhancement coefficient of carbon fiber reinforced polymer per unit thickness, which is positively correlated with the fiber volume fraction.

9. The method for dynamic non-pillar mining of close-range coal seams based on intelligent monitoring according to claim 1, characterized in that: The method further comprises the following steps: S6. Real-time assessment of overburden stability driven by multi-source data: S61: Based on the sensor data of step S1, construct the overburden stability index SI: Among them, σ avg is the mean stress; σ peak is the peak stress; δ max is the maximum displacement; L crit is the critical displacement threshold; When S62 and SI≤0.6, the early warning system is triggered, mining is suspended and reinforcement measures are initiated.

Images