Method for accurately judging drilling ruler and stratum trend through multiple parameters in underground coal mine

Abstract

The invention provides a method for accurately judging a drilling ruler and a stratum trend through multiple parameters in an underground coal mine. The method comprises the steps that firstly, multi-parameter data are collected and preprocessed; and step 2, real-time efficiency evaluation and lithology identification: taking the multi-parameter data preprocessed in the step 1 as input parameters, training the improved ResNet50-multiple LSTM model, optimizing network parameters through a back propagation algorithm, and taking the trained improved ResNet50-multiple LSTM model as a real-time efficiency evaluation and lithology identification model. The improvement method for improving the ResNet50 network comprises the step of respectively introducing CBAM into a residual block I and a residual block II of the ResNet50 network. The real-time efficiency evaluation method is as follows: the effective drilling time ratio = effective drilling time / total drilling operation time. Lithology includes coal seams, sandstones and fractures. And step 3, effective footage calculation. And step 4, stratum trend prediction and dynamic correction. According to the invention, accurate control of parameters is realized.

Description

Technical Field

[0001] This invention belongs to the field of underground drilling construction technology in coal mines, and relates to drilling rigs, specifically to a method for accurately determining drilling footage and formation direction in underground coal mines using multiple parameters. Background Technology

[0002] In underground coal mine drilling operations, precise control of the borehole trajectory, accurate determination of drilling footage, and real-time monitoring of geological formation information are core elements for improving drilling efficiency, ensuring operational safety, and achieving efficient resource extraction. However, traditional drilling techniques mainly rely on manual experience to adjust drilling parameters and determine geological information, which has significant shortcomings and cannot meet the needs of modern intelligent and precise coal mining.

[0003] Traditional coal mine drilling technology relies heavily on manual experience to judge drilling conditions. Monitoring core parameters (such as drill pipe rotation speed, drill pressure, and drilling speed) depends on operator perception, leading to significant subjective errors. For example, when drill bit wears or there are abrupt changes in formation lithology, the response time for manually adjusting drilling parameters is delayed, easily causing accidents such as drill string jamming and borehole collapse. Under complex geological conditions, such as fault zones, fracture zones, or high-hardness rock formations, manual experience is insufficient to accurately judge formation changes, causing borehole trajectories to deviate from the designed path and affecting resource extraction efficiency.

[0004] Traditional methods for obtaining stratigraphic information rely on limited resources. They primarily depend on geological compasses, localized strata exposures revealed through tunnels, or core sampling to determine stratigraphic strike and lithology. These methods have limitations in three-dimensional spatial positioning and cannot comprehensively and accurately reflect underground geological conditions. In complex geological structures such as folds and faults, traditional methods cannot obtain key parameters like dip and azimuth in real time, leading to a lack of scientific basis for borehole design and increasing drilling risks.

[0005] To overcome the limitations of traditional drilling techniques, Measurement While Drilling (MWD) technology has emerged. This technology integrates multiple sensors to collect multi-dimensional data in real time, including borehole trajectory, formation physical parameters, and drill string attitude, providing precise data support for drilling operations. However, MWD technology still faces many challenges in practical applications. In fault fracture zones, high-hardness rock formations, or areas with large water inflows, drill string vibration causes random drift in sensor data, affecting measurement accuracy. The lithological variations and fracture development in complex formations place higher demands on the adaptability of MWD systems, which current technologies cannot fully meet. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the present invention aims to provide a method for accurately determining drilling footage and formation direction in coal mines using multiple parameters, thereby solving the technical problem that the accuracy of existing methods for determining drilling footage and formation direction needs further improvement.

[0007] To solve the above-mentioned technical problems, the present invention adopts the following technical solution.

[0008] A method for accurately determining drilling footage and formation strike in coal mines using multiple parameters, comprising the following steps.

[0009] Step 1: Multi-parameter data acquisition and preprocessing.

[0010] Multi-parameter data is collected, and the collected multi-parameter data samples are preprocessed using wavelet transform to eliminate noise interference.

[0011] The multi-parameter data includes drilling rig status parameters, borehole trajectory parameters, geophysical parameters, and drilling dynamics parameters.

[0012] Step 2: Real-time efficiency assessment and lithology identification.

[0013] Step 201: Using the multi-parameter data preprocessed in Step 1 as input parameters, train the improved ResNet50-multiple LSTM model, optimize the network parameters through backpropagation algorithm, and use the trained improved ResNet50-multiple LSTM model as a real-time efficiency evaluation and lithology identification model.

[0014] The improvement method for the ResNet50 network is as follows: CBAM is introduced into Residual Block I and Residual Block II of the ResNet50 network respectively.

[0015] The real-time efficiency evaluation method is as follows: effective drilling time percentage = effective drilling time / total drilling operation time.

[0016] The lithology mentioned includes coal seams, sandstone, and fractures.

[0017] Step 202: After the real-time multi-parameter data is preprocessed in Step 1, it is input into the real-time efficiency evaluation and lithology identification model obtained in Step 201, and the real-time efficiency evaluation and lithology identification results are output.

[0018] Step 3: Calculate the effective advance.

[0019] Step 301: Based on the lithology identification results obtained in Step 2, determine whether the drilling section displacement is valid according to the displacement determination method; the drilling section displacement includes rod changing displacement, stuck drill displacement, drill retraction displacement, and idle displacement.

[0020] Step 302: Based on the lithology identification results obtained in Step 2 and the displacement judgment results obtained in Step 301, calculate the effective advance.

[0021] Step 4: Stratigraphic strike prediction and dynamic correction.

[0022] Step 401, Calculation of stratigraphic strike.

[0023] The dip direction of the strata is perpendicular to the strike of the strata. At the borehole trajectory point (x,y,z), the plane equation Ax+By+Cz+D=0 is fitted through the strata interface points of 3 different boreholes. The strike angle φ of the strata strike satisfies: φ=arctan(-A / B).

[0024] Step 402, Real-time trajectory fitting and deviation analysis.

[0025] The apex angle and azimuth angle measured while drilling will be converted into three-dimensional coordinates. , , ), Calculation of the deviation between the trajectory and the design path .

[0026] Step 403, dynamic correction.

[0027] When the location changes, dynamic stratigraphic correction is performed. The specific method for dynamic correction is as follows: .

[0028] Step 5: Multi-source parameter control.

[0029] Step 501: Based on the real-time efficiency evaluation results obtained in Step 2, dynamically adjust the drilling pressure through an adaptive control algorithm to improve efficiency.

[0030] Step 502: Based on the lithology judgment results obtained in Step 2, dynamically adjust the drilling pressure and rotation speed through an adaptive control algorithm to improve efficiency.

[0031] Step 503: Based on the lithology judgment results obtained in Step 2, dynamically adjust the drilling rig power through an adaptive control algorithm to reduce energy consumption.

[0032] Compared with the prior art, the present invention has the following technical effects.

[0033] (I) This invention achieves precise parameter control. It significantly improves the precision, intelligence, and safety of drilling operations. In the future, with the integrated application of technologies such as 5G communication and digital twins, this invention will further evolve towards "fully automated and unmanned" operations, promoting high-quality development in the coal industry.

[0034] (II) The real-time lithology identification accuracy of the present invention exceeds 95%: It integrates cross-validation of gamma rays (to distinguish coal seams from sandstone), resistivity (to classify sandstone from limestone) and vibration signals (to identify lithological hardness), and combines the improved ResNet50-multiple LSTM network model, reducing the lithology identification time from 5 minutes in the traditional method to within 3 seconds.

[0035] (III) The present invention achieves both efficiency and safety through efficiency optimization and adaptive control algorithm.

[0036] (IV) The present invention dynamically adjusts the drilling rig power according to the rock hardness to avoid energy waste caused by repeated drilling.

[0037] (V) The present invention makes a comprehensive breakthrough in the ability to correct borehole trajectory deviation and determine stratum strike: precise trajectory control under complex geological conditions. Traditional methods are prone to failure due to trajectory loss of control in complex geological conditions such as faults and fracture zones. The present invention achieves precise obstacle avoidance through the following technologies.

[0038] (VI) This invention can realize intelligent determination of stratum strike: based on the gamma value and trajectory data of three or more borehole interface points, the least squares method is used to fit the stratum strike plane equation, and combined with geological structural constraints (such as the safe distance of collapse column ≥ 5m), dynamic obstacle avoidance path planning is generated.

[0039] (VII) This invention can achieve automatic filtering of invalid displacement: by analyzing the slope change of the displacement-time curve, it can distinguish between effective drilling (stable slope) and idle / retracting drilling (abrupt slope), and automatically remove invalid footage data.

[0040] (VIII) The present invention can significantly improve the accuracy of drilling status identification and invalid footage determination: traditional methods lack drilling dynamics signal analysis and cannot distinguish abnormal states such as idle rotation and stuck drill. Attached Figure Description

[0041] Figure 1 This is a schematic diagram of the drilling rig feed process of the present invention.

[0042] Figure 2 This is a schematic diagram of the drilling trajectory of the present invention.

[0043] Figure 3 This is a schematic diagram showing the rock type, gamma ray, and drilling time at the bottom of the borehole drilled by the drilling rig of the present invention.

[0044] Figure 4 This is a schematic diagram of the structure of the improved ResNet50-multiple LSTM model of the present invention.

[0045] Figure 5 This is a schematic diagram of the traditional ResNet50 network residual block I and residual block II structure.

[0046] Figure 6 This is a schematic diagram of the improved ResNet50 network residual block I and residual block II of the present invention.

[0047] Figure 7 This is a CBAM diagram of the improved ResNet50 network of the present invention.

[0048] The specific content of the present invention will be further explained in detail below with reference to the embodiments. Detailed Implementation

[0049] It should be noted that, unless otherwise specified, all devices, instruments, functions, algorithms and networks in this invention are based on devices, instruments, functions, algorithms and networks known in the prior art.

[0050] The Necessity of Multi-Parameter Fusion Technology: Single-parameter measurements cannot meet the precise judgment requirements under complex geological conditions. Multi-parameter fusion technology, by integrating data from multiple sensors, enables comprehensive monitoring and precise control of drilling operations, becoming crucial for improving drilling efficiency and ensuring operational safety. 1. Precise Determination of Formation Strike: By combining multi-source data such as gamma-ray measurement, trajectory measurement, and resistivity measurement, a network model is constructed to accurately determine the formation strike. This helps optimize borehole trajectory design and improve resource extraction efficiency. By monitoring parameters such as formation dip and azimuth in real time, the borehole direction can be adjusted promptly to ensure the borehole extends along the designed path. 2. Improved Cutting Footage Accuracy: By monitoring parameters such as drill pressure and torque in real time, drilling efficiency is evaluated, providing a basis for optimizing drilling parameters.

[0051] The drilling rig and mud pump truck in this invention have a total of 3 pumps: Pump I, for rapid feed, rapid rotation, rapid lifting, and track travel; Pump II, for slow feed, slow rotation, and slow lifting; and Pump III, for clamping device, unhooking device, and chuck brake.

[0052] The hydraulic oil pumps in this invention are as follows: Pump I has a maximum displacement of 160 mL / r, a working pressure of 28 MPa, and a maximum speed of 210 r / min; Pump II has a maximum displacement of 71 mL / r, a working pressure of 26 MPa, and a maximum speed of 220 r / min; Pump III has a maximum displacement of 28 mL / r, a working pressure of 21 MPa, and a maximum speed of 300 r / min.

[0053] The feeding system in this invention is as follows Figure 1 As shown, the main components are the rotary circuit and the drilling circuit, including fast control and slow control, in order to achieve rotary feed, directional feed, and combined feed.

[0054] The drilling rig parameters in this invention are as follows: the rotational speed of the power head is mainly measured by a speed encoder; the rotational torque and feed / pull-out pressure are measured by a pressure sensor; and the feed / pull-out speed is measured by a proximity switch. The mud pump truck parameters in this invention are as follows: the pump pressure is mainly measured by a pressure sensor; and the pump flow rate is measured by a flow sensor. The auxiliary equipment parameters in this invention include: the return oil pressure is measured by a pressure sensor.

[0055] This invention employs an engineering parameter measuring instrument to collect data on drilling pressure, torque, internal and external annular pressure, drilling speed, vibration, and temperature of the drilling tools. It also employs a trajectory inclinometer to measure inclination angle, azimuth angle, and tool face angle.

[0056] In this invention, the drilling tool assembly for a downhole directional drilling rig consists of: a Ф120mm directional drill bit + a Ф89mm hydraulic screw motor + a Ф89mm lower non-magnetic drill rod + a Ф89mm engineering parameter measurement probe + a Ф89mm upper non-magnetic drill rod + a Ф89mm integral spiral measuring drill rod + a Ф89mm integral spiral measuring drill rod + a Ф89mm cable guide.

[0057] In this invention, the main drilling equipment for downhole operations includes: a ZDY12000LD directional drilling rig, a BLY500 mud pump truck, a Ф89mm cable-operated water supply system, a mining engineering parameter measurement-while-drilling system, a Ф89mm integral spiral measurement-while-drilling drill rod, a Ф89mm hydraulic screw drill bit, a Ф89mm upper non-magnetic drill rod, a Ф89mm lower non-magnetic drill rod, and a Ф120mm directional drill bit. The layout of the drilling rig's operating sensors is shown in Table 1. The data table for the drilling rig's hydraulic pumps is shown in Table 2. The formation structure for drilling operations using the drilling rig is shown in Table 3.

[0058] Table 1. Layout of drilling rig operation sensors according to the present invention

[0059] Table 2 Data Sheet of the Drilling Rig Hydraulic Pump of the Present Invention

[0060] Table 3. Drilling formation structure of the drilling rig of the present invention

[0061] In this invention, the drilling process of the downhole directional drilling rig includes the following steps.

[0062] 1. When designing the drilling tool assembly at the borehole location, do not connect the screw rod or magnetic drill rod. Only connect the drill bit to drill to a depth of 10 meters. Pull out the drill string, install the casing, and then cement the casing securely.

[0063] 2. Install wellhead drainage and slag removal devices.

[0064] 3. This set includes drill bits, screws, and non-magnetic components.

[0065] 4. After the entire drilling tool is connected, start the pump to test whether the mud pulse inclination meter has a signal.

[0066] 5. The mud pulse instrument signal is normal, the pump pressure value is stable, and directional drilling begins.

[0067] 6. After directional drilling is completed, the composite rotary pullback is pulled back 20cm for punching, and the pump is started for inclination measurement.

[0068] 7. After the inclination measurement is completed, stop the pump, unload the water, and connect a single pipe.

[0069] 8. After the single connection is completed, connect the water supply, start the pump, and lower the drill string to begin directional drilling. Drilling rig data extraction: Drilling rig parameters (rotation speed is obtained using an encoder, drilling rig torque is obtained using a pressure sensor, feed and pull-out pressure sensors, mud pump pressure and flow sensors).

[0070] In this invention, the drilling trajectory of the downhole directional drilling rig is as follows: Figure 2 As shown. The method for adjusting the drilling trajectory includes the following steps.

[0071] 1. When encountering a fault, based on the stratigraphic strike result obtained in step four, drill upwards or downwards to find the target layer, locate the target layer, determine whether it is a normal fault or a reverse fault, and calculate the fault displacement accordingly.

[0072] 2. When encountering a fracture zone, based on the formation strike results obtained in step four, observe the pump pressure drop, the absence of water return at the borehole opening, and the absence of slag return at the borehole opening. Increase the pump discharge rate and increase the drilling speed to penetrate the fracture zone.

[0073] 3. When encountering a collapse column, based on the stratigraphic strike results obtained in step four, locate the target stratum using the borehole trajectory. If the distance exceeds 5m, stop drilling, reorient the azimuth, and restart drilling to avoid the collapse column. Locate the target coal seam.

[0074] In this invention, the schematic diagram of the bottom lithology, gamma ray, and drilling time of the downhole directional drilling rig is as follows: Figure 3 As shown.

[0075] The following are specific embodiments of the present invention. It should be noted that the present invention is not limited to the following specific embodiments. All equivalent modifications made based on the technical solutions of this application fall within the protection scope of the present invention.

[0076] Example: This embodiment provides a method for accurately determining drilling footage and formation strike in coal mines using multiple parameters. The method includes the following steps.

[0077] Step 1: Multi-parameter data acquisition and preprocessing.

[0078] Multi-parameter data is collected, and the samples of the collected multi-parameter data are preprocessed using wavelet transform to eliminate noise interference. The multi-parameter data includes drilling rig status parameters, borehole trajectory parameters, geophysical parameters, and drilling dynamics parameters.

[0079] In step one, the drilling rig status parameters include drilling pressure, rotational speed, torque, hydraulic pump pressure, and mud pump pressure. In this embodiment, the specific monitoring method for the drilling rig status parameters is observation of parameters such as drilling pressure on the drilling rig's instrument panel. The specific function of the drilling rig status parameters is to reflect the drilling rig's working status and energy transfer efficiency, and to be used for determining the effectiveness of drilling footage and evaluating drilling efficiency.

[0080] Drill pressure (DP) refers to the axial pressure setpoint of the drill bit, used to break up rock cuttings. It is represented by f(t) and the unit is kN. In soft coal seams, excessive DP can lead to borehole collapse; in hard rock, insufficient DP results in slow drilling speed. The feed rate is displayed in real time by rapidly monitoring the feed and rotation values ​​of the hydraulic pump station I on the drilling rig.

[0081] Rotational speed: The rotational speed of the power head, represented by n(t) and measured in r / min, is used for real-time rock breaking. The rotational speed of the power head is displayed in real time by an encoder installed near the power head.

[0082] Torque: When a drill bit is drilling and breaking through rock strata, it receives a reverse force during rotation, denoted by T(t) and measured in kN·m. The torque value is displayed in real time via an engineering parameter torque sensor.

[0083] Mud pump pressure: reflects the degree to which the drill bit breaks through the rock formation and the cuttings inside the borehole, expressed as Q(t), in MPa.

[0084] In step one, the borehole trajectory parameters include well inclination, azimuth, tool face angle, and borehole depth. In this embodiment, the specific monitoring method for the borehole trajectory parameters is an engineering parameter instrument. The specific function of the borehole trajectory parameters is to draw a three-dimensional borehole trajectory, determine whether the borehole extends along the designed path, and identify formation strike deviations.

[0085] Inclination angle (a): This reflects the angle between the borehole axis and the vertical line, indicating the degree of inclination of the borehole trajectory. The unit is °.

[0086] Azimuth (φ): The angle between the horizontal projection of the borehole and the due north direction, reflecting the horizontal direction of the borehole, and the unit is °.

[0087] Tool face angle (γ): The angle between the cutting face of the drill bit and the drilling plane, used for directional drilling adjustment, in degrees.

[0088] Hole depth (L): Cumulative length of borehole, in meters.

[0089] In step one, the geophysical parameters include natural gamma values ​​and resistivity. In this embodiment, the specific monitoring method is a natural gamma sensor, a resistivity probe, and an acoustic logging device. The specific function is to identify lithology (coal seam, sandstone, limestone) and structure (collapse column, fracture) and locate the stratigraphic interface.

[0090] In step one, drilling dynamics parameters include vibration signals. In this embodiment, the specific monitoring method is a vibration sensor, which reflects changes in drilling resistance, abrupt changes in lithology, and the risk of stuck or lost drill bit, assisting in determining the effectiveness of the drilling footage.

[0091] Step 2: Real-time efficiency assessment and lithology identification.

[0092] Step 201: Using the multi-parameter data preprocessed in Step 1 as input parameters, train the improved ResNet50-multiple LSTM (Long Short-Term Memory) model, optimize the network parameters through backpropagation algorithm, and use the trained improved ResNet50-multiple LSTM model as a real-time efficiency evaluation and lithology identification model.

[0093] The real-time efficiency evaluation method is: effective drilling time percentage = effective drilling time / total drilling operation time.

[0094] Lithology includes coal seams, sandstone, and fractures.

[0095] In this embodiment, the structural diagram of the improved ResNet50-multiple LSTM model is shown below. Figure 4 As shown. The improvement method for the ResNet50 network is to introduce CBAM (Convolutional Block Attention Module) into Residual Block I and Residual Block II of the ResNet50 network respectively.

[0096] The ResNet50 (Residual Network with 50 layers) architecture consists of six core modules arranged in an orderly fashion. The first module uses a 7×7 convolutional kernel (stride set to 2, padding value set to 3, output channels 64), followed by batch normalization (BN) layers and ReLU activation function to enhance feature stability, and then a 3×3 max pooling layer (stride 2, padding 1) to achieve spatial dimensionality reduction. The subsequent four stages mainly consist of two types of residual modules stacked alternately: identity residual modules (Type I) and convolutional residual modules (Type II). Each stage contains one convolutional residual module as the basic unit, while the number of identity residual modules increases with each stage, from 2, 3, 5, to 2. Both types of residual modules use a three-layer convolutional structure of 1×1-3×3-1×1, with each layer followed by BN and ReLU to improve nonlinear fitting ability. The final module integrates spatial features through global average pooling and completes the multi-class classification task through a fully connected layer.

[0097] Traditional ResNet50 network feature extraction primarily relies on convolutional layers and residual connections to extract spatial features from images. While residual connections alleviate the vanishing gradient problem in deep networks, feature extraction remains limited to the spatial dimension, and the processing of all feature channels and spatial locations is uniform, lacking specificity. For complex scenes or images with interfering factors, traditional ResNet50 may fail to fully extract key features, resulting in limited classification performance.

[0098] In this embodiment, residual block optimization of the ResNet50 network is shown in the schematic diagram of the traditional ResNet50 network residual block I and residual block II. Figure 5 As shown, CBAM is introduced into residual block I and residual block II respectively, as follows: Figure 6 As shown.

[0099] In this embodiment, as Figure 7 As shown, CBAM includes channel attention and spatial attention mechanisms, which can adaptively adjust the weights of feature channels and spatial locations. Channel attention mechanism: Through global average pooling and global max pooling operations, it captures the dependencies between feature channels and assigns different weights to each channel, enabling the model to focus on more important feature channels. Spatial attention mechanism: Building upon the channel attention mechanism, it further captures the dependencies between different spatial locations on the feature map and assigns different weights to each spatial location, enabling the model to focus on more critical spatial regions. With the introduction of CBAM, the improved ResNet50 network can more accurately extract key features from images, suppress background noise and interference information, and improve the accuracy and robustness of feature extraction.

[0100] In this embodiment, as Figure 7 As shown, the CBAM attention mechanism is a fusion structure of channel attention and spatial attention, with its core goal of improving the performance of the ResNet50 network. It enhances the network's performance in visual tasks by adaptively adjusting the channel weights and spatial distribution of the input feature map. Its structure can be seen in the relevant illustrations.

[0101] Traditional ResNet50 networks lack temporal processing capabilities. ResNet50 cannot directly handle temporal dependencies in sequential data, thus limiting its performance in tasks that require temporal dependencies.

[0102] In this embodiment, an improved ResNet50 network with multiple LSTMs is introduced: the feature maps extracted by the improved ResNet50 are used as input to the multiple LSTM. LSTM units can capture long-term dependencies in sequence data and control the flow of information through gating mechanisms, thereby extracting dynamic changes and temporal features between frames. The advantages of multiple LSTMs: compared to a single LSTM unit, multiple LSTMs can learn more complex temporal patterns. Stacking LSTMs enhances the model's expressive power by increasing network depth. Fusion of temporal and spatial features: the spatial features extracted by the improved ResNet50 are fused with the temporal features extracted by the multiple LSTMs within the model, forming a more comprehensive feature representation, enabling the model to more accurately understand the overall meaning of the sequence data.

[0103] In this embodiment, the improved model supports an end-to-end training approach, meaning the entire process from raw input data to the final classification result can be completed within a unified framework. This training method simplifies the training process and improves training efficiency.

[0104] In this embodiment, the main algorithmic idea of ​​LSTM and the sample input set are as follows. Consistent with time In the output stage, the sigmoid activation function is used to generate the forget gate output. The output value of the sigmoid activation function varies between (0,1).

[0105] ; In the formula: This represents the sigmoid activation function; Indicates weight; The output state (also known as the hidden state) of the LSTM network at the previous time (t-1) is a cumulative representation of the network's historical time-series information. Represents the sample input set; Indicates bias.

[0106] In step 201, the improved ResNet50-multiple LSTM model extracts effective features from the improved ResNet50 network. The feature map output by the ResNet50 network is fused with feature data through multiple LSTMs, and finally outputs a real-time efficiency evaluation and lithology identification model through the ReLU activation function.

[0107] Effective features include azimuth gamma, resistivity, and drilling dynamics parameters; drilling dynamics parameters include vibration signal a(t) and displacement difference δ(t).

[0108] Azimuth Gamma: When the drill string rotates, the azimuth sensor calculates the drill string attitude in real time, and combines the time series data of gamma counting to determine the azimuth of the window (such as the upper / lower interface).

[0109] Resistivity: Characterizes the electrical conductivity of rock formations, measured in Ω. m. Coal seam resistivity (10–100 Ω) m) higher than sandstone (1~10Ω) m).

[0110] Vibration signal a(t): Triaxial vibration acceleration of the drill bit when breaking rock, in g. High-frequency vibration (100-500Hz) is significant during hard rock drilling, while low-frequency vibration (10-50Hz) is dominant in soft rock.

[0111] Displacement difference δ(t): The difference between the displacement of the power head and the actual advance, reflecting drill pipe slippage or compression, in mm. When δ(t) > 5 mm, it indicates a potential risk of stuck drill bit, and the drill rig displacement sensor is used to determine whether the power head is making any advance.

[0112] In this embodiment, for multi-channel data information, the improved ResNet50 can extract effective feature information, and LSTM processes time series data information. The fused improved ResNet50-LSTM model achieves accurate classification and discrimination of drilling data and engineering parameter data.

[0113] In this embodiment, after training, the model needs to be tested and evaluated. An independent test set is used to test the trained model, and the model performance is evaluated using metrics such as accuracy, precision, and recall (R). These metrics comprehensively reflect the model's classification ability and stability.

[0114] Step 202: After the real-time multi-parameter data is preprocessed in Step 1, it is input into the real-time efficiency evaluation and lithology identification model obtained in Step 201, and the real-time efficiency evaluation and lithology identification results are output.

[0115] Step 3: Calculate the effective advance.

[0116] Step 301: Based on the lithology identification results obtained in Step 2, determine whether the displacement of the drilling section is valid according to the displacement determination method; the displacement of the drilling section includes rod changing displacement, stuck drill displacement, drill retraction displacement, and idle displacement.

[0117] In step 301, the displacement determination method is as follows.

[0118] Step 30101, Rod replacement displacement: When drilling pressure f(t) = 0, rotational speed n(t) = 0, vibration signal a(t) = 0, displacement When the displacement of the pump rod is between 0.75 and 3.0 m, the pump pressure Q(t) is 0, and the torque T(t) is 0, then the displacement of the pump rod is determined to be invalid, i.e., the determination index is... Otherwise, the displacement of the rod replacement is determined to be a valid displacement, i.e., the determination index. .

[0119] Step 30102, stuck drill displacement: When the drilling pressure f(t) suddenly increases, the rotational speed n(t) = 0, the vibration signal a(t) = 0, the displacement change δ(t) = 0 m, the mud pump pressure Q(t) suddenly increases, and the torque T(t) suddenly increases, then the stuck drill displacement is determined to be an invalid displacement, i.e., the judgment index. Otherwise, the stuck drill displacement is determined to be a valid displacement, i.e., the judgment index. .

[0120] In this embodiment, a sudden increase refers to a gradual increase in drilling pressure and mud pump pressure.

[0121] Step 30103, Drill Retraction Displacement: When drilling pressure f(t) = 0, rotational speed n(t) > 0, vibration signal a(t) = 20Hz, displacement change δ(t) = 0m, mud pump pressure Q(t) suddenly increases and torque T(t) ≥ 2KN, then the drill retraction displacement is determined to be an invalid displacement, i.e., the determination index. Otherwise, the retraction displacement is determined to be a valid displacement, i.e., the determination index. .

[0122] Step 30104, Idle Displacement: When drilling pressure f(t) = 0, rotational speed n(t) > 0, vibration signal a(t) = 22Hz, displacement change δ(t) = 0m, mud pump pressure Q(t) suddenly increases and torque T(t) ≤ 2.4KN, then the idle displacement is determined to be invalid displacement, i.e., the judgment index. Otherwise, the idle displacement is determined to be a valid displacement, i.e., the determination index. .

[0123] Step 302: Based on the lithology identification results obtained in Step 2 and the displacement judgment results obtained in Step 301, calculate the effective advance.

[0124] In step 302, the effective footage (Eruler) is the sum of the displacements of each effective drilling segment, i.e., the effective footage. ; In the formula: Indicates effective advance; Indicates the first Displacement sensor records for each drilling section; Indicates the first Indicators for determining each drilling section; This indicates that the statement is valid. Indicates invalid.

[0125] Step 4: Stratigraphic strike prediction and dynamic correction.

[0126] Step 401, Calculation of stratigraphic strike.

[0127] The dip direction of the strata is perpendicular to the strike of the strata. At the borehole trajectory point (x,y,z), the plane equation Ax+By+Cz+D=0 is fitted through the strata interface points of 3 different boreholes. The strike angle φ of the strata strike satisfies: φ=arctan(-A / B). In the formula: A represents the component of the normal vector on the x-axis, reflecting the east-west dip trend of the strata. B represents the component of the normal vector on the y-axis, reflecting the dipping trend of the strata in the north-south direction; C represents the coefficient in the z-axis direction of the plane equation, which directly corresponds to the dip of the stratum interface (related to the dip angle of the stratum). When C≠0, there is an angle between the plane and the z-axis (vertical direction), that is, the stratum is dipped. D represents the constant term of the plane equation, which is determined by solving the x, y, and z coordinates of the three borehole formation interface points. It is used to fully define the position of the plane in three-dimensional space and ensure that the plane can accurately fit the three formation interface points.

[0128] Step 402, Real-time trajectory fitting and deviation analysis.

[0129] The apex angle and azimuth angle measured while drilling will be converted into three-dimensional coordinates. , , ), Calculation of the deviation between the trajectory and the design path ; ; ; ; ; In the formula: This indicates the deviation between the trajectory and the designed path; This represents the x-axis coordinate (east-west direction) of the borehole trajectory point at time t in the three-dimensional coordinate system during drilling measurements, reflecting the real-time east-west position of the borehole on the horizontal plane. This represents the y-axis coordinate (north-south direction) of the borehole trajectory point at time t in the three-dimensional coordinate system during drilling measurements, reflecting the real-time north-south position of the borehole on the horizontal plane. This represents the z-axis coordinate (vertical direction) of the borehole trajectory point at time t in the three-dimensional coordinate system during drilling measurements, reflecting the real-time depth of the borehole (the z-value increases with depth). This represents the time variable of measurements while drilling, characterizing the time progress of the drilling operation, and corresponding to the cumulative time from the start of drilling to the current moment; This represents the x-axis coordinate (east-west direction) at time t in the design path, i.e., the east-west target position that the borehole should reach at that time node; This represents the y-axis coordinate (north-south direction) at time t in the design path, which is the north-south target position that the borehole should reach at that time node; This represents the z-axis coordinate (vertical direction) at time t in the design path, which is the target depth that the borehole should reach at that time node; The x-axis coordinate (east-west direction) represents the starting point of the borehole, which is the initial east-west position at the start of drilling; The y-axis coordinate (north-south direction) represents the starting point of the borehole, which is the initial north-south position at the start of drilling. The z-axis coordinate (vertical direction) represents the starting point of the borehole, which is the initial depth at the start of drilling (usually set to the ground or reference surface depth). The apex angle (inclination angle) of the borehole at time t is the angle between the borehole axis and the vertical direction, which determines the vertical drilling trend of the borehole. It represents the azimuth angle of the borehole at time t, that is, the angle between the projection of the borehole axis on the horizontal plane and the due north direction, which determines the horizontal drilling direction of the borehole (related to the strike angle φ of the strata).

[0130] Step 403, dynamic correction.

[0131] When the location changes, dynamic stratigraphic correction is performed. The specific method for dynamic correction is as follows: ; In the formula: Indicates the apparent tilt angle; Indicates the angle of inclination; Indicates the angle between the borehole azimuth and the strike of the formation; This represents the correction factor that takes into account the anisotropy of the rock strata.

[0132] Step 5: Multi-source parameter control.

[0133] Step 501: Based on the real-time efficiency evaluation results obtained in Step 2, dynamically adjust the drilling pressure through an adaptive control algorithm to improve efficiency.

[0134] In step 501, the specific constraints for dynamic adjustment are as follows: when the effective drilling time percentage is ≤75%, the drilling pressure is gradually increased at a rate of 1kN per step until the efficiency recovers or the maximum drilling pressure limit (e.g., 15kN) is reached.

[0135] Specifically, in the drilling of limestone formations in a certain mine, this mode increased the net footage per shift from 80m to 112m, improving efficiency by 40%.

[0136] In this embodiment, the adaptive control algorithm is a commonly used adaptive control algorithm known in the art.

[0137] Step 502: Based on the lithology judgment results obtained in Step 2, dynamically adjust the drilling pressure and rotation speed through an adaptive control algorithm to improve efficiency.

[0138] In step 502, the specific constraints for dynamic adjustment are as follows: when the lithology is coal seam, adjust the drilling pressure f(t) = 5~10kN and the rotation speed n(t) = 60~100r / min (high rotation speed and low drilling pressure to reduce hole collapse); when the lithology is sandstone, adjust the drilling pressure f(t) = 11~15kN and the rotation speed n(t) = 40~60r / min (high drilling pressure and low rotation speed to improve rock breaking efficiency); when the lithology is fractured, adjust the drilling pressure f(t) to decrease, the rotation speed n(t) to decrease, and the mud pump pressure Q(t) to increase sharply (to reduce drill bit vibration).

[0139] Step 503: Based on the lithology judgment results obtained in Step 2, dynamically adjust the drilling rig power through an adaptive control algorithm to reduce energy consumption.

[0140] In step 503, the specific constraints for dynamic adjustment are as follows: when the lithology is a coal seam, the adjustment power is 60% of the rated power; when the lithology is a coal seam, the adjustment power is 80% of the rated power; when the lithology is a fracture, the pump pressure changes abruptly, that is, the pump pressure change ΔP>2MPa, and the mud discharge is increased by 30% to 50%.

Claims

1. A method for accurately determining drilling footage and formation strike in coal mines using multiple parameters, characterized in that... The method includes the following steps: Step 1: Multi-parameter data acquisition and preprocessing: Collect multi-parameter data, and preprocess the collected multi-parameter data samples using wavelet transform to eliminate noise interference; The multi-parameter data includes drilling rig status parameters, borehole trajectory parameters, geophysical parameters, and drilling dynamics parameters; Step 2, Real-time efficiency assessment and lithology identification: Step 201: Using the multi-parameter data preprocessed in Step 1 as input parameters, train the improved ResNet50-multiple LSTM model, optimize the network parameters through the backpropagation algorithm, and use the trained improved ResNet50-multiple LSTM model as a real-time efficiency evaluation and lithology identification model. The improvement method for the ResNet50 network is as follows: CBAM is introduced into Residual Block I and Residual Block II of the ResNet50 network respectively; The real-time efficiency evaluation method is as follows: effective drilling time percentage = effective drilling time / total drilling operation time; The lithology mentioned includes coal seams, sandstone, and fractures; Step 202: After the real-time multi-parameter data is preprocessed in Step 1, it is input into the real-time efficiency evaluation and lithology identification model obtained in Step 201, and the real-time efficiency evaluation and lithology identification results are output. Step 3, Calculation of Effective Advance: Step 301: Based on the lithology identification results obtained in Step 2, determine whether the drilling section displacement is valid according to the displacement determination method; the drilling section displacement includes rod changing displacement, stuck drill displacement, drill retraction displacement, and idle displacement. Step 302: Based on the lithology identification results obtained in Step 2 and the displacement judgment results obtained in Step 301, calculate the effective advance. Step 4, Stratigraphic strike prediction and dynamic correction: Step 401, Calculation of stratigraphic strike: The dip direction of the strata is perpendicular to the strike of the strata. At the borehole trajectory point (x,y,z), the plane equation Ax+By+Cz+D=0 is fitted through the strata interface points of 3 different boreholes. The strike angle φ of the strata strike satisfies: φ=arctan(-A / B). In the formula: A represents the component of the normal vector on the x-axis, reflecting the east-west dip trend of the strata. B represents the component of the normal vector on the y-axis, reflecting the dipping trend of the strata in the north-south direction; C represents the coefficient in the z-axis direction of the plane equation, which directly corresponds to the dip of the stratum interface; D represents the constant term in the plane equation; Step 402, Real-time trajectory fitting and deviation analysis: The apex angle and azimuth angle measured while drilling will be converted into three-dimensional coordinates. , , ), Calculation of the deviation between the trajectory and the design path ; ； ； ； ； In the formula: This indicates the deviation between the trajectory and the designed path; This represents the x-axis coordinate of the borehole trajectory point at time t in the three-dimensional coordinate system during drilling measurements, reflecting the real-time east-west position of the borehole on the horizontal plane. This represents the y-axis coordinate of the borehole trajectory point at time t in the three-dimensional coordinate system during drilling measurements, reflecting the real-time north-south position of the borehole on the horizontal plane. This represents the z-axis coordinate of the borehole trajectory point at time t in the three-dimensional coordinate system during measurement while drilling, reflecting the real-time depth of the borehole. This represents the time variable of measurements while drilling, characterizing the time progress of the drilling operation, and corresponding to the cumulative time from the start of drilling to the current moment; This represents the x-axis coordinate at time t in the design path, i.e., the east-west target position that the borehole should reach at that time node; This represents the y-axis coordinate at time t in the design path, which is the north-south target position that the borehole should reach at that time node. This represents the z-axis coordinate at time t in the design path, which is the target depth that the borehole should reach at that time node; The x-axis coordinate represents the starting point of the borehole, which is the initial east-west position at the start of drilling; The y-axis coordinate represents the starting point of the borehole, which is the initial north-south position at the start of drilling. The z-axis coordinate represents the starting point of the borehole, which is the initial depth at the start of drilling; It represents the apex angle of the borehole at time t, that is, the angle between the borehole axis and the vertical direction, which determines the vertical drilling trend of the borehole; The azimuth of the borehole at time t is the angle between the projection of the borehole axis onto the horizontal plane and the due north direction, which determines the horizontal drilling direction of the borehole. Step 403, Dynamic Correction: When the location changes, dynamic stratigraphic correction is performed. The specific method for dynamic correction is as follows: ; In the formula: Indicates the apparent tilt angle; Indicates the angle of inclination; Indicates the angle between the borehole azimuth and the strike of the formation; This represents the correction factor that takes into account the anisotropy of the rock strata.

2. The method for accurately determining drilling footage and formation strike in coal mines using multiple parameters as described in claim 1, characterized in that... In step one, the drilling rig status parameters include drilling pressure, rotational speed, torque, hydraulic pump pressure, and mud pump pressure; the drilling trajectory parameters include well inclination, azimuth, tool face angle, and hole depth; the geophysical parameters include natural gamma value and resistivity; and the drilling dynamics parameters include vibration signals.

3. The method for accurately determining drilling footage and formation strike in coal mines using multiple parameters as described in claim 1, characterized in that... In step 201, the improved ResNet50-multiple LSTM model extracts effective features using an improved ResNet50 network. The feature map output by the ResNet50 network is then fused with feature data through multiple LSTMs, and finally, a real-time efficiency evaluation and lithology identification model is output through the ReLU activation function.

4. The method for accurately determining drilling footage and formation strike in coal mines using multiple parameters as described in claim 3, characterized in that... In step 201, the effective features include azimuth gamma, resistivity, and drilling dynamics parameters; the drilling dynamics parameters include vibration signal a(t) and displacement difference δ(t).

5. The method for accurately determining drilling footage and formation strike in coal mines using multiple parameters as described in claim 1, characterized in that... In step 301, the displacement determination method is as follows: Step 30101, Rod replacement displacement: When drilling pressure f(t) = 0, rotational speed n(t) = 0, vibration signal a(t) = 0, displacement When the displacement of the pump rod is between 0.75 and 3.0 m, the pump pressure Q(t) is 0, and the torque T(t) is 0, then the displacement of the pump rod is determined to be invalid, i.e., the determination index is... ; Otherwise, the displacement of the rod replacement is determined to be a valid displacement, i.e., the determination index. ; Step 30102, stuck drill displacement: When the drilling pressure f(t) suddenly increases, the rotational speed n(t) = 0, the vibration signal a(t) = 0, the displacement change δ(t) = 0 m, the mud pump pressure Q(t) suddenly increases, and the torque T(t) suddenly increases, then the stuck drill displacement is determined to be an invalid displacement, i.e., the judgment index. ; Otherwise, the stuck drill displacement is determined to be a valid displacement, i.e., the judgment index. ; Step 30103, Drill Retraction Displacement: When drilling pressure f(t) = 0, rotational speed n(t) > 0, vibration signal a(t) = 20Hz, displacement change δ(t) = 0m, mud pump pressure Q(t) suddenly increases and torque T(t) = 2KN, then the drill retraction displacement is determined to be an invalid displacement, i.e., the determination index. ; Otherwise, the retraction displacement is determined to be a valid displacement, i.e., the determination index. ; Step 30104, Idle Displacement: When drilling pressure f(t) = 0, rotational speed n(t) > 0, vibration signal a(t) = 22Hz, displacement change δ(t) = 0m, mud pump pressure Q(t) suddenly increases and torque T(t) = 2.4KN, then the idle displacement is determined to be invalid displacement, i.e., the judgment index. ; Otherwise, the idling displacement is determined to be a valid displacement, i.e., the determination index. .

6. The method for accurately determining drilling footage and formation strike in coal mines using multiple parameters as described in claim 1, characterized in that... In step 302, the effective footage is the sum of the displacements of each effective drilling section, i.e., the effective footage. ; In the formula: Indicates effective advance; Indicates the first Displacement sensor records for each drilling section; Indicates the first Indicators for determining each drilling section; This indicates that the statement is valid. Indicates invalid.

7. The method for accurately determining drilling footage and formation strike in coal mines using multiple parameters as described in claim 1, characterized in that... It also includes step five; Step 5, Multi-source parameter control: Step 501: Based on the real-time efficiency evaluation results obtained in Step 2, dynamically adjust the drilling pressure through an adaptive control algorithm to improve efficiency. Step 502: Based on the lithology judgment results obtained in Step 2, dynamically adjust the drilling pressure and rotation speed through an adaptive control algorithm to improve efficiency; Step 503: Based on the lithology judgment results obtained in Step 2, dynamically adjust the drilling rig power through an adaptive control algorithm to reduce energy consumption.

8. The method for accurately determining drilling footage and formation strike in coal mines using multiple parameters as described in claim 7, characterized in that... In step 501, the specific constraint condition for the dynamic adjustment is: when the effective drilling time percentage is ≤75%, the drilling pressure is gradually increased at a rate of 1kN per step until the efficiency recovers or the maximum drilling pressure limit is reached.

9. The method for accurately determining drilling footage and formation strike in coal mines using multiple parameters as described in claim 7, characterized in that... In step 502, the specific constraints of the dynamic adjustment are as follows: when the lithology is coal seam, adjust the drilling pressure f(t) = 5~10kN and the rotation speed n(t) = 60~100r / min; when the lithology is sandstone, adjust the drilling pressure f(t) = 11~15kN and the rotation speed n(t) = 40~60r / min; when the lithology is fractured, adjust the drilling pressure f(t) to decrease, the rotation speed n(t) to decrease, and the mud pump pressure Q(t) to increase sharply.

10. The method for accurately determining drilling footage and formation strike in coal mines using multiple parameters as described in claim 7, characterized in that... In step 503, the specific constraints of the dynamic adjustment are as follows: when the lithology is coal seam, the adjustment power = 60% of the rated power; when the lithology is sandstone, the adjustment power = 80% of the rated power; when the lithology is fractured, the mud discharge is increased by 30% to 50% according to the sudden change in pump pressure, that is, the change in pump pressure ΔP > 2MPa.

Images