Green pressure-relief method for mine pressure behavior caused by multi-seam mining residual coal pillar

Abstract

The application discloses a green pressure relief method for mine pressure appearance caused by multi-coal seam mining residual coal pillars and belongs to the technical field of multi-coal seam mining. The intelligent decision system of multi-source data fusion is used to solve the problem of response lag of the traditional method. The coupling process of supercritical CO2 gas expansion pressure relief and biodegradable hydraulic fracturing technology is adopted to realize low pollution and coal pillar structure protection in the pressure relief process. Through the whole-process closed loop of "monitoring-prediction-decision-verification", the problem of multi-coal seam stress superposition control is broken through, and the method is especially suitable for the complex geological environment of mine pressure appearance caused by multi-coal seam superimposed mining.

Description

Technical Field

[0001] This invention relates to the field of multi-coal seam mining technology, and in particular to a green decompression method for mine pressure manifestation caused by coal pillars left over from multi-coal seam mining. Background Technology

[0002] In the process of mining multiple coal seams, the mining of the lower coal seam will create a high stress concentration zone below the coal pillars left by the upper coal seam. These coal pillars, as key structures supporting the goaf, bear enormous pressure from the overlying strata. They often cause severe mine pressure manifestation problems due to stress concentration, leading to instability of the coal pillar itself, severe deformation of adjacent roadways, and even major safety accidents such as rockbursts, seriously threatening mine safety and the lives of personnel.

[0003] Currently, conventional control methods for mine pressure manifestations mainly include depressurization techniques such as blasting, borehole depressurization, and hydraulic fracturing. However, these methods have significant limitations: while blasting depressurization is effective, the strong shock waves generated can easily disturb the stability of the surrounding rock, and are accompanied by large amounts of dust and harmful gases, which contradicts the concept of green mining; the depressurization range and effect of borehole depressurization are difficult to control precisely, often requiring the construction of a large number of boreholes, resulting in low engineering efficiency, and causing irreversible damage to the integrity of the coal pillar, affecting resource recovery rate; hydraulic fracturing has poor controllability of fracture propagation trajectory under complex geological conditions, and traditional chemical fracturing fluids pose a potential risk of polluting the underground environment.

[0004] At the monitoring and decision-making level, existing methods mostly rely on single-point, static monitoring means (such as single-point stress gauges) and human experience judgment. This approach cannot construct a three-dimensional dynamic evolution model of the stress field, resulting in a serious lag in risk warning. The timing, location, and intensity of stress relief measures all depend on qualitative experience, lacking data-driven precise decision support.

[0005] Specifically, existing technical solutions have profoundly exposed the aforementioned shortcomings, for example:

[0006] The invention patent with publication number CN117404088A, entitled "A Coal Seam and Coal Pillar Pressure Relief Method and Device Based on Rock Mass Strength," constructs a buffer pressure relief zone. Its pressure relief parameters rely entirely on initial, static rock mechanics test results. The entire scheme lacks a dynamic adjustment mechanism based on real-time data, making adaptive optimization impossible. Furthermore, the controllable shock wave and high-pressure water-sand mixture it employs are essentially still strong disturbances to the surrounding rock and non-green operations. The invention patent with publication number CN114382483A, entitled "A Method for Preventing Strong Mine Pressure in Working Faces with Hard Roofs Under Residual Coal Pillars," although it performs a one-time mine pressure prediction through a theoretical model, its static model cannot respond to dynamic stress changes during mining, resulting in rigid and delayed early warnings. Its ultimate reliance on shock waves and pre-splitting blasting methods also poses strong disturbances and environmental pollution risks.

[0007] In summary, existing technologies generally suffer from the following systemic defects:

[0008] 1) The decompression process is highly destructive and environmentally unfriendly: Traditional decompression methods may damage the coal pillar structure or cause pollution, which does not conform to the concept of green mining.

[0009] 2) The control strategy is crude and lacks precision: the pressure relief parameters rely on static design or experience and fail to match the dynamic evolution of the stress field, thus failing to achieve "pressure relief on demand".

[0010] 3) Disconnect between monitoring and control, and low level of intelligent decision-making: The closed-loop intelligent control system of "real-time perception - intelligent prediction - accurate decision-making - dynamic regulation" has not been formed, and it is impossible to achieve early warning and proactive prevention and control of risks.

[0011] Therefore, there is an urgent need for a comprehensive approach that integrates high-performance monitoring, intelligent trend prediction, intelligent decision-making, and green stress relief methods to achieve precise, green, and intelligent management of coal pillar stress manifestations left over from multi-coal seam mining. Summary of the Invention

[0012] The purpose of this invention is to provide a green stress relief method for mine pressure manifestation caused by coal pillars left over from multi-coal seam mining, so as to solve the problems in the background art.

[0013] To achieve the above objectives, the present invention provides a green stress relief method for mine pressure manifestation caused by coal pillars left over from multi-coal seam mining, comprising the following steps:

[0014] S1. Deploy a sensor network in the remaining coal pillar and surrounding area and connect it to the central control system. The central control system receives stress, displacement and environmental data monitored by the sensor network in real time.

[0015] S2. Geological parameters are imported into the central control system and historical mining data are loaded. An LSTM neural network model is constructed. The historical mining data and stress, displacement and environmental data are integrated into time series data in chronological order. After removing outliers, the data is aggregated. The aggregated time series data is used as input, and the LSTM neural network model is used to predict stress changes in future periods.

[0016] S3. Use fuzzy clustering analysis to determine the risk level, match the corresponding depressurization and mining strategies, and perform depressurization operations on the coal body.

[0017] S4. After depressurization, monitor stress and displacement data for 24 hours. If the stress of the remaining coal pillar is stable at 16~18MPa and the displacement rate is... If the pressure drops to 0.05 mm / h, the risk is considered eliminated; otherwise, a second depressurization operation is performed.

[0018] S5. After the risk is eliminated, the stress and displacement rate data monitored in real time by the sensor network are used as input. The mining sequence is optimized and determined according to the genetic algorithm. During the mining process, the stress and displacement data are dynamically monitored. Steps S4-S5 are iterated according to the monitoring results until all coal seams are mined.

[0019] Preferably, in step S1, the sensor network is distributed in a three-dimensional grid and includes micro-vibration sensors, fiber optic stress sensors, three-dimensional laser displacement gauges, and temperature and humidity sensors. The micro-vibration sensors are used to monitor coal seam fracture signals, with a sampling frequency of 100Hz. A set of fiber optic stress sensors is embedded every 4-6m along the coal pillar to synchronously collect vertical / horizontal stress with an accuracy of ±0.1MPa. Two sets of three-dimensional laser displacement gauges are arranged at the top and bottom of the coal pillar to monitor three-dimensional displacement with an accuracy of ±0.1mm. One temperature and humidity sensor is deployed every 8-10m on the roadway wall for environmental data calibration.

[0020] Preferably, in step S2, the geological parameters include the thickness of each coal seam and the compressive strength of the coal body, and the historical mining data includes the coordinates of adjacent goaf areas and the size of the remaining coal pillars; the time series data are aggregated at time intervals of 3 to 6 minutes.

[0021] Preferably, in S3, the risk level classification standard is as follows:

[0022] Level I: Stress 15MPa, displacement velocity 0.05 mm / h, no acoustic emission signal of coal body fracturing;

[0023] Level II: 15MPa stress 20MPa, displacement velocity of 0.05~0.2mm / h, micro-seismic event frequency 3 times / min;

[0024] Level III: 20MPa stress 27MPa, displacement velocity 0.2~0.5mm / h, micro-vibration event frequency 3~8 times / min;

[0025] Level IV: Stress 27MPa, displacement velocity 0.5 mm / h, frequency of microseismic events 8 times / min or abrupt growth.

[0026] Preferably, in step S3, when the risk level is Level I, the mining operation proceeds according to the original plan;

[0027] When the risk level is Level II, the mining speed is simultaneously reduced to 80% of the original plan, and regional micro-decompression is initiated, with gas expansion decompression as the main decompression method. Boreholes are drilled in the stress concentration area, with the drilling direction perpendicular to the main stress direction of the coal pillar, to carry out gas expansion decompression. Then, directional hydraulic fracturing is used as a backup method. If the stress rises to 90% of the original peak value after 2 hours of decompression, directional hydraulic fracturing is initiated in the residual stress area at a pressure of 15~20MPa and a fracture length controlled at 3~5m. The purpose of this operation is to release stress locally and avoid escalation of risk. Stress is released in an environmentally friendly and low-interference manner, and low-intensity hydraulic fracturing is only initiated when the effect is insufficient.

[0028] When the risk level is Level III, coal seam mining is suspended. Directional hydraulic fracturing is used as the core decompression method at a pressure of 25-35 MPa. The principal stress direction is determined based on microseismic CT inversion, and the fracture propagation path is designed. The target fracture length is 1.2-1.5 times the width of the coal pillar. Gas expansion decompression is used as an auxiliary decompression method. After directional hydraulic fracturing, gas expansion decompression is applied to the residual stress zone at a flow rate of 10-15 m³ / h. 3 / min, pressure 3~5MPa, time 5 min; The purpose of this operation is to precisely and quickly cut off the stress transmission path through high-intensity hydraulic fracturing, rapidly reduce concentrated stress, and use gas for local fine-tuning to avoid response delay;

[0029] When the risk level is IV, plan the mining evacuation route, activate the emergency ventilation system, drill shallow holes (5-8m deep) in the disaster-prone area, perform directional hydraulic fracturing and pressure relief operations, and implement high-pressure hydraulic cutting at a pressure of 40-50MPa for a period of time. For 1 minute, the fracture length is controlled to be 0.5 times the width of the coal pillar. Grouting reinforcement is carried out immediately after hydraulic fracturing, with a grouting pressure of [missing information]. 8MPa; The purpose of this operation is to use hydraulic fracturing to prepare for emergency events, grouting to reinforce and maintain structural stability, strengthen the coal pillar in the early stages of a disaster, and ensure the safe evacuation of personnel and equipment.

[0030] Preferably, the borehole diameter is 45~55mm, the borehole depth is 12~18m, and the spacing is 2~8m;

[0031] The gas expansion and depressurization utilizes supercritical CO2 with a phase change temperature of 30-35°C and an initial flow rate of 20-30 m³ / h. 3 The flow rate is 5~10 MPa per minute. During the depressurization process, the flow rate is dynamically adjusted according to the acoustic emission of the micro-vibration sensor. When the acoustic emission frequency is greater than 3 times / min, the dynamic flow rate is adjusted to 12~18 m³ / min. 3 / min; the pressure relief termination condition is that the stress reaches the target value, and the displacement rate... 0.05 mm / h, no rebound after 30 minutes of continuous monitoring.

[0032] Preferably, the directional hydraulic fracturing depressurization uses high-pressure water and a biodegradable solution containing 1.5-2.5% nanocellulose as the fracturing fluid, with a displacement of 0.5-1.2 m³ / h. 3 / min, the termination condition is that the crack extends to the target length and the peak stress of the coal pillar decreases to the safe threshold, displacement rate 0.1 mm / h, with no abnormal fluctuations for 1 hour.

[0033] During directional hydraulic fracturing, real-time monitoring of fracture propagation and error is crucial. If the deviation from the design path exceeds 5%, immediately stop the pump and adjust the drilling angle.

[0034] Preferably, the grouting reinforcement operation involves injecting a bio-based grouting material, which is a modified straw fiber grout with a water-cement ratio of 0.6:1 and an initial setting time of [missing information]. Degradation rate within 10 minutes and 1 year 10%, the grouting termination condition is: displacement rate 0.05mm / h or grouting volume reaching 2.5m 3 After 72 hours of grouting, core samples were taken to test whether the coal body bonding strength had improved.

[0035] Preferably, in step S5, the specific steps for using the genetic algorithm to optimize and determine the mining order are as follows:

[0036] 1) Use the risk level determination results of S3, the stress and displacement data after decompression in S4, geological parameters, and mining constraints as inputs;

[0037] 2) Encode the mining sequence as a chromosome, and determine the fitness function as follows:

[0038] ;

[0039] in, , These are the weighting coefficients;

[0040] 3) Perform selection, crossover, and mutation operations to output the optimal solution.

[0041] Preferably, in S5, the condition for risk relief is: the central control system continuously monitors for 24 hours and confirms that the stress of the remaining coal pillar is stable at 16~18MPa and the displacement rate is stable.

[0042] Therefore, the green stress relief method of the present invention for mine pressure manifestation caused by coal pillars left in multi-coal seam mining has the following beneficial effects:

[0043] (1) This invention uses LSTM neural network to fuse multi-source data (stress, displacement, environment) to achieve high-precision prediction of stress changes in future time periods. Combined with fuzzy clustering hierarchical (Level I-IV) dynamic matching stress relief strategy, the response speed is improved by 70% and the risk misjudgment rate is reduced by 85% compared with traditional experience judgment.

[0044] (2) In this invention, green and low-carbon operations are carried out by using supercritical CO2 gas expansion decompression and biodegradable hydraulic fracturing technology. No explosion is required, and the pressure release process can be precisely controlled. At low risk (Level II), the gas expansion is mainly controlled by fine regulation, taking into account both environmental protection and safety. At medium and high risk (Level III and IV), the hydraulic fracturing intensity is gradually upgraded. Combined with grouting and emergency measures, a multi-technology linkage disaster suppression system is formed, which not only reduces the environmental impact, but also ensures the stability of the decompression process. Dust emissions are reduced by 90%. The bio-based grouting material (modified straw fiber) has a degradation rate of <10% within 1 year, avoiding the long-term pollution risk of traditional chemical materials.

[0045] (3) In this invention, a closed loop of “monitoring-prediction-decision-execution-verification” is constructed. Based on traditional layered mining, the pressure relief strategy and mining sequence are dynamically determined according to monitoring data to avoid excessive stress concentration caused by mining multiple coal seams at the same time. It can effectively control the occurrence of mine pressure manifestation and has strong field applicability. It can adjust the pressure relief strategy in real time in complex mine environments to ensure safe and efficient mine mining.

[0046] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0047] Figure 1 This is a flowchart of an embodiment of the present invention. Detailed Implementation

[0048] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0049] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0050] Example

[0051] A mining area is undergoing superimposed mining of three coal seams (upper #1 coal seam, 2.5m thick; middle #2 coal seam, 3.2m thick; lower #3 coal seam, 4.0m thick), at a depth of 550m. The coal seam compressive strength is 15MPa. The remaining coal pillar of #2 coal seam measures 8m × 3.2m × 120m. A stress concentration zone has formed in the adjacent goaf, requiring green stress relief measures, such as... Figure 1 As shown, the specific steps are as follows:

[0052] S1. Deploy a sensor network in the remaining coal pillar and surrounding area and connect it to the central control system. The central control system receives stress, displacement, and environmental data monitored by the sensor network in real time. Specifically:

[0053] 1) Sensor deployment:

[0054] Within a 50m×50m×30m area surrounding the coal pillar remaining at #2 coal seam, a sensor network is deployed in a 10m×10m×5m three-dimensional grid.

[0055] Microseismic sensor (model MS-2000): 100Hz sampling frequency, 25 sensors in total, for monitoring coal seam fracture signals;

[0056] Fiber Bragg grating stress sensor: One set is embedded every 5m along the coal column (24 sets in total), with an accuracy of ±0.1MPa, to collect vertical and horizontal stress data in real time;

[0057] Three-dimensional laser displacement gauge: 2 sets each for the top and bottom plates (4 sets in total), with an accuracy of ±0.1mm, to monitor displacement changes;

[0058] Temperature and humidity sensors: 1 sensor is installed every 10 meters on the tunnel wall (10 sensors in total) to monitor the ambient temperature and humidity.

[0059] 2) Data initialization:

[0060] The central control system imports geological parameters: upper #1 coal seam, 2.5m thick; middle #2 coal seam, 3.2m thick; lower #3 coal seam, 4.0m thick), burial depth 550m, coal body compressive strength 15MPa. Historical mining data is loaded: location of adjacent goaf areas, dimensions of remaining coal pillars (8m wide, 3.2m high, 120m long), sensor calibration completed within 30 minutes of sensor activation, data transmitted at a frequency of 1 time / second.

[0061] S2. Geological parameters are imported into the central control system, and historical mining data is loaded. An LSTM neural network model is constructed, and the historical mining data is integrated with stress, displacement, and environmental data in chronological order to form time-series data. After outliers are removed, the data is aggregated. The aggregated time-series data is used as input, and the LSTM neural network model is used to predict the stress change trend in future periods. Specifically:

[0062] 1) Data preprocessing:

[0063] The real-time monitored stress, displacement, temperature and humidity data are integrated in chronological order, outliers are removed using the 3σ principle, and the data is aggregated at 5-minute intervals.

[0064] 2) Model training and prediction:

[0065] An improved LSTM neural network model is used in this embodiment. The LSTM neural network includes one input layer, two hidden layers, and one output layer. The number of neurons in the hidden layers is 128. The training iterations (epochs) are 1000, the batch size is 32, and the Adam optimizer is used. The model input consists of aggregated stress, displacement, and temperature / humidity time-series data from the past hour, and the output is the stress prediction value for the next 6 hours. Other specific parameters are not limited and can be adjusted according to actual conditions. The historical mining data for one month and the corresponding stress change data are used as the training set to input and train the LSTM neural network.

[0066] Based on the current data, the current peak stress is 25 MPa. The model predicts that a stress concentration zone will appear in the middle of the coal pillar within 6 hours, with the predicted value rising to 28 MPa (confidence level 92%).

[0067] S3. Use fuzzy clustering analysis to determine the risk level, match the corresponding decompression and mining strategies, and perform decompression operations on the coal body, specifically:

[0068] 1) Risk assessment:

[0069] Based on fuzzy clustering analysis, the current stress is 25 MPa and the displacement rate is 0.3 mm / h, which is classified as Level III risk (20 MPa ≤ stress < 27 MPa, displacement 0.2~0.5 mm / h).

[0070] 2) The central control system outputs the following instructions:

[0071] Mining suspended: Mining of coal seam #3 will be delayed at intervals of 24 hours.

[0072] Using directional hydraulic fracturing as the core pressure relief method, a 50mm diameter borehole with a depth of 15m was drilled in the stress concentration zone of the coal pillar, with a pressure set at 30MPa. The principal stress direction was determined based on microseismic CT inversion, and the designed fracture extension path was along the diagonal direction of the coal pillar, with a target fracture length of 10m. A biodegradable fluid with 2% nanocellulose was used as the fracturing fluid, with an initial displacement of 1m³. 3 The crack length is adjusted in real time according to the crack extension; crack extension is monitored in real time through microseismic CT inversion to ensure crack length error. 5%, when the peak stress decreases to 18 MPa, the displacement rate Stop fracturing when the fracturing rate reaches 0.1 mm / h and remains stable for more than 1 hour.

[0073] After hydraulic fracturing, the residual stress zone ( A supplementary gas expansion and depressurization process (15 MPa) is used for localized fine-tuning to prevent secondary stress concentration; supercritical CO2 gas is used, with a phase change temperature of 31.1℃, a pressure of 7.38 MPa, and an injection flow rate of 10~15 m³ / s. 3 / min, pressure 5MPa, duration 4min, when stress fluctuation Once the displacement rate reaches 5% and meets the standard, stop depressurizing.

[0074] In this embodiment, to ensure absolute safety during downhole operations and prevent CO2 leaks from causing personnel asphyxiation or other risks, this system implements the following strict safety control protocols: environmental safety thresholds and real-time monitoring:

[0075] Leakage concentration threshold: Multiple CO2 concentration sensors are deployed in the injection area and return airway. The system is configured with two alarm thresholds:

[0076] Level 1 warning (0.5% vol): When the monitored concentration reaches 0.5%, the central control system issues an audible and visual warning to alert operators to changes in airflow and automatically checks the sealing of the injection pipeline.

[0077] Level 2 Alarm and Automatic Response (1.5% vol): When the concentration reaches 1.5%, the system immediately triggers the automatic emergency procedure:

[0078] ① Automatically shut off the CO2 injection pump and close all solenoid valves in the injection pipeline.

[0079] ② The standby booster fan will automatically start, increasing the ventilation volume at the accident point to three times the normal level, forming a strong negative pressure drainage.

[0080] ③ Broadcast evacuation instructions to all areas.

[0081] The security control protocol described above is merely an example. Specific technical solutions can be modified according to actual circumstances, or technical substitutions can be made using conventional methods in the field. Specific security design schemes for the injection system are not limited in detail in this embodiment; conventional methods in the field can be employed.

[0082] S4. After depressurization, monitor stress and displacement data for 24 hours. If the coal pillar stress is stable at 16~18MPa and the displacement rate is... If the pressure drops to 0.05 mm / h, the risk is considered eliminated; otherwise, a second depressurization operation is performed, specifically as follows:

[0083] After 24 hours of monitoring, the stress dropped to 18 MPa after depressurization, and the displacement rate was 0.1 mm / h, which met the risk relief conditions (stress 16~18 MPa, displacement rate <0.05 mm / h), and proceeded to step S5; the depressurization data (stress curve, grouting parameters, mining sequence) were stored in the database of the central control system to optimize the LSTM neural network parameters and improve the accuracy of future predictions.

[0084] S5. After the risk is eliminated, the stress and displacement data monitored in real time by the sensor network are used as input. The mining sequence is optimized and determined according to the genetic algorithm. During the mining process, stress and displacement data are dynamically monitored. Steps S4-S5 are iterated based on the monitoring results until all coal seams are mined. Specifically:

[0085] Mining sequence optimization:

[0086] Input: Level III risk assessment result, post-decompression stress 18MPa, displacement 0.1mm / h, geological parameters and stress relaxation period T c ≥48h constraint;

[0087] Coding: The mining sequence (#1 coal, #3 coal) is encoded as a chromosome;

[0088] Fitness function: F = 0.7 × stress balance + 0.3 × mining efficiency; where stress balance is specifically quantified as the reciprocal of the variance of stress at all measuring points in the monitoring area, mining efficiency is quantified as the ratio of actual daily advance speed to standard speed (5m / d), and weighting coefficients are set based on experience or historical data.

[0089] Output: Prioritize mining #1 coal, with an advance speed of 3m / d (originally 5m / d).

[0090] The mining sequence has been replanned, with mining of the lower #3 coal seam suspended and the upper #1 coal face prioritized for advancement, at a reduced advance rate from 5 m / d to 3 m / d; the mining interval between adjacent coal seams must be ensured. Critical stress relaxation period (T) c =48h).

[0091] Dynamic monitoring and feedback:

[0092] The stress distribution model is updated every 2 hours during mining. If the stress of the coal pillar rises to 22MPa, the pressure relief operation is automatically triggered. If the system continues to monitor for 24 hours, it confirms that the stress of the remaining coal pillar is stable at 16~18MPa and the displacement rate is [not specified].

[0093] The implementation effect test was carried out in the mining area, and its safety indicators were: stress fluctuation controlled within 16~22MPa, and displacement rate stable <0.05mm / h;

[0094] Green indicators: CO2 utilization rate of 95%, dust emission reduction of 90%, and grouting material degradation rate of 9% in 1 year;

[0095] Efficiency indicators: resource recovery rate increased by 15%, mining cycle shortened by 28%, human intervention rate reduced by 70%, and mine pressure control time shortened by 40%.

[0096] Therefore, this invention provides a green decompression method for mine pressure manifestation caused by coal pillars left over from multi-coal seam mining. Through an intelligent decision-making system based on multi-source data fusion, it solves the problem of lag in response of traditional methods. It adopts a coupled process of supercritical CO2 gas expansion decompression and biodegradation hydraulic fracturing technology to achieve low pollution in the decompression process and protection of the coal pillar structure. Through a closed-loop process of "monitoring-prediction-decision-verification", it overcomes the problem of stress superposition control in multi-coal seams, and is especially suitable for complex geological environments where mine pressure manifestation is caused by multi-coal seam mining.

[0097] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A green stress relief method for mine pressure manifestation caused by coal pillars left over from multi-coal seam mining, characterized in that, Includes the following steps: S1. Deploy a sensor network in the remaining coal pillar and surrounding area and connect it to the central control system. The central control system receives stress, displacement and environmental data monitored by the sensor network in real time. S2. Geological parameters are imported into the central control system and historical mining data are loaded. An LSTM neural network model is constructed. The historical mining data and stress, displacement and environmental data are integrated into time series data in chronological order. After removing outliers, the data is aggregated. The aggregated time series data is used as input, and the LSTM neural network model is used to predict the stress change trend in future periods. S3. Use fuzzy clustering analysis to determine the risk level, match the corresponding decompression and mining strategies, and perform decompression operations on the coal seam; the risk level classification criteria are as follows: Level I: Stress 15MPa, displacement velocity 0.05 mm / h, no acoustic emission signal of coal body fracturing; Level II: 15MPa stress 20MPa, displacement velocity of 0.05~0.2mm / h, micro-seismic event frequency 3 times / min; Level III: 20MPa stress 27MPa, displacement velocity 0.2~0.5mm / h, micro-vibration event frequency 3~8 times / min; Level IV: Stress 27MPa, displacement velocity 0.5 mm / h, frequency of microseismic events 8 times / min or abrupt growth; When the risk level is Level I, mining operations shall proceed as originally planned. When the risk level is Level II, the mining speed is simultaneously reduced to 80% of the original plan, and regional micro-decompression is initiated, with gas expansion decompression as the main decompression method. Boreholes are drilled in the stress concentration area, with the drilling direction perpendicular to the main stress direction of the coal pillar, to carry out gas expansion decompression. Then, directional hydraulic fracturing is used as a backup method. If the stress rises to 90% of the original peak value after 2 hours of decompression, directional hydraulic fracturing is initiated in the residual stress area, with a pressure of 15~20MPa and a fracture length controlled at 3~5m. When the risk level is Level III, coal seam mining is suspended. Directional hydraulic fracturing is used as the core decompression method at a pressure of 25-35 MPa. The principal stress direction is determined based on microseismic CT inversion, and the fracture propagation path is designed. The target fracture length is 1.2-1.5 times the width of the coal pillar. Gas expansion decompression is used as an auxiliary decompression method. After directional hydraulic fracturing, gas expansion decompression is applied to the residual stress zone at a flow rate of 10-15 m³ / h. 3 / min, pressure 3~5MPa, time 5 min; When the risk level is IV, plan the mining evacuation route, activate the emergency ventilation system, drill shallow holes (5-8m deep) in the disaster-prone area, perform directional hydraulic fracturing and pressure relief operations, and implement high-pressure hydraulic cutting at a pressure of 40-50MPa for a period of time. For 1 minute, the fracture length is controlled to be 0.5 times the width of the coal pillar. Grouting reinforcement is carried out immediately after hydraulic fracturing, with a grouting pressure of [missing information]. 8MPa; The directional hydraulic fracturing pressure relief uses a biodegradable fluid containing 1.5-2.5% nanocellulose as the fracturing fluid, with a displacement of 0.5-1.2 m³ / h. 3 / min, the termination condition is that the crack extends to the target length and the peak stress of the coal pillar decreases to the safe threshold, displacement rate 0.1 mm / h, with no abnormal fluctuations for 1 hour; During directional hydraulic fracturing, real-time monitoring of fracture propagation and error is crucial. If the deviation from the design path exceeds 5%, immediately stop the pump and adjust the drilling angle. The grouting reinforcement operation involves injecting a bio-based grouting material, specifically a modified straw fiber grout. The grouting is terminated based on the displacement rate. 0.05mm / h or grouting volume reaching 2.5m 3 ; S4. After depressurization, monitor stress and displacement data for 24 hours. If the stress of the remaining coal pillar is stable at 16~18MPa and the displacement rate is... If the pressure drops to 0.05 mm / h, the risk is considered eliminated; otherwise, a second depressurization operation is performed. S5. After the risk is eliminated, the stress and displacement rate data monitored in real time by the sensor network are used as input. The mining sequence is optimized and determined according to the genetic algorithm. During the mining process, the stress and displacement data are dynamically monitored. Steps S4-S5 are iterated according to the monitoring results until all coal seams are mined.

2. The green stress relief method for mine pressure manifestation caused by coal pillars left in multi-coal seam mining according to claim 1, characterized in that: In S1, the sensor network is distributed in a three-dimensional grid and includes micro-vibration sensors, fiber optic stress sensors, three-dimensional laser displacement gauges, and temperature and humidity sensors. Among them, the micro-vibration sensors are used to monitor coal fracture signals, with a sampling frequency of 100Hz; a set of fiber optic stress sensors is embedded every 4-6m along the coal pillar to synchronously collect vertical / horizontal stress with an accuracy of ±0.1MPa; two sets of three-dimensional laser displacement gauges are arranged at the top and bottom of the coal pillar to monitor three-dimensional displacement with an accuracy of ±0.1mm; and one temperature and humidity sensor is deployed every 8-10m on the roadway wall for environmental data calibration.

3. A green stress relief method for mine pressure manifestation caused by coal pillars left in multi-coal seam mining according to claim 1, characterized in that: In S2, the geological parameters include the thickness of each coal seam and the compressive strength of the coal body. The historical mining data includes the coordinates of adjacent goaf areas and the size of the remaining coal pillars. The time series data is aggregated at time intervals of 3 to 6 minutes.

4. A green stress relief method for mine pressure manifestation caused by coal pillars left in multi-coal seam mining according to claim 3, characterized in that: The borehole diameter is 45~55mm, the borehole depth is 12~18m, and the spacing is 2~8m; The gas expansion and depressurization utilizes supercritical CO2 with a phase change temperature of 30-35°C and an initial flow rate of 20-30 m³ / h. 3 The flow rate is 5~10 MPa per minute. During the depressurization process, the flow rate is dynamically adjusted according to the acoustic emission of the micro-vibration sensor. When the acoustic emission frequency is greater than 3 times / min, the dynamic flow rate is adjusted to 12~18 m³ / min. 3 / min; the pressure relief termination condition is that the stress reaches the target value, and the displacement rate... 0.05 mm / h, no rebound after 30 minutes of continuous monitoring.

5. A green stress relief method for mine pressure manifestation caused by coal pillars left in multi-coal seam mining according to claim 4, characterized in that: The modified straw fiber grout has a water-cement ratio of 0.6:1 and an initial setting time of [missing information]. 10 minutes.

Images