Unfavorable geological exploration method based on inclined drilling

Abstract

The invention relates to the technical field of stratum detection and track control in front of a tunnel face of an expressway tunnel, and provides an unfavorable geological exploration method based on inclined drilling. The method aims at solving the problems that in the prior art, a static geologic model is disjointed with an actual stratum, passive response is made to stratum changes, a stratum subdivision adjustment plan is lacked, and a multi-drill-hole collaborative iteration mechanism does not exist. The method comprises the following steps: firstly, collecting multi-source geological data to construct a dynamic geological model capable of being iteratively updated, generating an initial trajectory and presetting a differentiated stratum adjustment plan; integrating sensing, execution and control components to complete calibration and simulation test; then drilling is conducted according to the initial track, and initial calibration, predictive adjustment and real-time response closed-loop control are sequentially executed; and finally, the effect is verified through imaging and rock core analysis, the model is optimized through drilling data inversion, and multi-drill-hole cooperative detection precision iteration is achieved. Stratum detection accuracy and trajectory control precision are improved, and support is provided for safe and efficient development of tunnel construction.

Description

Technical Field

[0001] This invention relates to the field of geological exploration and trajectory control technology in front of the tunnel face of highways, specifically a method for exploring adverse geological conditions based on inclined boreholes. Background Technology

[0002] In highway tunnel construction, geological detection and trajectory control ahead of the tunnel face are crucial for construction safety and efficiency. However, existing technologies suffer from several shortcomings, making it difficult to meet the demands for accurate detection and efficient control. Specific technical problems include: First, existing technologies mostly employ static geological models with fixed parameters that cannot be updated to reflect actual geological changes during drilling. This leads to a disconnect between the model representation and the actual geological conditions, resulting in insufficient accuracy of the geological data provided for subsequent trajectory design and adjustment. Second, responses to geological changes are often passive, with trajectory adjustments only initiated after changes are detected. This delay in adjustment can easily cause trajectory deviations, failing to proactively mitigate the risk of trajectory errors caused by geological changes. Third, there is a lack of detailed adjustment plans for different geological change scenarios. Using a uniform adjustment logic and parameters to address various situations such as sudden changes in hard and soft rock or the proximity of fracture zones results in poor targeted adjustments and difficulty in ensuring control accuracy. Fourth, there is no systematic multi-hole collaborative optimization mechanism. Data from preceding boreholes cannot provide optimization support for subsequent boreholes, making it difficult to continuously improve detection accuracy.

[0003] Therefore, a method for exploring adverse geological conditions based on inclined boreholes is proposed. This method involves constructing an iteratively updatable dynamic geological model and continuously refining the model parameters using real-time drilling data. A predictive adjustment mechanism is introduced to anticipate changes in the strata ahead based on real-time formation parameters and the predictive model, and to adjust the trajectory in advance. Differentiated adjustment plans are designed for different strata change scenarios. A multi-bore collaborative iterative mechanism of "data inversion - model optimization - subsequent borehole accuracy improvement" is established to overcome the limitations of existing technologies and achieve accurate and efficient strata detection and trajectory control ahead of the tunnel face. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method for exploring adverse geological conditions based on inclined boreholes.

[0005] To achieve the above-mentioned objective, the present invention provides the following technical solution: a method for exploring adverse geological conditions based on inclined boreholes, comprising the following steps: Step 1: Dynamic geological model construction and trajectory design: Collect multi-source geological data to initialize a three-dimensional dynamic geological model and preset model iteration trigger conditions; Based on the dynamic geological model, generate an initial trajectory and positioning curve through a geological-trajectory coupling algorithm, and preset trajectory adjustment plans for different stratigraphic types; Step 2: Component Integration and Debugging: Select and install the core sensing, actuation, and control components, complete the calibration of each component and simulated formation testing, and verify the collaborative working effect of the components and the accuracy of the prediction module; Step 3 Closed-loop control operation: Drill according to the initial trajectory, and sequentially perform initial drilling model calibration, predictive adjustment and real-time response control to correct trajectory deviation; Step 4: Quantification of Results and Iteration of Scheme: Based on the drilling termination conditions, stop the operation, verify the accuracy of target detection through borehole imaging and core analysis, update the dynamic geological model by inverting the drilling data, and perform multi-bore collaborative detection based on the updated model to achieve accuracy iteration.

[0006] Preferably, the multi-source geological data acquisition in step 1 includes: using seismic wave detection equipment and ground-penetrating radar to scan a preset area in front of the tunnel face to obtain three-dimensional data of depth, stratum type, and wave velocity; retrieving regional geological reports to extract mechanical parameters of different stratum types, wherein the stratum types include at least hard rock, soft rock, and fracture zones.

[0007] Preferably, the dynamic geological model construction in step 1 adopts a three-dimensional modeling framework, and the construction logic includes: establishing a three-dimensional coordinate system with the center point of the tunnel face as the origin, filling the unsampled area of ​​the stratigraphic layer data with an interpolation algorithm; marking the stratigraphic type probability of each coordinate point; embedding a stratigraphic trend prediction submodule to calculate the stratigraphic intensity change trend, and performing deviation correction processing on abnormal data points.

[0008] Preferably, the model iteration triggering condition in step 1 is: when the preset drilling depth and the threshold acoustic detection module detects that the attenuation rate of the reflected wave reaches the preset threshold, the real-time drilling data is called and the model probability parameters are corrected by weighted averaging.

[0009] Preferably, the initial trajectory generation in step 1 includes: using the center of the target detection area as the endpoint, employing a path planning algorithm to avoid high-risk geological areas, with the objective function being the shortest trajectory length + the lowest probability of contact with risk areas; calculating the minimum radius of curvature based on the borehole diameter and inclination angle, and the positioning curve being formed by dividing the drilling path into sections at preset intervals and connecting the center points of each section.

[0010] Preferably, the adjustment plan in step 1 includes: Hard rock and soft rock abrupt change adjustment: When the formation hardness difference is detected to reach the preset threshold and the distance between the calibration line and the positioning curve cross section exceeds the allowable deviation, the execution component is pre-adjusted in advance, and the pre-adjustment direction is the side with higher formation hardness. Breaking zone proximity adjustment: When the acoustic detection detects that the distance to the breaking zone has reached the warning value, the allowable deviation threshold between the calibration line and the positioning curve is tightened to improve the response speed of the actuator.

[0011] Preferably, the core components in step 2 include: Formation parameter sensing components: including at least a hardness sensor, an acceleration sensor and a multi-frequency acoustic module, used to collect formation hardness, drilling vibration frequency and forward formation reflected wave data, respectively; Depth measurement component: Utilizes fiber optic sensing technology to achieve real-time monitoring of drilling depth and drill rod bending status; Track adjustment component: Employs a hydraulic actuator arranged circumferentially along the drill bit to achieve inclination and azimuth adjustment; Control component: With a programmable logic controller as the core, it integrates a prediction model. The prediction model outputs the probability of formation type and short-term hardness prediction value based on drilling process data, and then controls the action of the trajectory adjustment component.

[0012] Preferably, the calibration in step 2 includes: Hardness sensor calibration: Calibrate using a standard hardness block under standard conditions to keep measurement error within a preset range; Depth measurement component calibration: Establish the mapping relationship between sensor signal changes and drill pipe bending angle using a three-dimensional coordinate measuring machine; Calibration line generation calibration: Ensures that the distance measurement error of drill bit feature points meets the accuracy requirements.

[0013] Preferably, the closed-loop control in step 3 includes: Initial drilling calibration: Drilling parameters are set according to different formation types. After each preset drilling distance, the deviation between the calibration line and the positioning curve is compared. For deviations exceeding the threshold, the execution component is activated for adjustment. Predictive adjustment: Based on real-time data from sensor components, the system predicts changes in the geological formation ahead using a predictive model, and adjusts the trajectory in advance to control deviations; Real-time response control: When encountering prediction deviation, switch to adaptive adjustment mode to improve the response speed of the execution components and dynamically calculate the adjustment amount based on deviation feedback.

[0014] Preferably, the effect quantification and iteration in step 4 includes: Drilling termination: The drilling will stop when the attenuation rate of the acoustic probe reflected wave reaches the termination threshold and the trajectory endpoint deviation meets the requirements. Results verification: The target stratum boundary error was detected by in-hole imaging equipment, and the detection accuracy was verified by the core recovery rate; Model iteration: The geological model boundary is corrected by inverting the drilling data, and the prediction model parameters are adjusted to reduce the prediction error; the subsequent borehole trajectory is generated based on the updated model to achieve iterative improvement of detection accuracy.

[0015] Compared with the prior art, the present invention provides a method for exploration of adverse geological conditions based on inclined boreholes, which has the following beneficial effects: 1. In this scheme, by constructing an iteratively updatable dynamic geological model and continuously correcting the model parameters in combination with real-time drilling data, the geological characterization is always consistent with the actual stratum conditions, providing accurate geological basis for trajectory design and adjustment, and breaking through the limitation of existing static models that cannot adapt to dynamic changes in strata.

[0016] 2. In this scheme, by introducing a predictive adjustment mechanism, the changes in the strata ahead are predicted based on real-time collected stratum parameters and prediction models, and trajectory adjustment is initiated in advance. This avoids the trajectory deviation problem caused by passive response in existing technologies, and significantly improves the timeliness and accuracy of trajectory control.

[0017] 3. In this solution, differentiated adjustment logic is designed based on different geological change scenarios (such as abrupt changes between hard and soft rock, or proximity of fracture zones), including deviation threshold setting and adjustment of the response speed of execution components, to achieve precise adaptation and adjustment to different geological changes. Compared with the unified adjustment method of existing technologies, this significantly improves the targeting and accuracy of trajectory control.

[0018] 4. In this scheme, a multi-hole collaborative iterative mechanism of "data inversion - model optimization - subsequent accuracy improvement" is constructed. The geological model and prediction parameters are corrected by using the previous drilling data, which provides an optimization basis for subsequent borehole exploration and realizes continuous iterative improvement of exploration accuracy. Existing technologies lack such a systematic accuracy optimization mechanism and cannot achieve the progressive improvement of technical effects among multiple boreholes. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the steps of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see Figure 1 A method for exploring adverse geological conditions based on inclined boreholes includes the following steps: Step 1: Dynamic Geological Model Construction and Trajectory Design 1. Multi-source data acquisition and model initialization Data source: A commercially available seismic CT scanner (model: YCS40A, detection depth: 50m, resolution: 1m) and a ground-penetrating radar (model: RD1500, resolution: 0.5m) were used to scan a 50m area in front of the tunnel face of the highway tunnel to obtain stratigraphic data (data format: a three-dimensional table of "depth-stratum type-wave velocity", sampling interval: 0.5m); at the same time, publicly available geological reports for this area (such as "Geological Survey Report of XX Highway Tunnel", No.: GS2024-038) were retrieved to extract mechanical parameters (including Poisson's ratio and elastic modulus, the values ​​of which are the average values ​​of the same type of strata in the report) for hard rock (compressive strength 45-60MPa), soft rock (15-30MPa), and fracture zone (≤10MPa).

[0022] Model Construction: A 3D dynamic geological model was constructed using the Python + Gempy open-source framework (version 3.0). The specific implementation logic is as follows: ① A 3D coordinate system was established with the center point of the tunnel face as the origin (X-axis parallel to the tunnel strike, Y-axis perpendicular to the tunnel strike, Z-axis perpendicular to the ground); ② Stratigraphic data was imported into the framework in the format of "coordinate-stratigraphic type-mechanical parameters", and the "kriging interpolation algorithm" was used to fill the unsampled areas; ③ The probability of stratigraphic type at each coordinate point was labeled (e.g., 80% probability of hard rock and 20% probability of soft rock at a borehole depth of 15m, the probability calculation is based on the stratigraphic type voting of the three sampling points around this point); The "stratigraphic trend prediction submodule" was embedded—based on the mechanical parameters of three adjacent borehole data points, a linear fitting algorithm was used (formula: Calculate the trend of formation strength variation. Explanation of characters in the formula: σ represents the formation compressive strength (MPa) at borehole depth L. The formation compressive strength (MPa) at the initial borehole location (L=0) is the average of three evenly distributed sampling points at the initial location, and L is the borehole depth (m, measured with the borehole opening point as the reference). If there are outliers (deviations exceeding 20% ​​of the mean) among three adjacent data points, the median replacement method is used.

[0023] Iteration trigger conditions: Preset model update rules: Every 3m of drilling (based on the depth measured by the fiber optic grating sensor), or when the acoustic detection module detects a sudden change in the attenuation rate of the reflected wave ≥30% (indicating a new geological body; attenuation rate calculation method: ...). When drilling is in progress, real-time drilling data (hardness, vibration frequency, with data sampling frequency consistent with the sensor) is automatically retrieved. The model probability parameters are corrected using a weighted average method (hardness weight 0.6, vibration frequency weight 0.4), avoiding the problem of "static model decoupling from actual formation" in existing technologies. Note: γ represents the reflected wave attenuation rate (%). The amplitude of the reflected wave before the abrupt change. This represents the amplitude of the reflected wave after the abrupt change.

[0024] 2. Intelligent trajectory generation and adjustment plan Trajectory Generation Algorithm: Based on a dynamic model, an initial trajectory is generated through a "geology-trajectory coupling algorithm." The core steps are: ① The endpoint is the center of the target fracture zone (X=120m, Y=35m, Z=+8m, coordinates based on the model coordinate system); ② Avoid areas in the model where the probability of fracture zone is ≥60%, and the avoidance path planning adopts the "A* algorithm," with "shortest trajectory length + lowest probability of fracture zone" as the objective function; ③ Calculate the inclined trajectory with a minimum radius of curvature ≥15m (radius of curvature calculation is based on a borehole diameter of 100mm, formula: The characters in the formula are explained as follows: R is the radius of curvature (m), d is the borehole diameter (m), and θ is the inclination angle (°); the initial trajectory parameters are: opening point (X=100m, Y=30m, Z=+5m), inclination angle 30°, azimuth angle 15°, and hole depth 25m.

[0025] Meanwhile, based on the initial design trajectory, a supplementary positioning curve generation logic is added: the drilling path is divided into sections at 0.5m intervals along the drilling direction (consistent with the formation sampling interval), the center point coordinates of each section are extracted (determined by the projection position of the design trajectory on that section), and all the center points of the sections are connected to form a continuous positioning curve (covering the hole depth range of 0-25m, such as the hole depth of 0m corresponding to the opening point coordinates, and the hole depth of 25m corresponding to the target fracture zone center coordinates), which serves as the reference path for the drill bit's movement.

[0026] Pre-set adjustment plan (revised deviation judgment logic): Differentiated adjustment logic is set for different stratigraphic types (no such detailed plan exists in existing technology, demonstrating creativity): Calibration line definition: The three-dimensional coordinates of the center point of the drill bit tip (front end in the drilling direction) and the center point of the drill bit end (drill rod connection end) are collected in real time by fiber optic grating sensors, and the two points are connected to form a calibration line; the calibration line is updated in real time as the drill bit advances, and the coordinate data is refreshed every 0.1 seconds.

[0027] 1. Hard rock to soft rock abrupt change (hardness difference ≥ 15 MPa, hardness difference is the difference between the average hardness value in the previous 0.5 seconds and the current hardness value): Real-time comparison of the overlap between the calibration line and the positioning curve. If the distance between the calibration line and the positioning curve at any cross-section is ≥ 0.3 m (preset error threshold), the push rod pre-adjustment is initiated 0.5 seconds in advance. Pre-adjustment amount = hardness difference × 0.02 (formula: For example, a 20MPa difference corresponds to a 0.4mm pre-adjustment amount, with the pre-adjustment direction being "shifted towards the side with higher formation hardness". Explanation of characters in the formula: Preset amount (mm). The difference in hardness is (MPa).

[0028] 2. Approaching a fracture zone (sonic detection distance ≤ 5m, detection distance is calculated based on "reflection wave propagation time × sound wave propagation speed in the formation", sound wave speed is taken as the default value of the current formation type): tighten the allowable deviation threshold between the calibration line and the positioning curve from 0.5m to 0.2m (deviation is the distance between the calibration line and the positioning curve at the corresponding cross section), and increase the push rod response speed from 500ms to 300ms (response speed is "time from receiving the command to the push rod action").

[0029] Step 2: Feasible component integration and debugging 1. Selection and installation of core components Miniature hardness sensor (HS-01): measuring range 10-200MPa, response time 0.1s, accuracy ±5%; three sensors are evenly arranged around the front end of the drill bit at 120° intervals; fixed with M3 threads, tightening torque 2.5N·m, sensor probe flush with the front end of the drill bit (error ≤0.1mm); sensor wire (model RVVP2×0.5mm²) runs along the inner wall of the drill rod, and is fixed every 50cm with high-temperature resistant cable ties (temperature resistance ≥150℃), the end of the wire is connected to the analog input interface (channels AI1-AI3) of the ground control box.

[0030] Accelerometer (ADXL355): Range ±10g, accuracy 0.1mg, sampling rate 1kHz; installed at the center of the drill bit tail; before installation, wipe the bonding surface with alcohol (to remove oil); use epoxy resin (model E-44) for bonding and fixing; curing conditions are 25℃ for 24 hours; the sensor data interface (SPI protocol) is shared with the hardness sensor, and data is transmitted through time-division multiplexing (hardness data duty cycle 60%, vibration data duty cycle 40%).

[0031] Multi-frequency acoustic module (SW-05): operating frequency 10-50kHz (can be set via control box, 50kHz for hard rock, 30kHz for soft rock, and 10kHz for fractured zone), detection radius 5-10m, reflected wave resolution 0.1m; installed at the rear of the drill bit 5cm from the cutting edge, with the acoustic transmitter facing the drilling direction and three receivers arranged circumferentially (spaced 120° apart); both transmitter and receiver are waterproof and sealed (sealing ring model 10×1.5mm O-ring, material nitrile rubber) to prevent drilling fluid from entering.

[0032] Miniature hydraulic push rod (HY-03): Stroke 0-5mm, thrust 5kN, response time ≤100ms; 3 sets of push rods are evenly distributed at 120° at the connection between the drill bit and the drill pipe. The push rod base is fixed to the end face of the drill pipe with M5 bolts (tightening torque 5N·m). The top of the push rod is connected to the tail of the drill bit through a ball joint (model SA30); the hydraulic pipeline of the push rod (model PU pipe Φ4×1mm) is arranged along the inner wall of the drill pipe and connected to the ground hydraulic station (pressure 0.8MPa).

[0033] Fiber Bragg Grating (FBG) Sensor + Demodulator: The FBG-1550 fiber optic sensor operates at a wavelength of 1520-1570nm with an accuracy of ±0.1nm. Three FBG sensors are embedded axially along the inner wall of each drill rod section (3m in length), spaced 0.5m apart. One sensor is used to acquire the coordinates of the center point of the drill bit tip (near the cutting edge of the drill bit), and another sensor is used to acquire the coordinates of the center point of the drill bit end (near the drill rod connection end). The sensor is fixed in a slot on the inner wall of the drill rod (slot depth 0.5mm) with epoxy resin (model E-51). The two ends of the fiber optic cable are connected to the demodulator (FBG-200) via SC connectors (model SC-UPC). The demodulator has a sampling frequency of 1Hz, and the data is transmitted to the control box via an RS485 interface.

[0034] Control and Prediction Module: Centered on an industrial control box (PLC-S7-1200, CPU model 1214C), it integrates an LSTM model developed based on TensorFlow 2.0. The model input features are a 4-dimensional vector of "hardness (MPa), vibration frequency (Hz), depth (m), and time (s)," with two hidden layers (64 neurons per layer). The output is "stratum type probability and hardness prediction value after 10cm." The training data comes from publicly available tunnel drilling datasets (such as 300 sets of tunnel drilling data in the "China Tunnel Engineering Geological Database"), with 1000 training iterations and mean squared error (MSE) as the loss function. After training, the model weight file (.h5 format) is imported into the PLC's memory card. The control box is installed on the ground operating platform and receives data from various sensors via Ethernet (TCP / IP protocol, baud rate 115200bps), outputting 4-20mA analog signals (corresponding to push rod stroke 0-5mm) to control the hydraulic push rod action.

[0035] 2. Equipment linkage commissioning Calibration steps: (1) Hardness sensor calibration: In an environment of 20-25℃, use standard hardness blocks (20MPa, 50MPa, 100MPa, accuracy ±1MPa) to contact the sensor probe one by one, with a contact pressure of 50N and a contact time of 3s; adjust the amplification factor of the sensor (range 0.8-1.2) through the parameter adjustment function of the ground control box (calibration program in PLC, path "menu-calibration-hardness sensor") so that the error between the sensor display value and the nominal value of the standard hardness block is ≤5%; save the parameters to the permanent storage area of ​​PLC after calibration.

[0036] (2) Fiber Optic Measurement Calibration: Fix the drill rod on a three-dimensional coordinate measuring machine (model LeitzPMM-C, accuracy ±0.001mm), and manually control the drill rod to bend to a known angle (1°, 2°, 3°, the angle is read by the measuring machine); record the wavelength change value of the fiber optic grating sensor at different bending angles, and use the "least square method" to fit and generate a "wavelength change - bending angle" calibration curve (formula: ), to ensure curve fit Note: In the formula, α represents the bending angle (°). The wavelength change is in nm; after calibration, the curve is loaded using the demodulator's "Curve Import" function for subsequent coordinate conversion.

[0037] (3) Calibration line generation calibration: After fiber measurement calibration, fix the standard distance between the drill tip and the end of the drill bit (preset to 0.5m) using a three-dimensional coordinate measuring instrument, input it into the control box, and ensure that the distance calculated from the coordinates of the two points collected by the fiber sensor is ≤0.1mm away from the standard distance error, so as to avoid the calibration line length deviation affecting the judgment.

[0038] Simulated strata test: In the laboratory, three simulated strata were constructed: "hard rock (50MPa concrete, cement:sand:stone = 1:2:3), soft rock (20MPa gypsum, gypsum:water = 10:3), and fractured zone (8MPa crushed stone mixed mortar, crushed stone:sand:cement:water = 5:3:1:1)" (each layer is 1m thick and 2m×2m×2m in size); the drilling speed was set to 5cm / min (controlled by the feed motor of the drilling rig, speed 10r / min) to verify the collaborative working effect of each module.

[0039] Prediction module verification: Input the vibration frequency (150Hz, sampling time 10s) of the hard rock section into the control box. The LSTM model should be able to accurately predict that after 10cm, it will enter the soft rock section (corresponding to vibration frequency 80Hz). The prediction accuracy is ≥90% (accuracy is calculated as "number of correct predictions / total number of tests × 100%", and the number of tests is ≥30).

[0040] Component adjustment verification: When the hardness sensor detects that the formation hardness has decreased from 50MPa to 20MPa (hardness change rate 5MPa / s), first compare the deviation between the current calibration line and the positioning curve. If the deviation is ≥0.2m, the control box automatically commands the hydraulic push rod to extend by 0.4mm (through the PLC output of a 20mA analog signal), so that the drill bit inclination angle is adjusted from 30° to 30.8° (the inclination angle is measured by the fiber optic grating sensor). The angle adjustment error is ≤0.2° (the error is the absolute value of "measured inclination angle - target inclination angle"), and the deviation between the calibration line and the positioning curve after adjustment is ≤0.1m.

[0041] Step 3: Specific Operations of Closed-Loop Control 1. Initial drilling and model calibration (hole depth 0-5m, laying the foundation for accuracy) Start the drilling rig (model: XY-44) and set the drilling parameters according to the initial formation type: 8kN thrust in hard rock section (controlled by hydraulic station pressure, 1.2MPa) and 300r / min rotation speed (controlled by the drilling rig spindle motor); reduce the thrust to 6kN (0.9MPa) and 250r / min in soft rock section and start drilling according to the designed trajectory.

[0042] For every 1 meter drilled (based on the depth measured by the fiber Bragg grating sensor), the center point coordinates of the current drill bit tip and tip are obtained through the fiber optic demodulator, and a real-time calibration line is generated. The real-time calibration line is then compared with the positioning curve section by section, and the maximum deviation between the two is calculated. If the maximum deviation is ≤0.2m: maintain the current drilling parameters, and at the same time input the measured hardness data at this depth (e.g., 48MPa, take the average of 3 sensors) into the dynamic geological model, and correct the probability of hard rock in the range of 5-10m hole depth using the "Bayesian update method" (updated from the initial 80% to 85%).

[0043] If the maximum deviation is >0.2m (e.g., the measured deviation is 0.3m): immediately start the single-sided hydraulic push rod to shorten by 1mm (through the PLC output of a 4mA analog signal), and adjust the drilling azimuth angle by +1° (the azimuth angle adjustment direction is determined by "the side of the calibration line deviating from the positioning curve". If the calibration line deviates towards the +X direction, then adjust towards the -X direction). The entire adjustment process is completed within 500ms (the time from detecting the deviation to the end of the adjustment). At the same time, record the adjustment parameters (push rod extension and retraction, angle change value) to the model's "historical database" to provide a reference for subsequent handling of similar deviations (when encountering the same deviation again, the historical parameters will be called first).

[0044] 2. Predictive adjustment (hole depth 5-20m, overcoming the limitations of passive response in existing technologies) Data acquisition: The hardness sensor collects formation hardness data every 0.1s (e.g., 48MPa at t=60s, dropping to 46MPa at t=60.1s, and the data is the average of the three sensors). The accelerometer synchronously collects vibration frequency (145Hz at t=60s, dropping to 130Hz at t=60.1s, and the data is the result of filtering the original sampled values ​​using the Kalman filter algorithm). All data is transmitted to the ground control box in real time via Ethernet (transmission delay ≤100ms).

[0045] Prediction calculation: The LSTM model is based on the feature combination of "hardness decreases by 2MPa within 200ms + vibration frequency decreases by 15Hz within 200ms" (feature window is 200ms, sliding step size is 0.1s), and outputs the prediction result: "The next 30cm drilling section will enter soft rock formation, with a predicted hardness of 22MPa, and the deviation between the calibration line and the positioning curve will reach 0.3m in the next 30cm". The confidence level of the prediction result must be ≥85% (the confidence level is determined by the probability value output by the model). If the confidence level is <85%, the sampling time is increased to 500ms and the prediction is re-predicted.

[0046] Early intervention: If the prediction deviation is ≥0.2m (allowable threshold), the control box, based on the prediction result, instructs the left hydraulic push rod to extend by 1.5mm (outputting a 16mA analog signal) and the right hydraulic push rod to shorten by 0.5mm (outputting an 8mA analog signal), completing the adjustment of the drill bit inclination angle from 30° to 30.5° within 500ms; after drilling 30cm, the hardness sensor measures the formation hardness to be 23MPa, with a deviation from the predicted value ≤4% (deviation calculation method: Furthermore, the actual deviation between the calibration line and the positioning curve is ≤0.1m, effectively avoiding trajectory deviation caused by lack of prior adjustment in existing technologies (where the deviation is typically above 0.2m). Explanation of characters in the formula: δ represents the hardness deviation rate (%). The measured hardness (MPa) is given. To predict hardness (MPa).

[0047] 3. Real-time response (with guaranteed accuracy in the vicinity of the target area for borehole depths of 20–25 m) If a prediction deviation occurs (e.g., at a hole depth of 22m, the acoustic detection module detects the breakage zone signal 0.2s in advance, and the hardness sensor's measured hardness drops sharply to 9MPa, deviating from the predicted "breakage zone encountered at 25m", with a deviation distance of 3m): The control box immediately switched from "predictive mode" to "PID adaptive control mode," with PID parameters set according to the "Ziegler-Nichols tuning method": proportional coefficient... (Initial value, which can be dynamically adjusted according to the magnitude of the deviation; when the deviation is >0.3m) When the deviation is <0.1m ), integral coefficient (Integration time 10s), differential coefficients (Differential time 2s); at the same time, the response speed of the hydraulic push rod is increased from 500ms to 300ms (achieved by shortening the opening of the throttle valve in the hydraulic pipeline) to quickly respond to sudden changes in the formation.

[0048] The drill bit inclination angle is adjusted from 30.5° to 32° within 500ms (the adjustment range is calculated by the PID algorithm, formula: ),in The actual deviation between the calibration line and the positioning curve; the real-time trajectory deviation (deviation between the calibration line and the positioning curve) fed back by the fiber optic demodulator decreased from 0.4m to 0.15m, meeting the accuracy requirements for target area exploration (deviation ≤ 0.3m). Explanation of characters in the formula: The extension / retraction amount of the push rod (mm). This is the proportionality coefficient. The integral coefficient is... These are the differential coefficients. Let t be the trajectory deviation (m) and t be the time (s).

[0049] Step 4: Quantifiable Results and Iterable Solutions 1. Drilling Termination and Hit Verification Termination condition: The multi-frequency acoustic module detects a reflected wave attenuation rate ≥ 60% (indicating that the core area of ​​the fracture zone has been entered; the attenuation rate is calculated in the same way as in step one, using the formula: Furthermore, if the calibration line generated by the coordinates of the drill bit tip center point acquired by the fiber optic sensor deviates from the endpoint of the positioning curve (hole depth 24-25m) by ≤0.3m (preset threshold), drilling operations are stopped (achieved by controlling the power off of the drill rig spindle motor). Explanation of characters in the formula: γ is the reflected wave attenuation rate (%). The amplitude of the reflected wave before the abrupt change. This represents the amplitude of the reflected wave after the abrupt change.

[0050] Verification steps: (1) Borehole imaging verification: The entire borehole wall was scanned and photographed using a borehole imager (model: ZJ-200, resolution 0.1mm) at a scanning speed of 5cm / s. The image storage format was JPEG. The image showed that the length of the fracture zone in the borehole was 6m, which was ≤3% compared with the 5.8m predicted by the dynamic geological model (error calculation method: Simultaneously, it was confirmed that the trajectory of the calibration line was consistent with the actual path of the borehole wall, with no significant deviation (deviation ≤ 0.15m), proving the accuracy of the target area detection. Explanation of characters in the formula: The error is the length of the fracture zone (%). The measured length of the fracture zone (m) is given. To predict the length (m) of the fracture zone.

[0051] (2) Core analysis and verification: Core samples were retrieved from the borehole using a "double-tube single-action drill" at a sampling speed of 3 cm / min. The core recovery rate in the fractured zone area was ≥80% (the recovery rate was calculated as follows: ,in (This refers to the length of the fracture zone distribution within the borehole). Explanation of characters in the formula: Core recovery rate (%) The total length of the core sample taken (m). The drilling depth (m) of the fractured zone section was observed; the particle size distribution characteristics of the rock core were observed (measured by sieve analysis, sieve size 0.1mm, 1mm, 5mm), which completely corresponded to the "disordered reflected wave area" (area where the reflected wave amplitude fluctuation is >20%) recorded by the acoustic detection, further confirming that there were no missed detections.

[0052] 2. Data inversion and multi-hole collaboration Model Update: The entire drilling data (hardness-depth curve at a depth of 25m, trajectory deviation time series, and acoustic reflectivity data in CSV format) was input into the dynamic geological model. The three-dimensional boundary parameters of the target fracture zone were corrected using the finite element inversion method—the Z-coordinate of the fracture zone in the original model was +8m, and after the update it was +7.8m, with the lateral distribution range expanding from 5m to 5.2m (the boundary parameter correction was based on the principle of minimizing the residual between measured data and model prediction data). At the same time, the cause of deviation in each depth segment was analyzed using the deviation source tracing algorithm. The algorithm logic is as follows: ① Calculate the residual between the actual deviation of the "calibration line-positioning curve" and the model prediction deviation in each depth segment; ② Analyze the correlation between the residual and formation parameters (hardness, vibration frequency) (a correlation coefficient absolute value > 0.7 is considered a strong correlation); ③ Adjust the mapping coefficient for strongly correlated parameters. For example, a trajectory deviation of 0.2m at a borehole depth of 18m originates from a deviation in the mapping coefficient between vibration frequency and hardness (the original coefficient was 0.8, which was adjusted to 0.75 after optimization). The optimized coefficient is then input into the LSTM model to re-predict the stratigraphic changes in the area. The error between the predicted and measured values ​​is ≤5% (error calculation method is the same as in step three, formula: Note: In the formula, δ represents the parameter error rate (%). These are measured values. These are predicted values.

[0053] Verification of the second borehole: Based on the updated dynamic model, the trajectory parameters and positioning curve of the second exploration borehole (target: the extended section of the fractured zone) were generated: opening point X=102m, Y=32m, inclination angle 28°, borehole depth 28m (trajectory generation algorithm is the same as in step one); drilling was carried out using optimized adjustment parameters (hardness mutation threshold of the fractured zone area 12MPa, hydraulic push rod response speed 300ms); the final deviation between the borehole calibration line and the positioning curve was 0.1m, an improvement of 33% compared to 0.15m of the first borehole (improvement rate calculation method: This verifies the effectiveness of the iterative mechanism of "front hole data - model optimization - rear hole accuracy improvement" in this scheme. Explanation of characters in the formula: η represents the accuracy improvement rate (%). The deviation of the first hole is (m). The deviation of the second hole is (m).

[0054] Finally, it should be noted that the above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Any obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A method for prospecting unfavorable geology based on inclined boreholes, characterized in that: The method comprises the following steps: Step 1: dynamic geological model construction and trajectory design: collect multi-source geological data to initialize a three-dimensional dynamic geological model, and preset model iteration trigger conditions; generate an initial trajectory and positioning curve based on the dynamic geological model through a geological-trajectory coupling algorithm, and preset trajectory adjustment plans for different stratum types; Step 2: component integration and debugging: select core sensing, execution and control components for installation, complete component calibration and simulated stratum testing, and verify the collaborative working effect of the components and the accuracy of the prediction module; Step 3: closed-loop control operation: drill according to the initial trajectory, sequentially perform initial drilling model calibration, predictive adjustment and real-time response control, and correct trajectory deviation; Step 4: effect quantification and scheme iteration: stop the operation based on the drilling termination condition, verify the target detection accuracy through borehole imaging and core analysis, update the dynamic geological model through drilling data inversion, and perform multi-borehole collaborative detection based on the updated model to realize precision iteration.

2. The method according to claim 1, wherein: The multi-source geological data collection in step 1 comprises: using seismic wave detection equipment and a geological radar to scan a preset range in front of a tunnel face, and obtaining depth-stratum type-wave velocity three-dimensional data; and calling a regional geological report to extract mechanical parameters of different stratum types, wherein the stratum types at least include hard rock, soft rock and a fracture zone.

3. The method according to claim 1, wherein: The dynamic geological model construction in step 1 adopts a three-dimensional modeling framework, and the construction logic comprises: establishing a three-dimensional coordinate system with a center point of the tunnel face as the origin, filling unsampled areas of stratum layered data by using an interpolation algorithm; labeling stratum type probabilities of each coordinate point; embedding a stratum trend prediction sub-module, calculating a stratum strength change trend, and performing deviation correction processing on abnormal data points.

4. The method according to claim 1, wherein: The model iteration trigger condition in step 1 is that when either of the following conditions is met: a preset depth is drilled, and a threshold acoustic wave detection module detects that a reflection wave attenuation rate reaches a preset threshold, the real-time drilling data is called, and the model probability parameters are corrected by using a weighted average method.

5. The method for exploration of unfavorable geology based on inclined borehole according to claim 1, characterized in that: The initial trajectory generation in step 1 comprises: taking a center of a target detection area as an end point, avoiding high-risk stratum areas by using a path planning algorithm, and taking the shortest trajectory length + the lowest risk area contact probability as a target function; calculating a minimum curvature radius according to a borehole diameter and an inclination angle, and the positioning curve is obtained by dividing a drilling path into cross sections at a preset interval, and connecting centers of the cross sections.

6. The method for exploration of unfavorable geology based on inclined borehole according to claim 1, characterized in that: The adjustment plan in step 1 comprises: hard rock and soft rock mutation adjustment: when it is detected that a stratum hardness difference reaches a preset threshold, and a calibration line and a positioning curve cross section distance exceeds an allowed deviation, a pre-adjustment of an execution component is started in advance, and a pre-adjustment direction is a stratum hardness higher side; fracture zone proximity adjustment: when an acoustic wave detects that a fracture zone distance reaches a warning value, the allowed deviation threshold of the calibration line and the positioning curve is tightened, and the response speed of the execution component is improved.

7. The method for exploration of unfavorable geology based on inclined borehole according to claim 1, characterized in that: The core components in step 2 comprise: a stratum parameter sensing component: at least including a hardness sensor, an acceleration sensor and a multi-frequency acoustic wave module, which are respectively used for collecting stratum hardness, drilling vibration frequency and front stratum reflection wave data; a depth measurement component: using an optical fiber sensing technology to realize real-time monitoring of drilling depth and drill pipe bending state; Trajectory adjustment component: hydraulic actuator is used to achieve inclination and azimuth adjustment; Control component: programmable logic controller is used as the core, and a prediction model is integrated, which outputs the formation type probability and short-term hardness prediction value based on drilling process data, and then controls the action of the trajectory adjustment component.

8. The method for exploration of unfavorable geology based on inclined borehole according to claim 1, characterized in that: The calibration in step 2 includes: Hardness sensor calibration: standard hardness blocks are used to calibrate in a standard environment, so that the measurement error is controlled within a preset range; Depth measurement component calibration: a three-dimensional coordinate measuring instrument is used to establish a mapping relationship between the change of the sensing signal and the bending angle of the drill pipe; Calibration line generation calibration: ensures that the distance measurement error of the drill bit feature point meets the accuracy requirement.

9. The method for exploration of unfavorable geology based on inclined borehole according to claim 1, characterized in that: The closed-loop control in step 3 includes: Initial drilling calibration: drilling parameters are set according to different formation types, and the deviation between the calibration line and the positioning curve is compared every preset distance, and the deviation exceeding the threshold value starts the adjustment of the execution component; Predictive adjustment: based on the real-time data of the sensing component, the formation change in front is predicted through the prediction model, and the trajectory is adjusted in advance to control the deviation; Real-time response control: when the predicted deviation is encountered, the adaptive adjustment mode is switched to, the response speed of the execution component is improved, and the adjustment amount is dynamically calculated according to the deviation feedback.

10. The method for exploration of unfavorable geology based on inclined borehole according to claim 1, characterized in that: The effect quantification and iteration in step 4 include: Drilling termination: when the reflection wave attenuation rate of the acoustic wave detection reaches the termination threshold value and the trajectory end point deviation meets the requirements, the machine is stopped; Effect verification: the boundary error of the target formation is detected by the borehole imaging device, and the detection accuracy is verified by the core recovery rate; Model iteration: the drilling data is inverted to correct the geological model boundary, the prediction model parameters are adjusted to reduce the prediction error, and the subsequent drilling trajectory is generated based on the updated model to realize the iterative improvement of the detection accuracy.

Images