Mine movable high-pressure water jet precision roof cutting and pressure releasing method

Abstract

The present application relates to the technical field of coal mining, and particularly relates to a mine movable high-pressure water jet precision roof cutting pressure relief method, comprising collecting roof geological parameters of a construction area, water jet system performance parameters, hydraulic fracturing system rated parameters and mine environment parameters, constructing a CNN model, outputting self-adaptive iteration step and depth-distance coupling coefficient, and correcting drilling horizontal spacing according to a comparison result of a current pressure approximation coefficient and a preset target interval, and simultaneously correcting drilling depth according to the coupling coefficient, so that actual fracturing pressure quickly approximates to rated pressure in multiple feedback rounds, drilling spacing is gradually increased from a conservative initial value to an optimal value, drilling density is maximally reduced, roof cutting period is shortened, and working condition adaptation capability of different mine environments is significantly improved under the premise of ensuring roof cutting effect and dynamic safety redundancy.

Description

Technical Field

[0001] This invention relates to the field of coal mining technology, and in particular to a mobile high-pressure water jet method for precise roof cutting and pressure relief in mining. Background Technology

[0002] He Manchao and his team proposed the theory of short-arm roof cutting, overturning the traditional approach of leaving coal pillars to support the roof. They clarified the technical path of cutting off the stress transfer between the goaf and the roadway through directional pre-splitting of the roof, utilizing the roof's own weight to collapse and form a self-stabilizing structure. This is the "110-N00" method of self-forming roadway mining without coal pillars. This method combines roof cutting and stress relief with roadway support, using constant-resistance anchor cables to strengthen the support. The working face is advanced, and the roof is directionally pre-splitted. After the working face passes, the roof automatically collapses along the cut to form the roadway side, achieving multi-purpose use of a single roadway and pillarless mining. Under the technical framework established by He Manchao's team, domestic research institutions and enterprises have further optimized and iterated, forming diverse branches of roof cutting and stress relief technology.

[0003] The team led by Dong Guowei at Xi'an University of Science and Technology invented a "method for hydraulic fracturing and roof cutting in coal mine preparation roadways" (publication number CN120007249B). By identifying the key layer above the coal seam and calculating the minimum overhang length, the method achieves precise positioning of hydraulic fracturing and solves the problem of long-term stability of preparation roadways.

[0004] The team led by Zhang Huiyuan of Guizhou Panjiang Refined Coal Co., Ltd. proposed a "directional pre-splitting blasting roof cutting and pressure relief method for roadway retention" with the publication number CN104806246B. This method optimizes the layout of blasting holes and the design of directional circular pipes, reducing the cost and complexity of roadway backfilling.

[0005] The team led by Yu Bin of Datong Coal Mine Group proposed a "precise roof cutting and pressure relief method on the ground" with the publication number CN110080769B. This method solves the problem of roof cutting in hard roofs of roadways near the ground by using ground drilling and perforation fracturing technology, and expands the applicable scenarios of the technology.

[0006] The team led by Zhang Xiangyang from Anhui University of Science and Technology proposed a "Pre-cutting and Pressure Relief Method for the Final Mining Stage of Coal Seam Working Faces" (publication number CN113863927B). This method optimizes drilling and blasting schemes for the hard roof of extra-thick coal seams, improving the extraction rate and safety in the final mining stage.

[0007] Waterjet roof cutting combined with hydraulic fracturing and depressurization has been widely used in the support of hard roof roadways in high-gas environments due to its advantages of low disturbance and high safety. However, the design of construction parameters for waterjet roof cutting currently lacks dynamic adaptability. Existing methods mostly calculate borehole spacing based on static geological parameters or empirical formulas, without fully considering the actual performance of the waterjet roof cutting system and the rated pressure constraints of the hydraulic fracturing equipment. In order to ensure successful roof cutting, an overly conservative borehole spacing design is often adopted, resulting in excessive borehole density, increased construction workload, and longer roof cutting period, which seriously restricts the efficiency of coal mining.

[0008] To address this, the inventors proposed a mobile high-pressure water jet method for precise roof cutting and pressure relief in mining. With the core objective of approximating the rated pressure to the actual fracturing pressure, the method utilizes the strong feature extraction and nonlinear fitting capabilities of CNN to optimize parameter modeling and feedback correction processes. This maximizes the reduction of drilling density and shortens the roof cutting period while ensuring the roof cutting effect and safety redundancy. Summary of the Invention

[0009] The purpose of this invention is to provide a mobile, high-pressure water jet method for precise roof cutting and pressure relief in mining, thereby solving the problems mentioned in the background art. The specific technical solution is as follows:

[0010] The purpose of this invention is to provide a mobile high-pressure water jet method for precise roof cutting and pressure relief in mining, comprising the following steps:

[0011] S1: Collect the geological parameters of the roof, the performance parameters of the water jet system, the rated parameters of the hydraulic fracturing system, and the mine environmental parameters of the construction area, and form input feature data after preprocessing;

[0012] S2: Construct a CNN model, extract nonlinear coupling features between parameters based on the input feature data, output a local cutability coefficient matrix and a global cutability coefficient that characterize the difficulty of cutting the roof in different areas of the roof, and output a safety redundancy coefficient that matches the mine environment parameters.

[0013] S3: Based on the global cutability coefficient, water jet system performance parameters and preset conventional effective cut radius, calculate the borehole spacing benchmark value, and generate initial borehole parameters according to a conservative coefficient. The initial borehole parameters include borehole lateral spacing, borehole angle, borehole depth, borehole position distribution and top cutting cycle step.

[0014] S4: Perform water jet top cutting construction according to the initial drilling parameters. During the fracturing construction, the actual fracturing pressure of the hydraulic fracturing system is collected in real time, and the pressure approximation coefficient between the actual fracturing pressure and the rated pressure is calculated.

[0015] S5: Input historical construction parameters, historical pressure approximation coefficients, current local cutability coefficient matrix and real-time water jet performance parameters into the CNN parameter optimization network, output adaptive iteration step size and depth-spacing coupling coefficient, and correct the borehole lateral spacing according to the comparison result between the current pressure approximation coefficient and the preset target interval, and simultaneously correct the borehole depth according to the coupling coefficient.

[0016] S6: Repeat steps S4 to S5 for multiple rounds of iterative optimization until the pressure approximation coefficient stabilizes within the preset target range and the borehole parameter fluctuations meet the stability conditions, generate the optimal borehole parameters and execute subsequent construction.

[0017] Preferably, in step S1,

[0018] The geological parameters of the roof include: Protodyakonov coefficient f, joint density Jd, and average roof stress σ. avg Top plate thickness h, rock fracture angle α, uniaxial compressive strength σ c ;

[0019] The performance parameters of the water jet system include: jet pressure. Jet flow rate Abrasive mixing ratio ;

[0020] The rated parameters of the hydraulic fracturing system include: rated pressure. ;

[0021] The mine environmental parameters include: mine disaster level, roadway cross-sectional dimensions, and support structure parameters.

[0022] Preferably, the CNN model in step S2 includes:

[0023] The parameter feature extraction network uses a 1D convolutional neural network, which contains 4 convolutional layers and 3 pooling layers, and outputs a 64-dimensional feature vector to extract nonlinear coupling features between different parameters and historical sequence trend features.

[0024] The cutability coefficient prediction network includes three fully connected layers and outputs a 32×32-dimensional local cutability coefficient matrix, which is used to output the local cutability coefficient matrix and the global cutability coefficient based on the deep spatial feature vector.

[0025] The safety redundancy prediction network, using the MobileNet architecture, is used to output the safety redundancy coefficient β based on mine environment parameters.

[0026] Preferably, the initial borehole spacing reference value in step S3 The calculation formula is:

[0027] ;

[0028] in, This is the standard effective kerf radius; The global tangibility coefficient; For reference cutability coefficient; These are empirical coupling coefficients;

[0029] The effective action coefficient of the water jet is calculated using the following formula:

[0030] ;

[0031] Where a, b, and c are weighting coefficients, and the sum of a, b, and c is 1; For the rated jet pressure, For the rated jet flow rate, The rated abrasive mixing ratio is given; and the value of F is not greater than 1.

[0032] Preferably, the initial borehole lateral spacing in step S3 The calculation formula is:

[0033] ;

[0034] in, This is a conservative coefficient, with a value of 0.8 to 0.9. 0.8 is used for mines prone to rock bursts, and 0.9 is used for conventional mines.

[0035] Initial drilling angle The calculation formula is:

[0036] ;

[0037] in, The value ranges from 0.8° to 1.2°, and ∈[60°, 90°];

[0038] The formula for calculating the initial borehole depth D1 is:

[0039] ;

[0040] in To allow for a certain depth to penetrate;

[0041] The initial borehole location distribution is determined based on the local cutability coefficient matrix, and boreholes are densely distributed in areas where the cutability coefficient is lower than a preset threshold.

[0042] Preferably, the pressure approximation coefficient The calculation formula is:

[0043] ;

[0044] in This represents the actual crack initiation pressure. The rated pressure; the pressure approximation coefficient The preset target range is [0.85, 1.05].

[0045] Preferably, the rule for correcting the lateral spacing of the boreholes in step 5 is based on the pressure approximation coefficient. The value of ΔS is used to adjust the CNN parameters and optimize the adaptive iteration step size ΔS of the network output. CNN Specifically:

[0046] like ∈[0.85, 1.05], keep the current parameters, ΔS CNN Take the minimum value;

[0047] like If the pressure is less than 0.85, it is determined to be too low; therefore, the lateral spacing of the boreholes should be increased. = +ΔS CNN ;

[0048] like A value >1.05 indicates excessive pressure; therefore, the lateral spacing of the boreholes should be reduced. = ΔS CNN ;

[0049] ΔS CNN The value range is 0.05~0.3m.

[0050] Preferably, the formula for correcting the drilling depth in step 5 is:

[0051] ;

[0052] in, The depth-space coupling coefficient for optimizing the synchronous output of the CNN parameters is set, with a value ranging from 0.3 to 0.6.

[0053] Corrected borehole lateral spacing Meet safety constraints:

[0054] ∈[0.7×X base X base ×β];

[0055] Corrected drilling depth Meet safety constraints:

[0056] ∈[D min D max ];

[0057] Where β is the safety redundancy coefficient, D min To ensure the minimum depth for the cut-off effect, D max To protect the maximum allowable depth of the upper critical layer.

[0058] Preferably, in step 6, the pressure approximation coefficient The criteria for determining whether borehole parameters are stable within a preset target range and meet the stability conditions are as follows:

[0059] Three consecutive rounds of construction The values ​​all fall within the range of [0.85, 1.05], and the fluctuation range of the lateral spacing of the boreholes does not exceed 0.1m, and the fluctuation range of the borehole depth does not exceed 0.2m.

[0060] Preferably, the optimal drilling parameters include the optimal lateral spacing S. opt Optimal drilling angle θ opt Optimal drilling depth D opt Optimal tipping cycle step size L opt The method also includes continuously monitoring the cut penetration rate, stress release rate, and roof collapse rate for effect acceptance after generating the optimal drilling parameters; the CNN model is pre-trained based on the historical construction dataset of the mining area, and the pre-training loss function adopts the mean square error, and it stops when the loss value is less than 0.001; incremental learning fine-tuning is performed after each optimization stage is completed during the construction process.

[0061] The present invention provides a mobile high-pressure water jet precision roof cutting and pressure relief method for mines, which has the following beneficial effects:

[0062] This invention extracts features such as roof joint distribution, stress concentration zones, and rock texture using a lightweight CNN model. Combined with the actual performance parameters of the water jet system, it establishes nonlinear coupling features of roof conditions, construction parameters, and fracturing pressure response. Taking the rated pressure of the hydraulic fracturing equipment as the sole optimization target, the CNN parameter optimization network calculates the pressure approximation coefficient in real time and outputs an adaptive iteration step size and weight combination. This allows the actual fracturing pressure to rapidly approach the rated pressure in multiple rounds of feedback. The borehole spacing is gradually increased from a conservative initial value to the optimal value. Under the premise of ensuring the roof cutting effect and dynamic safety redundancy, the borehole density is minimized and the roof cutting period is shortened. At the same time, the adaptability to different mine environments is significantly improved. Attached Figure Description

[0063] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0064] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0065] The following detailed description, in conjunction with the accompanying drawings and specific embodiments, provides a more detailed explanation of the present invention: a mobile high-pressure water jet method for precise roof cutting and pressure relief in mining. The advantages and features of the present invention will become clearer from the following description. It should be noted that the accompanying drawings are all in a very simplified form and use non-precise proportions, and are only used to facilitate and clarify the explanation of the embodiments of the present invention.

[0066] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0067] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Therefore, the illustrations only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the state, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0068] This embodiment provides a mobile high-pressure water jet method for precise roof cutting and pressure relief in mines, such as... Figure 1 As shown, it includes the following steps:

[0069] Step 1: Multi-source data acquisition and CNN preprocessing:

[0070] 1.1 Collect core input data, including:

[0071] Geological parameters of the roof: Protodyakonov coefficient f, joint density Jd, average stress of the roof σ avg Top plate thickness h, rock fracture angle α, uniaxial compressive strength σ c ;

[0072] Water jet top cutting system performance parameters: jet pressure Jet flow rate Abrasive mixing ratio ;

[0073] Hydraulic fracturing system rated parameters: Rated pressure ;

[0074] Mine environmental parameters: mine disaster level, roadway cross-sectional dimensions, and support structure parameters.

[0075] 1.2 Data Preprocessing:

[0076] The parameter data is normalized and mapped to the interval [0, 1] to form a one-dimensional parameter vector.

[0077] Step 2, CNN Model Construction and Feature Extraction:

[0078] Construct a lightweight multi-task CNN model that shares a backbone feature extraction network, including the following output functions:

[0079] 2.1 Parameter Feature Extraction Network: A 1D convolutional neural network is adopted, containing 4 convolutional layers and 3 pooling layers. The input is a normalized 1D parameter vector, and the output is a 64-dimensional deep feature vector. This network extracts the nonlinear coupling features between different parameters through convolution operations, including the interaction relationship between geological parameters and equipment parameters, trend features in historical sequences, etc.

[0080] 2.2. Cutability Coefficient Prediction Network: The 64-dimensional spatial feature vector is concatenated with the one-dimensional parameter vector, processed through 3 fully connected layers, and output as a 32×32-dimensional local cutability coefficient matrix K. 局部 Each element corresponds to the shearability coefficient of a 0.1m × 0.1m top plate region, with a value range of [0.1, 1.0]. The global shearability coefficient is also calculated. =mean(K 局部 ).

[0081] 2.3 Safety Redundancy Level Prediction: Using the MobileNet architecture, the inputs are normalized mine environmental parameters and real-time roof monitoring data, stress mutation rate dσ / dt, and microseismic event frequency N. The output is the safety redundancy level and corresponding coefficient β.

[0082] Low redundancy: β=1.05;

[0083] Medium redundancy: β=1.15;

[0084] High redundancy: β=1.25;

[0085] 2.4 Model Training and Updates:

[0086] Based on the historical construction dataset of the mining area, which contains more than 1,000 sets of input data and output label samples of different roof types and working conditions, pre-training was performed. The loss function adopted was mean square error, and the iteration stopped when the loss value was ≤0.001.

[0087] During the underground construction process, incremental learning is used to fine-tune the model after each optimization stage is completed.

[0088] The total number of model parameters is controlled within 1 million, the inference time is ≤100ms, it is compatible with mining embedded controllers, and supports offline inference and breakpoint resume training.

[0089] Step 3: Generate initial conservative drilling parameters:

[0090] 3.1 Construct a CNN feature-parameter coupled model to calculate the initial borehole spacing baseline value. :

[0091] ;

[0092] in, For conventional effective slit radius, 0.3~0.5m is used for rigid top slabs; The global tangibility coefficient; For reference, the cutability coefficient is set to 0.5; The coefficient is an empirical coupling coefficient, which is set to 0.5~0.8 based on historical data fitting.

[0093] The effective action coefficient of the water jet is calculated using the following formula:

[0094] ;

[0095] Where a, b, and c are weighting coefficients, and the sum of a, b, and c is 1. Based on the fitting of 200 sets of field test data, a=0.5, b=0.3, and c=0.2 are taken. For the rated jet pressure, For the rated jet flow rate, The rated abrasive mixing ratio is given; and the value of F is not greater than 1.

[0096] 3.2. Based on the conservative coefficient Generate initial borehole lateral spacing:

[0097] ;

[0098] in, This is a conservative coefficient, with a value of 0.8 to 0.9. 0.8 is used for mines prone to rock bursts, and 0.9 is used for conventional mines.

[0099] 3.3. Combining the rock fracture angle α with the local shearability distribution extracted by CNN, the initial borehole angle is generated:

[0100] ;

[0101] in, The value ranges from 0.8° to 1.2°, and ∈[60°, 90°].

[0102] 3.4. Based on the top plate thickness h, drilling angle θ1, and local cutability coefficient, generate the initial drilling depth:

[0103] ;

[0104] in To allow for a penetration depth, the value ranges from 0.3 to 0.5 m.

[0105] 3.5. Based on the local K matrix, the stress concentration area and the high shearability area are located. Combined with the support avoidance rules, the initial borehole location distribution is generated: the high shearability area with K local matrix ≥ 0.8 is arranged according to S1, and the low shearability area with K local matrix ≤ 0.4 is densified by 10%.

[0106] 3.6 Calculate the initial cutting cycle step size:

[0107] ;

[0108] in This represents the effective propagation radius of hydraulic fracturing.

[0109] Step 4: Monitoring the fracturing process and correcting the CNN feedback:

[0110] 4.1 Drilling, waterjet fracturing, and hydraulic fracturing operations were performed according to the initial conservative drilling parameters. The actual fracturing pressure of the roof was collected in real time using a high-pressure water pressure sensor. Calculate the pressure approximation coefficient :

[0111] ;

[0112] 4.2. A CNN parameter optimization and prediction network is constructed using the ShuffleNet architecture, with historical construction parameters from multiple rounds as input:

[0113] { , ..., };{ , ..., ;{ , ..., };

[0114] Historical pressure approximation coefficient: { n−3 , ..., n};

[0115] Current local tangency matrix K 局部 ;

[0116] Real-time performance parameters of water jet;

[0117] Output:

[0118] Adaptive iteration step size ΔS CNN ∈[0.05, 0.3]m;

[0119] Depth-Spacing Coupling Coefficient ∈[0.3, 0.6];

[0120] In the coupled model The optimal weight combination with F.

[0121] 4.3. Revised Rules:

[0122] like ∈[0.85, 1.05], keep the current parameters, ΔS CNN Take the minimum value;

[0123] like If the pressure is less than 0.85, it is determined to be too low; therefore, the lateral spacing of the boreholes should be increased. = +ΔS CNN ;

[0124] like A value >1.05 indicates excessive pressure; therefore, the lateral spacing of the boreholes should be reduced. = ΔS CNN ;

[0125] ΔS CNN The value range is 0.05~0.3m.

[0126] 4.4 Drilling Depth Linkage Correction:

[0127] ;

[0128] 4.5. Corrected transverse spacing of boreholes Meet safety constraints:

[0129] ∈[0.7×X base X base ×β];

[0130] Corrected drilling depth Meet safety constraints:

[0131] ∈[D min D max ];

[0132] Where β is the safety redundancy coefficient, D min To ensure the minimum depth for the cut-off effect, D max To protect the maximum allowable depth of the upper critical layer.

[0133] Step 5: Optimal Drilling Parameter Generation and Stable Execution:

[0134] 5.1 Repeat step 4 for multiple rounds of CNN feedback optimization until the convergence condition is met:

[0135] Convergence conditions during construction: 3 consecutive rounds of construction The values ​​all fall within the range of [0.85, 1.05], and the fluctuation range of the lateral spacing of the boreholes does not exceed 0.1m, and the fluctuation range of the borehole depth does not exceed 0.2m;

[0136] Acceptance criteria: Cut penetration rate ≥ 95%, stress release rate ≥ 90%, and roof collapse rate ≥ 90%.

[0137] 5.2 Generating optimal drilling parameters: optimal lateral spacing S opt Optimal drilling angle θ opt Optimal drilling depth D opt Optimal tipping cycle step size L opt .

[0138] 5.3. Perform subsequent top cutting and depressurization operations according to the optimal parameters, continuously collect data, and fine-tune the CNN model online once after each optimization stage is completed; if the geological conditions change suddenly and cause ξ to exceed the range of [0.85, 1.05], return to step 4 to restart the optimization.

[0139] The following example demonstrates the practical application of this method in the 8305 longwall face of a conventional coal mine, aiding in understanding the method. This longwall face has an advance length of 800m, mining the No. 3 coal seam, with a hard sandstone roof of Protodyakonov coefficient f=7.5 (relatively developed joints, joint density 2.5 joints / m²), uniaxial compressive strength σc=75MPa, rock fracture angle α=63°, roof thickness h=7.0m, and average roof stress σavg=28MPa. The rated parameters of the water jet roof cutting system are as follows: =50MPa =100L / min =20%, the rated pressure of the hydraulic fracturing system P is 15MPa; the mine environment is a conventional mine, and a CNN model is deployed using a mining embedded controller.

[0140] The specific steps of the mining CNN-enhanced adaptive feedback low-density borehole roof cutting and pressure relief control method in this embodiment are as follows:

[0141] Step 1: Multi-source data acquisition and preprocessing

[0142] 1.1 Collect core input data:

[0143] Geological parameters of the roof: f=7.5, Jd=2.5, σ avg =28MPa, h=7.0m, α=63°, σ c =75MPa;

[0144] Water jet system performance parameters: =42MPa =85L / min =16%;

[0145] Rated parameters of hydraulic fracturing system: =15MPa;

[0146] Mine environmental parameters: Conventional mine, roadway cross-section dimensions 5.0m × 3.5m;

[0147] Historical construction data: There is no historical data for the initial construction phase, so pre-trained model parameters are used.

[0148] 1.2 Data Preprocessing:

[0149] The above parameters are normalized and mapped to the [0,1] interval to form a one-dimensional parameter vector:

[0150] [f′,Jd′,σ avg ′,h′,α′,σ c ′, ′, ′, ′, ′,Environmental parameters′]=

[0151] [0.75,0.5,0.56,0.7,0.7,0.75,0.84,0.85,0.8,1.0,0.5]

[0152] Step 2: CNN Model Construction and Feature Extraction

[0153] 2.1 Start the lightweight CNN model. Input the above one-dimensional parameter vector into the parameter feature extraction network. After passing through 4 layers of one-dimensional convolutional layers with kernel sizes of 3, 5, 7, 5, and 3 pooling layers, output a 64-dimensional deep feature vector.

[0154] 2.2 The cutability coefficient prediction network processes the 64-dimensional feature vector through 3 fully connected layers, outputting:

[0155] Global cutability coefficient =0.52;

[0156] 32×32-dimensional local tangency coefficient matrix K 局部 K 局部 The high-cuttable zone with a strength ≥0.8 accounts for approximately 15% of the top slab area. 局部 The low cutability region (≤0.4) accounts for about 10%, and the rest are medium cutability regions.

[0157] 2.3 The safety redundancy level prediction network inputs mine environmental parameters and real-time roof monitoring data, stress mutation rate = 0.08 MPa / h, micro-seismic event frequency N = 2 times / h, and output redundancy β = 1.15.

[0158] 2.4 The model has been pre-trained based on the historical construction dataset of the mining area, with a pre-training loss value of 0.0008, which meets the accuracy requirements.

[0159] Step 3: Generate initial conservative drilling parameters

[0160] 3.1 Calculate the effective action coefficient of the water jet:

[0161] F=0.5×(42 / 50)+0.3×(85 / 100)+0.2×(16 / 20)=0.835;

[0162] 3.2 Take the conventional effective kerf radius =0.45m, empirical coupling coefficient =0.67, reference cutability coefficient =0.5, calculate the initial borehole spacing reference value:

[0163] =0.45×(1+0.67×0.835×(0.52 / 0.5))=0.712m;

[0164] 3.3 Conservative coefficient γ = 0.9, initial borehole lateral spacing:

[0165] S1 = 0.9 × 0.712 ≈ 0.64 m;

[0166] 3.4 Considering the rock strata fracture angle α = 63° and the local shearability distribution, δ = 1.0° is taken (the median value is taken in the medium shearability region), and the initial borehole angle is:

[0167] θ1 = 63° + 1.0° = 64°;

[0168] 3.5 Based on the top plate thickness h=7.0m, drilling angle θ1=64° and local cutability coefficient, the penetration depth ΔD=0.4m is taken, and the initial drilling depth is:

[0169] D1 = 7.0 / sin64° + 0.4 ≈ 8.19m;

[0170] 3.6 Generating the initial borehole location distribution based on the K-local matrix:

[0171] The highly cuttable zone is laid out with S1=0.64m;

[0172] The low cutability area is densified by 10%, with a spacing of 0.58m;

[0173] The spacing between the cuttable regions is 0.64m.

[0174] 3.7 Calculate the initial cutting cycle step distance, taking Ry = 1.4m:

[0175] L1=1.4×(1.2+0.67×0.835)=2.46m;

[0176] Step 4: Monitoring the fracturing process and CNN feedback correction

[0177] 4.1 Drilling, waterjet fracturing, and hydraulic fracturing operations were performed according to the initial conservative drilling parameters. The actual fracturing initiation pressure of the roof was collected using a high-pressure water pressure sensor. =11.2MPa, calculate the pressure approximation factor:

[0178] =11.2 / 15=0.747;

[0179] judge =0.747<0.85, indicating low pressure; therefore, the lateral spacing of the boreholes should be increased.

[0180] 4.2 CNN Parameter Optimization Prediction Network Input: Historical construction parameters from three rounds, historical pressure approximation coefficients, current local tangency coefficient matrix, and real-time water jet performance parameters; Output:

[0181] Adaptive iteration step size ΔS CNN =0.22m;

[0182] Depth-Spacing Coupling Coefficient =0.45;

[0183] In the coupled model The weight is adjusted to 0.45, and the F weight is adjusted to 0.55.

[0184] 4.3 According to the correction rules:

[0185] = +ΔS CNN =0.64 + 0.22 = 0.86m;

[0186] 4.4 Drilling Depth Linkage Correction:

[0187] =9.46;

[0188] 4.5 Safety Constraint Verification:

[0189] Spacing constraints: Upper bound Xbase×β=0.712×1.15=0.819m, lower bound 0.7×Xbase=0.498m;

[0190] Since it exceeds the upper limit by 0.819m, we take S2 ​​= 0.82m;

[0191] Depth constraint: Take D min =7.5m, D max =10.0m, D2=9.46m satisfies the constraints.

[0192] Step 5: Second round of optimization

[0193] 5.1 The second round of construction was carried out according to the corrected parameters S2=0.82m and D2=9.46m, and the actual crack initiation pressure was collected. =12.8MPa, calculate:

[0194] ξ = 12.8 / 15 = 0.853;

[0195] ξ=0.853∈[0.85,1.05], which is determined to be close to the rated pressure.

[0196] 5.2 CNN Parameter Optimization: Network Output ΔS CNN =0.05m, λ=0.45 (keep).

[0197] After revision 5.3:

[0198] = +0.05=0.87m

[0199] Security constraint verification: =0.87m exceeds the upper limit of 0.819m, so take... =0.82m.

[0200] 5.4 Deep Linkage Correction:

[0201] = ×(1+0.45×(0.82-0.82) / 0.82)=9.46m;

[0202] Step 6: Third round of optimization and convergence

[0203] 6.1, according to =0.82m At a depth of 9.46m, the third round of construction was carried out, and actual crack initiation pressure was collected. =13.1MPa, calculate:

[0204] ξ = 13.1 / 15 = 0.873;

[0205] ξ=0.873∈[0.85,1.05].

[0206] 6.2 The ξ values ​​for the three consecutive rounds are 0.747, 0.853, and 0.873, respectively, with the latter two rounds stabilizing within the target range; Spacing fluctuation: 0.64m in the first round → 0.82m in the second round, 0.82m in the second round → 0.82m in the third round, satisfying the convergence condition that the spacing fluctuation amplitude for the three consecutive rounds is ≤0.1m; Depth fluctuation: 8.19m in the first round → 9.46m in the second round, 9.46m in the second round → 9.46m in the third round, satisfying the convergence condition that the depth fluctuation amplitude for the three consecutive rounds is ≤0.2m.

[0207] 6.3 Continue to the next round of optimization and convergence until the ξ value of three consecutive construction rounds falls within the interval [0.85, 1.05], and finally generate the optimal parameters.

[0208] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. 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 be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A mobile high-pressure water jet method for precise roof cutting and pressure relief in mining, characterized in that, Includes the following steps: S1: Collect the geological parameters of the roof, the performance parameters of the water jet system, the rated parameters of the hydraulic fracturing system, and the mine environmental parameters of the construction area, and form input feature data after preprocessing; S2: Construct a CNN model, extract nonlinear coupling features between parameters based on the input feature data, output a local cutability coefficient matrix and a global cutability coefficient that characterize the difficulty of cutting the roof in different areas of the roof, and output a safety redundancy coefficient that matches the mine environment parameters. S3: Based on the global cutability coefficient, water jet system performance parameters and preset conventional effective cut radius, calculate the borehole spacing benchmark value, and generate initial borehole parameters according to a conservative coefficient. The initial borehole parameters include borehole lateral spacing, borehole angle, borehole depth, borehole position distribution and top cutting cycle step. S4: Perform water jet top cutting construction according to the initial drilling parameters. During the fracturing construction, the actual fracturing pressure of the hydraulic fracturing system is collected in real time, and the pressure approximation coefficient between the actual fracturing pressure and the rated pressure is calculated. S5: Input historical construction parameters, historical pressure approximation coefficients, current local cutability coefficient matrix and real-time water jet performance parameters into the CNN parameter optimization network, output adaptive iteration step size and depth-spacing coupling coefficient, and correct the borehole lateral spacing according to the comparison result between the current pressure approximation coefficient and the preset target interval, and simultaneously correct the borehole depth according to the coupling coefficient. S6: Repeat steps S4 to S5 for multiple rounds of iterative optimization until the pressure approximation coefficient stabilizes within the preset target range and the borehole parameter fluctuations meet the stability conditions, generate the optimal borehole parameters and execute subsequent construction.

2. The method for precise roof cutting and pressure relief using a mobile high-pressure water jet in mining, as described in claim 1, is characterized in that... In step S1, The geological parameters of the roof include: Protodyakonov coefficient f, joint density Jd, and average roof stress σ. avg Top plate thickness h, rock fracture angle α, uniaxial compressive strength σ c ; The performance parameters of the water jet system include: jet pressure. Jet flow rate Abrasive mixing ratio ; The rated parameters of the hydraulic fracturing system include: rated pressure. ; The mine environmental parameters include: mine disaster level, roadway cross-sectional dimensions, and support structure parameters.

3. The method for precise roof cutting and pressure relief using a mobile high-pressure water jet in mining, as described in claim 2, is characterized in that... The CNN model in step S2 includes: The parameter feature extraction network uses a 1D convolutional neural network, which contains 4 convolutional layers and 3 pooling layers, and outputs a 64-dimensional feature vector to extract nonlinear coupling features between different parameters and historical sequence trend features. The cutability coefficient prediction network includes three fully connected layers and outputs a 32×32-dimensional local cutability coefficient matrix, which is used to output the local cutability coefficient matrix and the global cutability coefficient based on the deep spatial feature vector. The safety redundancy prediction network, using the MobileNet architecture, is used to output the safety redundancy coefficient β based on mine environment parameters.

4. The method for precise roof cutting and pressure relief using a mobile high-pressure water jet in mining, as described in claim 1, is characterized in that... The initial borehole spacing reference value in step S3 The calculation formula is: ； in, This is the standard effective kerf radius; The global tangibility coefficient; For reference cutability coefficient; These are empirical coupling coefficients; The effective action coefficient of the water jet is calculated using the following formula: ； Where a, b, and c are weighting coefficients, and the sum of a, b, and c is 1; For the rated jet pressure, For the rated jet flow rate, The rated abrasive mixing ratio is given; and the value of F is not greater than 1.

5. A mobile high-pressure water jet precision roof cutting and pressure relief method for mining as described in claim 4, characterized in that, The initial borehole lateral spacing in step S3 The calculation formula is: ； in, This is a conservative coefficient, with a value of 0.8 to 0.

9. 0.8 is used for mines prone to rock bursts, and 0.9 is used for conventional mines. Initial drilling angle The calculation formula is: ； in, The value ranges from 0.8° to 1.2°, and ∈[60°, 90°]; The formula for calculating the initial borehole depth D1 is: ； in To allow for a certain depth to penetrate; The initial borehole location distribution is determined based on the local cutability coefficient matrix, and boreholes are densely distributed in areas where the cutability coefficient is lower than a preset threshold.

6. A method for precise roof cutting and pressure relief using a mobile high-pressure water jet in mining, as described in claim 1, is characterized in that... The pressure approximation coefficient The calculation formula is: ； in This represents the actual crack initiation pressure. The rated pressure; the pressure approximation coefficient The preset target range is [0.85, 1.05].

7. A method for precise roof cutting and pressure relief using a mobile high-pressure water jet in mining, as described in claim 6, is characterized in that... The rule for correcting the lateral spacing of the boreholes in step 5 is based on the pressure approximation coefficient. The value of ΔS is used to adjust the CNN parameters and optimize the adaptive iteration step size ΔS of the network output. CNN Specifically: like ∈[0.85, 1.05], keep the current parameters, ΔS CNN Take the minimum value; like If the pressure is less than 0.85, it is determined to be low; therefore, the lateral spacing of the boreholes should be increased. = +ΔS CNN ; like A value >1.05 indicates excessive pressure; therefore, the lateral spacing of the boreholes should be reduced. = ΔS CNN ; ΔS CNN The value range is 0.05~0.3m.

8. A method for precise roof cutting and pressure relief using a mobile high-pressure water jet in mining, as described in claim 7, is characterized in that... The formula for correcting the borehole depth in step 5 is: ； in, The depth-space coupling coefficient for optimizing the synchronous output of the CNN parameters is set, with a value ranging from 0.3 to 0.

6. Corrected borehole lateral spacing Meet safety constraints: ∈[0.7×X base ,X base ×β]; Corrected drilling depth Meet safety constraints: ∈[D min ,D max ]; Where β is the safety redundancy coefficient, D min To ensure the minimum depth for the cut-off effect, D max To protect the maximum allowable depth of the upper critical layer.

9. A mobile high-pressure water jet precision roof cutting and pressure relief method for mining as described in claim 8, characterized in that, In step 6, the pressure approximation coefficient The criteria for determining whether borehole parameters are stable within a preset target range and meet the stability conditions are as follows: Three consecutive rounds of construction The values ​​all fall within the range of [0.85, 1.05], and the fluctuation range of the lateral spacing of the boreholes does not exceed 0.1m, and the fluctuation range of the borehole depth does not exceed 0.2m.

10. A mobile high-pressure water jet precision roof cutting and pressure relief method for mining as described in claim 9, characterized in that, The optimal drilling parameters include the optimal lateral spacing S. opt Optimal drilling angle θ opt Optimal drilling depth D opt Optimal tipping cycle step size L opt The method also includes continuously monitoring the cut penetration rate, stress release rate, and roof collapse rate for effect acceptance after generating the optimal drilling parameters; the CNN model is pre-trained based on the historical construction dataset of the mining area, and the pre-training loss function adopts the mean square error, and it stops when the loss value is less than 0.001; incremental learning fine-tuning is performed after each optimization stage is completed during the construction process.

Images