Methods and systems for prevention and intelligent support of high-stress disasters in deep mines

Abstract

This invention discloses a method and system for preventing and controlling high-stress disasters in deep mines and for intelligent support, belonging to the field of rock mass control technology in mining. The method and system adopt an integrated collaborative architecture of active pressure relief and intelligent support. First, fracture-inducing holes are constructed in the stress concentration zone, employing a step-by-step air-water pressure relief process: compressed air is injected to create fractures, gently releasing some elastic potential energy; then high-pressure water is injected to expand the fractures, achieving directional fracture penetration. Simultaneously, sensor anchors with built-in distributed optical fibers are used to collect multi-dimensional deformation data of the rock mass in real time and input into an LSTM prediction model to predict future support load trends. Finally, the system automatically adjusts the support force of the electro-hydraulic controlled hydraulic support according to the prediction commands. This invention effectively overcomes the shortcomings of traditional single-stage strong pressure relief methods that easily lead to surrounding rock instability and the serious lag in passive support, achieving precise prevention and dynamic adaptive control of high-stress disasters in deep mines.

Description

Technical Field

[0001] This invention relates to the field of rock mass control in mining, and in particular to a method and system for preventing and controlling high-stress disasters and providing intelligent support in deep mines. Background Technology

[0002] With the increasing scarcity of shallow mineral resources, mining operations both domestically and internationally are gradually extending to deeper depths, with some mines reaching depths of 2000-2500m or even over 3000m. Deep rock masses experience extremely high geostress, resulting in rock conditions completely different from those in shallower areas. As depth increases, they are highly susceptible to sudden mining disasters such as tunnel deformation, rock bursts, and collapses.

[0003] Most existing support technologies are traditional passive support (such as surrounding rock support technology, anchor injection support, etc.), which have the following defects: First, traditional support is limited by the limitation of one-time support and fixed parameters, and cannot respond to the complex and ever-changing dynamic rock environment. Traditional support systems are prone to damage and failure in just about a month. Second, existing stress relief technologies are still mostly in the stage of manual operation. Single stress relief (such as blasting or single-step hydraulic fracturing) is prone to rock mass vibration and instability problems due to the instantaneous release of energy. Moreover, the stress relief range is difficult to control precisely. Summary of the Invention

[0004] This invention provides a method and system for preventing and controlling high-stress disasters and for intelligent support in deep mines, solving the technical problems of rock mass control and the lag of traditional support in high-stress environments in deep mines.

[0005] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows: a method for prevention and intelligent support of high-stress disasters in deep mines, comprising the following steps:

[0006] S1. Preliminary preparation: Detect the rock mass around the deep tunnel to determine the stress concentration points, and construct fracture holes at the stress concentration points;

[0007] S2, Step-by-step decompression-compressed air fracturing: Compressed air is injected into the fracturing holes, allowing the gas to penetrate along the original weak surface inside the rock mass and form an initial fracture network, so as to release part of the elastic potential energy of the rock mass;

[0008] S3, Step-by-step pressure relief-hydraulic fracturing: After the compressed air fracturing is completed, high-pressure water is injected into the fracturing hole. The high-pressure water seeps along the initial fracture network to produce a splitting effect, thereby achieving directional expansion and connection of fractures.

[0009] S4. Selection and deployment of fiber optic sensing anchor bolts: By deploying sensing anchor bolts with built-in distributed optical fibers around the roadway in the pressure relief area, the strain data of the rock mass is collected in real time and converted into the rock mass deformation rate and cumulative deformation.

[0010] S5. Algorithm Platform and Hydraulic Support Control: The transformed rock mass deformation data is input into a pre-trained LSTM deep learning prediction model to predict the trend of support load changes within a set time period. Based on the prediction results, control commands are sent to the electro-hydraulic control system of the hydraulic support to automatically adjust the hydraulic supply pressure of the column to dynamically adjust the support force.

[0011] Furthermore, in step S1, the construction parameters for the fracture-inducing holes are: hole depth set to 8-15m, hole diameter set to 90-110mm, and hole spacing set to 3-5m.

[0012] Furthermore, in step S2, the specific control conditions for compressed air fracturing are as follows: the injection pressure is set according to the rock mass hardness: the pressure is set at 15-25 MPa for soft rock, 25-35 MPa for medium-hard rock, and 35-45 MPa for hard rock; the fracturing pressure holding time is 1-2 minutes per meter of hole depth.

[0013] Furthermore, during the fracturing process, the orifice pressure is monitored in real time. If the pressure drop is greater than 5 MPa, it is determined that the fracture has been completed, and the fracturing is terminated in advance. If the pressure continues to rise and there is no downward trend, the pressure is reduced by 10% to 15%, and the pressure stabilization time after the pressure reduction is extended by 50%.

[0014] Furthermore, in step S3, the specific control conditions for hydraulic fracturing are as follows: high-pressure water injection is performed within 12 hours after compressed air fracturing is completed; the set high-pressure water pressure is 10-15 MPa higher than the compressed air fracturing pressure, initially injected at a low speed of 5-8 L / min, and increased to 10-15 L / min after the pressure stabilizes; when the injection volume reaches 3-5 times the volume of the fracturing hole, or when the pressure shows 2-3 periodic fluctuations, it is determined that the fracture has been fully expanded and water injection is stopped.

[0015] Furthermore, in step S5, the logic of advance prediction and adaptive adjustment is as follows: when the rock mass deformation rate is ≥0.3mm / h, the LSTM model predicts that the support load will increase by 20% to 30% in the next 1 to 2 hours; if the predicted load increases by more than 10%, an instruction to increase the support force by 5% to 15% is sent to the hydraulic support; if the predicted load decreases by more than 10%, an instruction to decrease the support force by 5% to 10% is sent; the model calls the pressure-time curve data of the decompression stage in real time, and if the stress recovery is >10% within 1 hour after decompression, the support load prediction threshold is automatically lowered by 15%, triggering the support force increase instruction in advance.

[0016] Another objective of this invention is to provide a support system for a method of preventing and intelligently supporting high-stress disasters in deep mines, the system comprising:

[0017] The system comprises a step-by-step pressure relief subsystem and an intelligent support subsystem. The step-by-step pressure relief subsystem includes a compressed air device, a high-pressure water injection device, and a borehole sealing device. The compressed air device is used to inject compressed air into the fracture-inducing borehole for initial fracture initiation. The high-pressure water injection device is used to inject high-pressure water into the borehole after the initial fracture initiation for secondary fracture expansion and fracturing.

[0018] The intelligent support subsystem includes fiber optic sensing anchor bolts, an AI algorithm platform, and electro-hydraulic hydraulic supports. The fiber optic sensing anchor bolts are deployed around the tunnel to collect multi-dimensional physical quantity data of the rock mass in real time and transmit it to the AI ​​algorithm platform. The AI ​​algorithm platform has a built-in LSTM prediction model to analyze the received data and output the predicted support load and control commands. The electro-hydraulic hydraulic supports are communicatively connected to the AI ​​algorithm platform to receive the control commands and automatically adjust the support force.

[0019] Furthermore, the compressed air equipment is a screw air compressor with a rated pressure of 30-50 MPa; the high-pressure water injection equipment is a high-pressure plunger pump with a rated pressure of 50-80 MPa.

[0020] Furthermore, the fiber optic sensing anchor is an intrinsically safe anchor for mining with built-in distributed optical fiber, and the rod body is made of 20MnSi steel. Its sensing accuracy is ±2με. The deployment spacing is 3-5m in the direction of the roadway, 2-3m in the direction of the roadway dip, and the spacing is increased to 1.5-2m in the stress concentration area.

[0021] The beneficial effects of this invention are as follows:

[0022] 1. A new step-by-step collaborative decompression process is proposed, which involves first creating fractures with compressed air and then expanding them with high-pressure hydraulic fracturing. Through the complementary advantages and orderly action of the air-water medium, the elastic potential energy of the rock mass is released in a stable and controllable manner, which significantly reduces the risk of disasters caused by high ground stress from the source.

[0023] 2. An intelligent support system was constructed that integrates distributed optical fiber sensing for real-time monitoring, LSTM deep learning algorithm for intelligent prediction, and electro-hydraulic control system for precise execution, realizing a fundamental shift in support mode from passive response to active prediction and dynamic control. Attached Figure Description

[0024] Figure 1 This is a schematic diagram illustrating the compressed air-induced cracking principle in an embodiment of the present invention;

[0025] Figure 2 This is a schematic diagram of the hydraulic fracturing principle in an embodiment of the present invention;

[0026] Figure 3 This is a flowchart of the step-by-step depressurization operation in an embodiment of the present invention;

[0027] Figure 4 This is an operation flowchart of the intelligent support system in an embodiment of the present invention;

[0028] Figure 5 This is a logic control diagram of the intelligent support system in an embodiment of the present invention. Detailed Implementation

[0029] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings. Identical components are indicated by the same reference numerals.

[0030] It should be noted that the terms front, back, left, right, up, and down used in the following description refer to the directions in the attached diagram, while the terms inside and outside refer to the directions toward or away from the geometric center of a specific component.

[0031] To make the content of this invention easier to understand, the technical solutions in the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings.

[0032] In the field of deep resource extraction, rock mass control under high stress environments is a core challenge for ensuring production safety and efficiency. Traditional stress relief and support technologies often fail to accurately match dynamic changes in the rock mass and lead to uncontrolled energy release, resulting in safety hazards such as tunnel deformation and roof collapse. Therefore, a collaborative technology system based on precise stress relief and intelligent support has emerged. Its core working principle can be divided into two parts: a step-by-step stress relief principle and an intelligent support system principle. Through technological collaboration, it achieves safe and controllable extraction of deep rock masses.

[0033] like Figures 1 to 2 As shown, this application provides a method for prevention and intelligent support of high-stress disasters in deep mines. The core method is to first create fractures using air fracturing, followed by hydraulic fracturing to enlarge the fractures. By precisely controlling construction parameters, controllable stress relief of deep high-stress rock masses is achieved. The specific scheme is as follows:

[0034] Step-by-step pressure relief principle:

[0035] The step-by-step decompression technology is a core process designed to address the imbalance between energy accumulation and release in deep, high-stress rock masses. Through the synergistic effect of compressed air fracturing and hydraulic fracturing, it completes the stress release and fracture control of the rock mass in stages, fundamentally solving the drawbacks of excessive energy impact and uncontrollable decompression range in traditional single-stage decompression technology.

[0036] S1. Preliminary preparation: Detect the rock mass around the deep tunnel to determine the stress concentration points, and construct fracture holes at the stress concentration points.

[0037] Specifically, before implementing the process, a comprehensive inspection of the rock mass surrounding the deep roadway must be conducted using specialized stress monitoring equipment (such as borehole stress gauges and rock mass stress sensors) to accurately determine the stress concentration points. These stress concentration points are typically distributed in high-stress areas at the roadway face, sides, and roof. Based on the stress values ​​and distribution range, key stress relief areas need to be delineated to provide precise location data for subsequent drilling and fracturing processes, avoiding poor stress relief or resource waste caused by blind construction.

[0038] S2, Step-by-step decompression-compressed air fracturing: Compressed air is injected into the fracturing holes, allowing the gas to penetrate along the original weak surfaces inside the rock mass and form an initial fracture network, thereby releasing some of the elastic potential energy of the rock mass.

[0039] Specifically, after locating the stress concentration points, the technical team will drill holes within the stress concentration areas of deep roadways. After drilling, the team will first inject compressed air at a stable pressure of 30–50 MPa into the pre-drilled rock ducts (e.g., through a dedicated high-pressure pipeline). Figure 1 Compared to traditional explosive blasting, compressed air releases energy more gently and controllably. It slowly penetrates along the original weak points (such as bedding and joints) within the rock mass, gradually opening up tiny fractures and forming an initial fracture network. The core objective of this process is to create pathways for subsequent hydraulic fracturing. By pre-breaking the local integrity of the rock mass, the resistance to subsequent high-pressure water injection is reduced, avoiding concentrated energy impacts caused by excessively hard rock. Simultaneously, the formation of the initial fracture zone can release some of the rock mass's elastic potential energy in advance, reducing the risk of stress abrupt changes in subsequent processes.

[0040] S3. Step-by-step depressurization - hydraulic fracturing: After the compressed air fracturing is completed, high-pressure water is injected into the fracturing hole. The high-pressure water seeps along the initial fracture network to produce a splitting effect, realizing the directional expansion and connection of fractures.

[0041] Specifically, after entering the hydraulic fracturing propagation stage, the system switches to high-pressure water injection mode, continuously delivering high-pressure water at a pressure of 50-80 MPa to the initial fracture zone. At this time, the high-pressure water seeps along the pre-set fracture network and generates a fracturing effect, achieving directional propagation and connection of fractures (e.g., Figure 2During this process, the elastic potential energy accumulated in the rock mass is slowly released through the expansion of fractures. Pressure monitoring data shows that this technology can reduce the stress value of the rock mass in high-stress areas by 30% to 50% (data from field tests at the Zhangxiaolou Mine in Xuzhou, with a test section depth of 1050m and a rock mass compressive strength of 85MPa). Furthermore, the stress release rate can be precisely controlled by adjusting the water pressure. In addition, hydraulic fracturing can precisely define the pressure relief range by controlling the injection time and pressure. For example, in coal mining, the pressure relief range can be controlled within 5 to 8 meters around the roadway based on the working face advance speed, ensuring both effective pressure relief and avoiding excessive damage to the surrounding rock stability.

[0042] This step-by-step strategy of pre-fracture followed by propagation is essentially a refined management of the rock mass energy release process. By controlling the energy input intensity and range in two steps, it can not only effectively avoid instability problems such as rock mass vibration and fragment ejection caused by a single violent decompression, but also flexibly adjust process parameters according to different rock mass hardness and stress distribution characteristics. It is applicable to various deep resource mining scenarios such as metal mines and coal mines, providing core technical support for safe mining under high stress environments.

[0043] Principle of intelligent support system:

[0044] Intelligent support systems are proactive prevention and control technologies for dynamic deformation and stress fluctuation problems in deep rock masses. They break through the limitations of traditional passive support (such as anchor bolts and masonry arches) which rely on one-time support and fixed parameters. By constructing a closed-loop operation mechanism of dynamic monitoring, real-time analysis, and adaptive control, they achieve dynamic matching between support strength and rock mass condition, upgrading the support system from static defense to proactive response, and significantly improving the safety and efficiency of deep mining.

[0045] This invention provides a support system for preventing and intelligently supporting high-stress disasters in deep mines, comprising:

[0046] The system consists of a step-by-step pressure relief subsystem and an intelligent support subsystem. The step-by-step pressure relief subsystem includes a compressed air device, a high-pressure water injection device, and a borehole sealing device. The compressed air device is used to inject compressed air into the fracture-inducing borehole for initial fracture initiation. The high-pressure water injection device is used to inject high-pressure water into the borehole after the initial fracture initiation for secondary fracture expansion and fracturing.

[0047] The intelligent support subsystem includes fiber optic sensing anchor bolts, an AI algorithm platform, and electro-hydraulic hydraulic supports. The fiber optic sensing anchor bolts are deployed around the tunnel to collect multi-dimensional physical data of the rock mass in real time and transmit it to the AI ​​algorithm platform. The AI ​​algorithm platform has a built-in LSTM prediction model to analyze the received data and output the predicted support load and control commands. The electro-hydraulic hydraulic supports are connected to the AI ​​algorithm platform to receive control commands and automatically adjust the support force.

[0048] Through the above system, the following intelligent support operations can be performed:

[0049] S4. Selection and Deployment of Fiber Optic Sensing Anchor Bolts: By deploying sensing anchor bolts with built-in distributed optical fibers around the roadway in the decompression area, the strain data of the rock mass is collected in real time and converted into the rock mass deformation rate and cumulative deformation.

[0050] Specifically, the first key component of the system's operation is fiber optic sensing monitoring. The technical team will deploy distributed fiber optic sensing anchor bolts at critical locations such as the roof and sidewalls of the roadway in the decompression area. These anchor bolts embed ultra-fine fiber optic sensors, enabling real-time acquisition of multi-dimensional data on rock deformation, stress changes, and temperature fluctuations. Compared to traditional point sensors, distributed fiber optic sensing technology offers a wider monitoring range (up to 1000m for a single fiber) and higher accuracy (deformation monitoring accuracy down to 0.1mm), capturing minute changes in the rock mass from all angles. For example, when the roadway roof experiences a 0.5mm subsidence, the fiber optic sensor can transmit the data to the central control system within one second, providing real-time data support for subsequent analysis. Simultaneously, through the collaborative monitoring of multiple sensor sets, a three-dimensional distribution model of the rock mass stress field can be constructed, accurately locating potential risk areas.

[0051] S5. Algorithm Platform and Hydraulic Support Control: The transformed rock mass deformation data is input into a pre-trained LSTM deep learning prediction model to predict the trend of support load changes within a set time period. Based on the prediction results, control commands are sent to the electro-hydraulic control system of the hydraulic support to automatically adjust the hydraulic supply pressure of the column to dynamically adjust the support force.

[0052] Specifically, in the AI ​​prediction algorithm stage, the system inputs real-time data collected by fiber optic sensors (such as rock deformation rate and stress change amplitude) into a pre-trained machine learning model. This model, built upon massive amounts of deep mining case data, can identify the correlation between rock deformation and support load through deep learning. When the rock deformation rate exceeds 0.3 mm / h, based on Hooke's law in rock mechanics, the deformation is directly proportional to the stress. Combined with field data fitting, it predicts a 20%-30% increase in support load within 1-2 hours (fitting formula: σ = 0.6v + 1.2, where σ is the load increase and v is the deformation rate, in mm / h). For example, research shows that when the rock deformation rate in a certain area exceeds 0.3 mm / h, the model automatically predicts that the support load in that area will increase by 20%-30% within the next 1-2 hours. Compared to manual analysis, the AI ​​algorithm has a faster response time (data processing time less than 10 seconds) and higher prediction accuracy (error rate less than 5%), enabling it to identify potential risks that are difficult to detect through traditional manual inspections, thus gaining valuable time for support control. In addition, the model has self-learning capabilities and can continuously optimize its algorithm by accumulating on-site data to adapt to the changing patterns of rock masses in different mining scenarios.

[0053] Hydraulic Support Adaptive Control: Finally, the system enters the hydraulic support adaptive control phase. Based on the control commands output by the AI ​​algorithm, the electro-hydraulic control system of the hydraulic support automatically adjusts the hydraulic supply pressure of the support column to achieve real-time adjustment of the support force. For example, when the AI ​​predicts that the support load in a certain area will increase, the system will automatically increase the support force of the hydraulic support to ensure that the support and rock mass deformation are synchronized; if the rock mass stress in a certain area is detected to be stabilizing, the system will appropriately reduce the support force to avoid material waste caused by over-support. This dynamic control mechanism keeps the support strength always within the optimal range, preventing roadway deformation due to insufficient support force and resource waste due to excessive support force. Field application data (as shown in Table 2-1) shows that compared with traditional passive support, the intelligent support system can reduce roadway maintenance costs by 25%–30% and reduce the roof accident rate by more than 80% (compared to the traditional passive support scheme in Xinwen Suncun Mine, with a statistical period of 6 months and a roadway length of 500m in the intelligent support system application section). It also reduces the frequency of manual intervention and improves mining operation efficiency.

[0054] Table 2-1 Comparison of data between hydraulic support and traditional support

[0055] index Traditional passive support Hydraulic support adaptive adjustment Increase / Decrease Amount tunnel maintenance costs Baseline value (set to 100%) 70%～75% Reduced by 25% to 30% Roof accident rate Baseline value (set to 100%) ＜20% A decrease of more than 80% Frequency of human intervention High frequency (requires regular inspection and adjustment) Low frequency (intervention only when abnormalities occur) Reduce by more than 60% Support and rock mass matching degree Low (fixed parameters, prone to mismatch) High (dynamic matching) Increase by more than 70%

[0056] From a technical perspective, the core value of intelligent support systems lies in achieving integrated closed-loop control of perception, decision-making, and execution. By deeply integrating the real-time sensing capabilities of fiber optic sensors, the intelligent decision-making capabilities of AI algorithms, and the precise execution capabilities of hydraulic supports, it completely changes the passive response of traditional support technologies, providing intelligent solutions for rock mass control in deep, high-stress environments, and laying a key technological foundation for future unmanned mining.

[0057] To further implement the above embodiments, the present invention will be further described in detail through the following implementation methods:

[0058] 1. Preliminary Preparation: Based on the rock mass mechanical parameters (such as compressive strength and elastic modulus) and stress distribution characteristics, determine the parameters of the fracturing holes. The hole depth is set to 8-15m, which is based on the distribution law of stress concentration zones in deep rock masses (when the burial depth reaches 1000m, the thickness of the stress concentration zone is about 12m), aiming to ensure that the borehole penetrates the high-stress layer; the hole diameter is determined to be 90-110mm, which is suitable for the output torque of the mining hydraulic drilling rig (the output torque must be greater than or equal to 300N・m), and also meets the high-pressure hose interface specifications; the hole spacing is set to 3-5m, which is based on the stress diffusion radius of the rock mass (about 2-3m), to avoid stress superposition between holes or the appearance of pressure relief blind zones.

[0059] Compressed air equipment uses screw-type air compressors with a rated pressure of 30–50 MPa, equipped with high-pressure hoses (pressure resistance rating not less than 60 MPa) and orifice sealing devices; hydraulic fracturing equipment uses high-pressure plunger pumps with a rated pressure of 50–80 MPa, equipped with a flow control system (accuracy ±0.5 L / min) and a pressure sensor (accuracy ±0.1 MPa). Before construction, the equipment undergoes no-load testing to check the pressure display, flow control, and sealing performance, ensuring no leaks or malfunctions.

[0060] Subsequently, the specific procedures for depressurization in stages (such as...) will be carried out. Figure 3 It is divided into two parts: compressed air fracturing and hydraulic fracturing.

[0061] 2. Step-by-step depressurization - compressed air fracturing:

[0062] Connect the high-pressure hose of the compressed air equipment to the sealing device at the orifice of the fracture hole, ensuring a tight connection and using double sealing rings to prevent air leakage. Set the pressure parameters according to the rock mass hardness: 15-25 MPa for soft rock (compressive strength < 60 MPa), 25-35 MPa for medium-hard rock (60-100 MPa), and 35-45 MPa for hard rock (> 100 MPa). Adjust the fracturing time according to the hole depth, 1-2 minutes per meter of hole depth to ensure sufficient initial fracture development. Start the equipment and slowly increase the pressure to the set pressure, maintaining a stable pressure for the preset time. During the process, pressure changes are monitored in real time using a pressure sensor at the orifice. If the pressure drops sharply (a decrease of >5 MPa), it is determined that the fracture has been completed, and fracturing can be terminated early. If the pressure continues to rise without any downward trend, the pressure should be reduced by 10%–15%, and the pressure stabilization time after the reduction should be extended by 50% (from 1–2 minutes per meter of hole depth to 1.5–3 minutes) to ensure that the initial fracture develops fully. If the pressure continues to rise after the reduction (an increase of >5 MPa within 10 minutes), fracturing should be stopped, and the orifice should be checked for blockage. Rock debris should be cleaned, and construction should be restarted to avoid excessive rock fragmentation. After shutting down the equipment, the pressure should be slowly released to atmospheric pressure, the connecting devices should be removed, and the pressure-time curve and any abnormalities of this fracturing operation should be recorded.

[0063] 3. Step-by-step pressure relief - hydraulic fracturing:

[0064] Within 12 hours of completing compressed air fracturing (to prevent fracture closure), connect the hose of the high-pressure water injection equipment to the fracturing hole sealing device, and replace the water-use sealing ring with a suitable one to ensure reliable sealing. The water pressure setting should be 10–15 MPa higher than the air fracturing pressure: 25–35 MPa for soft rock, 35–45 MPa for medium-hard rock, and 45–55 MPa for hard rock. The water injection rate should be adjusted according to the fracture development (as shown in Table 3-1), initially injecting at a low rate of 5–8 L / min, and increasing to 10–15 L / min after the pressure stabilizes to avoid sudden pressure increases that could cause rock mass collapse. Start the equipment to inject high-pressure water, and monitor the injection volume using a flow sensor. When the injection volume reaches 3–5 times the hole volume or the pressure shows 2–3 periodic fluctuations, it is determined that the fracture has fully expanded, and water injection should be stopped. During the process, monitor the displacement of the rock mass surface in real time. If the displacement exceeds 5 mm / h, immediately reduce the water pressure by 5–10 MPa to prevent secondary disasters. After shutting down the equipment, gradually depressurize, dismantle the equipment, and seal the fracturing hole. Simultaneously, water pressure, water injection rate, water injection volume, and rock mass displacement data are recorded to generate a single-hole pressure relief report.

[0065] Table 3-1 Selection Table for Step-by-Step Pressure Relief Technology

[0066] soft rock <60 Pressure 15-25 MPa, fracture occurs in 1-2 minutes per meter of hole depth. Pressure 25–35 MPa, initial water injection 5–8 L / min, stabilizing to 10–15 L / min Hole depth 8–12 m, hole diameter 90 mm, hole spacing 4–5 m medium hard rock 60～100 Pressure 25-35 MPa, fracture occurs in 1-2 minutes per meter of hole depth. Pressure 35–45 MPa, initial water injection 5–8 L / min, stabilizing to 10–15 L / min Hole depth 10–13 m, hole diameter 100 mm, hole spacing 3.5–4.5 m Hard rock >100 Pressure 35-45 MPa, fracture occurs in 1-2 minutes per meter of hole depth. Pressure 45–55 MPa, initial water injection 5–8 L / min, stabilizing to 10–15 L / min Hole depth 12–15 m, hole diameter 110 mm, hole spacing 3–4 m Anomaly handling threshold - A sudden pressure drop exceeding 5 MPa prematurely terminates the rupture process; a sustained increase in pressure reduces the rupture rate by 10%–15%. When the water injection volume reaches 3-5 times the orifice volume and the pressure fluctuates 2-3 times, the injection should be stopped; when the rock mass displacement is >5mm / h, the water pressure should be reduced by 5-10MPa. -

[0067] 4. Selection and Deployment of Fiber Optic Sensing Anchor Bolts: Intrinsically safe fiber optic sensing anchor bolts for mining are selected. The bolt body is made of 20MnSi steel (diameter 20-22mm, length 2.5-3.5m), with built-in distributed optical fibers. Its sensing accuracy is ±2με (compliant with GB / T34960-2017 "General Technical Requirements for Fiber Optic Sensing Systems", calibrated by the National Mining Machinery Quality Supervision and Inspection Center), with a measurement range of 5000-5000με, IP68 protection rating, and suitability for humid and dusty underground environments. Fiber optic sensing anchor bolts are deployed bidirectionally along the roadway's strike and dip. The strike spacing is set at 3-5m: this spacing matches the pressure relief hole spacing (3-5m) to ensure comprehensive coverage of the pressure relief influence area. The dip spacing is set at 2-3m: this setting is based on the development width of transverse rock fractures (approximately 1.5-2.5m), aiming to avoid missed detection of local deformation. In stress concentration areas, the deployment is denser, with the spacing reduced to 1.5-2m. Resin anchoring agent is used during installation to ensure that the anchoring force of the anchor rod is ≥80kN and the fit between the rod body and the rock mass is >95%, so as to avoid the distortion of monitoring data due to gaps (as shown in Table 3-2).

[0068] Table 3-2 Equipment List

[0069] screw air compressor LG-22 / 50 Kaishan Group Explosion-proof underground (ExdI) Rated pressure of 50 MPa, meeting the requirements for fracturing hard rock. High-pressure plunger pump 3DSY-100 / 80 Dezhou Deyuan Mining water corrosion resistance Rated pressure 80MPa, suitable for hydraulic fracturing parameters Industrial-grade servers ThinkSystemSR860 Lenovo Ground control room (0-40℃) Supports GPU expansion (NVIDIA A100) to meet AI computing needs. Intrinsically safe fiber optic sensing anchor for mining FBG-MA-20 China Coal Technology & Engineering Chongqing Research Institute Dampness (RH≤95%) and dust in the mine The pole is made of 20MnSi steel, and the anchoring force is ≥80kN.

[0070] 5. Algorithm Platform and Hydraulic Support Control: Built on an industrial-grade server (CPU: Intel Xeon Gold 6330, GPU: NVIDIA A100), it incorporates an LSTM prediction model. Input parameters include rock mass deformation rate, cumulative deformation, and stress value after decompression. Output parameters are the support load change trend for the next hour. AI prediction accuracy is ≥90% (based on 12 sets of deep mine monitoring datasets, containing 50,000 deformation-load samples, with a test set error rate <8%). The platform supports real-time data storage (storage capacity ≥10TB), curve visualization, and anomaly alarms (alarm response time <1 second). Electro-hydraulic hydraulic supports (support resistance 3000~6000kN) are selected, equipped with intrinsically safe controllers, pressure sensors (measurement range 0~80MPa, accuracy ±0.2MPa), and electro-hydraulic valves (response time <0.3 seconds). The controller and AI algorithm platform are connected via industrial Ethernet to achieve real-time command reception and execution status feedback.

[0071] Intelligent support operation procedure: Intelligent support operation procedure (e.g.) Figure 4 It is mainly divided into four parts: system deployment and initialization, real-time monitoring and data transmission, AI prediction and adaptive adjustment, and regular inspection and maintenance.

[0072] System Deployment and Initialization: After installing the fiber optic sensing anchor bolts according to the design plan, connect the fiber optic cable to the downhole fiber optic junction box and connect it to the AI ​​algorithm platform via the backbone fiber optic cable (single-mode fiber, transmission distance ≥10km). Establish communication between the hydraulic support controller and the platform, and complete equipment address allocation and parameter initialization (e.g., setting the initial support force of the support to 60%–70% of the rated value). Then, start the system for no-load testing, sending simulated adjustment commands to the hydraulic support (e.g., adjusting the support force ±10%) to verify the consistency between the controller response and the support action. Collect rock mass deformation data under no-load conditions through the fiber optic sensing system to confirm stable data transmission. During initialization, key parameters for step-by-step depressurization (e.g., air fracturing pressure, rock mass stress value after hydraulic fracturing) are uploaded to the AI ​​algorithm platform via industrial Ethernet as the 'initial stress input parameters' of the LSTM model. If the stress decreases by more than 30% after depressurization, the initial support force is set to 60% of the rated value (normally 60%–70%).

[26] Reduce excessive support.

[0073] Real-time monitoring and data transmission: Fiber optic sensing anchors collect rock strain data at a sampling frequency of 1Hz and transmit it to an AI algorithm platform via fiber optic cable. The platform filters (removes high-frequency noise) and normalizes the raw data, converting it into rock deformation rate (mm / h) and cumulative deformation (mm). Hydraulic support pressure sensors collect support force data every 5 seconds and upload it to the platform in real time. This data is then correlated with the rock deformation data and stored to form a real-time deformation-load monitoring curve.

[0074] AI Prediction and Adaptive Adjustment: The AI ​​algorithm platform analyzes monitoring data every 5 minutes and predicts the support load change within the next hour based on the LSTM model. If the predicted load increases by more than 10%, a support force increase command is sent to the hydraulic support (increasing by 5% to 15%); if the predicted load decreases by more than 10%, a support force decrease command is sent (decreasing by 5% to 10%). After receiving the command, the hydraulic support controller adjusts the hydraulic fluid supply to the column through the electro-hydraulic valve to achieve adaptive adjustment of the support force. The AI ​​model calls the pressure-time curve data of the depressurization process in real time. If stress rebound is detected after depressurization (e.g., stress rebound > 10% within 1 hour), the support load prediction threshold is automatically lowered by 15%, triggering the support force increase command in advance. After the adjustment is completed, the actual support force data is fed back to the platform, forming an adjustment closed loop.

[0075] Table 3-3 Logic Table for Monitoring and Adjusting Parameters of Intelligent Support System

[0076] Rock mass deformation rate 1Hz <0.3mm / h ≥0.3mm / h It is predicted that the load will increase by 20% to 30% and the support force will increase by 8% to 15% within 1 to 2 hours. Cumulative deformation of rock mass 1Hz <5mm (within 24 hours) ≥5mm (within 24 hours) Combined with deformation rate, the load is increased by 15%–25%, and the support force is increased by 5%–12%. Hydraulic support force 5 seconds / time Rated value 60%~100% Less than 60% of the rated value or more than 100% of the rated value. <60%, immediately increase by 10%–15%; >100%, slowly decrease by 5%–10%. Fiber optic sensor data packet loss rate Real-time monitoring ＜0.1% ≥0.1% Triggering a device self-test command prompts a check of the fiber optic interface.

[0077] Regular inspections and maintenance: Daily visual inspections of fiber optic sensing anchors are conducted to check for bending and loose fiber optic interfaces; any damaged anchors are replaced immediately. Weekly optical fiber loss is measured using an optical time domain reflectometer (OTDR) to ensure a loss rate <0.5dB / km. Every 3 days, the sealing performance, column extension / retraction status, and electro-hydraulic valve operation of the hydraulic supports are checked; any leaks or jamming are addressed by shutting down the system for maintenance. Monthly data backups are performed on the AI ​​algorithm platform, and the LSTM model is iteratively trained based on newly added monitoring data (no less than 1000 sets) to optimize prediction accuracy. When the fiber optic data packet loss rate is ≥0.1%, the system automatically switches to emergency support mode, maintaining the current support force of the hydraulic supports and triggering an audible and visual alarm. When the electro-hydraulic valve malfunctions (response time >0.3 seconds), the controller activates the manual control interface, allowing on-site personnel to adjust the support force via the control panel with an adjustment accuracy of ±5% of the rated value.

[0078] like Figure 5 As shown, the process of intelligent support from deployment to operation and maintenance (the following process) is clearly presented, enabling technicians to quickly understand the system operation steps and execution sequence.

[0079] (1) The intelligent support is divided into four core stages: system deployment and initialization, real-time monitoring and data transmission, AI prediction and adaptive adjustment, and regular inspection and maintenance. The order and relationship of each stage are clearly defined.

[0080] (2) Use step-by-step graphics to show key operations, so that on-site personnel can quickly grasp the key points of execution.

[0081] (3) The intelligent support closed loop of "perception-analysis-execution-maintenance" is presented intuitively, highlighting the system's automation and intelligence features.

[0082] (4) Mark the regular maintenance nodes, clarify the safety operation and maintenance requirements such as equipment maintenance and data backup, and ensure the stable operation of the system.

[0083] In the diagram, C2 represents the electro-hydraulic control system of the hydraulic support, which is responsible for receiving instructions from the AI ​​platform, adjusting the hydraulic supply pressure of the column, adjusting the support force, and transmitting the actual support force data back to the platform to complete the "decision-execution-feedback" closed loop.

[0084] C3 stands for AI algorithm analysis and command issuance platform, which is responsible for processing fiber optic sensor data, running LSTM prediction models, and generating support and adjustment commands.

[0085] The synergistic effect of C2 / C3: C3 (AI decision-making) outputs control commands—C2 (electro-hydraulic control execution) adjusts actions—C2 provides feedback on the actual state—C3 continuously optimizes, and the two work together to achieve adaptive support.

[0086] In summary, this application proposes and constructs an integrated control system for active pressure relief and intelligent support, achieving a technological leap from passive defense to active regulation.

[0087] In terms of pressure relief, the system adopts a step-by-step collaborative process of compressed air fracturing and hydraulic fracturing. Through the orderly combination of pilot gas fracturing and subsequent hydraulic fracturing, the system achieves a stable and controllable release of the elastic potential energy of the rock mass, effectively reducing the stress concentration by 30%-50% and mitigating the risk of disaster from the source.

[0088] In terms of support, the system integrates distributed fiber optic sensing, LSTM deep learning algorithms, and electro-hydraulic control technology to construct a closed-loop control architecture of real-time perception, intelligent prediction, and precise execution. This system can dynamically capture rock mass deformation, predict load changes in advance, and automatically adjust support parameters, significantly improving the adaptability and reliability of the support system.

[0089] The key advantage of this application lies in integrating decompression and support from traditionally independent processes into a data-driven, collaborative, and interconnected whole. Through process collaboration, technology integration, and system optimization, it achieves refined, intelligent, and efficient control of deep, high-stress rock masses, providing a practical technical path and system solution for the safe and efficient mining of deep resources, and has significant value for widespread application.

[0090] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for preventing and intelligently supporting high-stress disasters in deep mines, characterized in that, Includes the following steps: S1. Preliminary preparation: Detect the rock mass around the deep tunnel to determine the stress concentration points, and construct fracture holes at the stress concentration points; S2, Step-by-step decompression-compressed air fracturing: Compressed air is injected into the fracturing holes, allowing the gas to penetrate along the original weak surface inside the rock mass and form an initial fracture network, so as to release part of the elastic potential energy of the rock mass; S3, Step-by-step pressure relief-hydraulic fracturing: After the compressed air fracturing is completed, high-pressure water is injected into the fracturing hole. The high-pressure water seeps along the initial fracture network to produce a splitting effect, thereby achieving directional expansion and connection of fractures. S4. Selection and deployment of fiber optic sensing anchor bolts: By deploying sensing anchor bolts with built-in distributed optical fibers around the roadway in the pressure relief area, the strain data of the rock mass is collected in real time and converted into the rock mass deformation rate and cumulative deformation. S5. Algorithm Platform and Hydraulic Support Control: The transformed rock mass deformation data is input into a pre-trained LSTM deep learning prediction model to predict the trend of support load changes within a set time period. Based on the prediction results, control commands are sent to the electro-hydraulic control system of the hydraulic support to automatically adjust the hydraulic supply pressure of the column to dynamically adjust the support force.

2. The method for prevention and intelligent support of high-stress disasters in deep mines according to claim 1, characterized in that, In step S1, the construction parameters for the fracture-inducing holes are: hole depth set to 8-15m, hole diameter set to 90-110mm, and hole spacing set to 3-5m.

3. The method for prevention and intelligent support of high-stress disasters in deep mines according to claim 1, characterized in that, In step S2, the specific control conditions for compressed air fracturing are as follows: the injection pressure is set according to the rock mass hardness: the pressure is set at 15-25 MPa for soft rock, 25-35 MPa for medium-hard rock, and 35-45 MPa for hard rock; the fracturing pressure holding time is 1-2 minutes per meter of hole depth.

4. The method for prevention and intelligent support of high-stress disasters in deep mines according to claim 3, characterized in that, During the fracturing process, the orifice pressure is monitored in real time. If the pressure drop is greater than 5 MPa, it is judged that the fracture has been completed and the fracturing is terminated in advance. If the pressure continues to rise and there is no downward trend, the pressure is reduced by 10% to 15%, and the pressure stabilization time after the pressure reduction is extended by 50%.

5. The method for prevention and intelligent support of high-stress disasters in deep mines according to claim 4, characterized in that, In step S3, the specific control conditions for hydraulic fracturing are as follows: high-pressure water injection is performed within 12 hours after compressed air fracturing is completed; the set high-pressure water pressure is 10-15 MPa higher than the compressed air fracturing pressure, initially injected at a low speed of 5-8 L / min, and increased to 10-15 L / min after the pressure stabilizes; when the injection volume reaches 3-5 times the volume of the fracturing hole, or when the pressure shows 2-3 periodic fluctuations, it is determined that the fracture has been fully expanded and water injection is stopped.

6. The method for prevention and intelligent support of high-stress disasters in deep mines according to claim 1, characterized in that, In step S5, the logic of advance prediction and adaptive adjustment is as follows: when the rock mass deformation rate is ≥0.3mm / h, the LSTM model predicts that the support load will increase by 20% to 30% in the next 1 to 2 hours; if the predicted load increases by more than 10%, an instruction to increase the support force by 5% to 15% is sent to the hydraulic support; if the predicted load decreases by more than 10%, an instruction to decrease the support force by 5% to 10% is sent; the model calls the pressure-time curve data of the decompression stage in real time, and if the stress recovery is >10% within 1 hour after decompression, the support load prediction threshold is automatically lowered by 15%, triggering the support force increase instruction in advance.

7. A support system applying the method for prevention and intelligent support of high-stress disasters in deep mines as described in any one of claims 1-6, characterized in that, The system includes: The system comprises a step-by-step pressure relief subsystem and an intelligent support subsystem. The step-by-step pressure relief subsystem includes a compressed air device, a high-pressure water injection device, and a borehole sealing device. The compressed air device is used to inject compressed air into the fracture-inducing borehole for initial fracture initiation. The high-pressure water injection device is used to inject high-pressure water into the borehole after the initial fracture initiation for secondary fracture expansion and fracturing. The intelligent support subsystem includes fiber optic sensing anchor bolts, an AI algorithm platform, and electro-hydraulic hydraulic supports. The fiber optic sensing anchor bolts are deployed around the tunnel to collect multi-dimensional physical quantity data of the rock mass in real time and transmit it to the AI ​​algorithm platform. The AI ​​algorithm platform has a built-in LSTM prediction model to analyze the received data and output the predicted support load and control commands. The electro-hydraulic hydraulic supports are communicatively connected to the AI ​​algorithm platform to receive the control commands and automatically adjust the support force.

8. The deep mine high-stress disaster prevention and intelligent support system according to claim 7, characterized in that: The compressed air equipment is a screw air compressor with a rated pressure of 30-50 MPa; the high-pressure water injection equipment is a high-pressure plunger pump with a rated pressure of 50-80 MPa.

9. The deep mine high-stress disaster prevention and intelligent support system according to claim 7, characterized in that: The fiber optic sensing anchor is an intrinsically safe anchor for mining with built-in distributed optical fiber. The anchor body is made of 20MnSi steel, and its sensing accuracy is ±2με. The deployment spacing is 3-5m in the direction of the roadway, 2-3m in the dip direction of the roadway, and the spacing is increased to 1.5-2m in stress concentration areas.

Images