Intelligent directional drilling machine adaptive learning measurement and control method and system for coal mine underground

Abstract

The application discloses a kind of intelligent directional drilling rig adaptive learning measurement and control method and system for underground coal mine.The actual build-up rate is calculated by measuring interval by interval acquisition while drilling measurement data and trajectory deviation data;Based on strata build-up gain coefficient and the distance between measurement unit and drill bit, near-bit attitude feedforward compensation is carried out to correct trajectory deviation and eliminate measurement lag;Combined with deviation gain coefficient, comprehensive compensation angle and single comprehensive correction coefficient, the target tool face angle is calculated and the drilling rig is executed;After drilling is completed, the integrated control error of the fusion trajectory deviation and tool face error is calculated, and according to the error dominant representation and build-up rate abnormality, the attribution mapping is carried out, the parameter set to be corrected is selected, and the gradient descent algorithm is used to update the parameters in reverse and feedback the next measurement interval to form adaptive regulation.It solves the control delay and over-correction problem caused by measurement lag, avoids parameter oscillation caused by blind correction of all parameters, and improves the trajectory control accuracy and working condition adaptability in complex strata.

Description

Technical Field

[0001] This invention relates to the field of coal mine geological exploration and drilling engineering technology, and in particular to an adaptive learning measurement and control method and system for intelligent directional drilling rigs in coal mines. It can be applied to the measurement and control while drilling and automatic trajectory correction of various types of directional drilling in coal mines, such as gas drainage holes, water exploration holes, and holes for investigating hidden geological factors that may cause disasters. Background Technology

[0002] Intelligent directional drilling technology is a core supporting means for underground gas control, water hazard prevention, and exploration of hidden geological factors in coal mines. With the deepening of intelligent construction in coal mines, underground directional drilling is developing towards long distances, high precision, and adaptation to complex strata, which puts forward higher requirements for the fully autonomous trajectory control accuracy and dynamic adaptive capability of directional drilling rig systems.

[0003] Currently, the tool face angle control of intelligent directional drilling rigs in coal mines generally adopts a method based on vertical deviation. , left and right deviation The calculation is performed using a simple proportional algorithm, which belongs to the open-loop control mode with fixed parameters and is the closest existing technology in this field. In the measurement while drilling (MWD) stage, the measurement probe is constrained by the build-up tool and interference from ferromagnetic materials (specifically referring to measurement probes using magnetic sensors), and is usually positioned 3.5 to 7 meters behind the drill bit. The control system directly uses the hysteresis attitude data collected by the measurement probe as the current actual attitude of the drill bit, and uses this to calculate the tool face control commands and issue them for execution.

[0004] Existing technologies have the following significant drawbacks in engineering practice: First, measurement lag leads to control inaccuracies. There is a physical distance of several meters between the measurement unit and the drill bit. The collected data reflects the trajectory information of the historical position, which cannot match the actual posture of the drill bit at the current cutting position. Directly correcting based on the lag data will produce phase deviation, which can easily lead to overcorrection or response lag, resulting in increased trajectory fluctuations.

[0005] Second, the correction system for interference factors is fragmented. Existing tool face solution models do not integrate multiple physical interferences such as screw bend angle, drill pipe friction resistance, formation anisotropy, and drill bit size. They are mostly scattered, single-item empirical corrections, lacking a unified mathematical correction model. As a result, the accuracy of the build-up rate prediction is low, and the trajectory deviation often exceeds 1° in complex formations.

[0006] Third, there is a lack of online self-learning capabilities. Most of the parameters in the existing control algorithms are preset fixed values, which cannot be optimized in reverse according to the actual drilling results. When faced with changes in working conditions such as changes in formation lithology, drill bit wear, and increased drill string length, they cannot be dynamically adapted, resulting in a high frequency of corrections (usually more than 3 times / 10 meters), low drilling efficiency, and an increased risk of stuck drill.

[0007] Fourth, poor equipment compatibility. The structural parameters of magnetic probes and gyroscope probes from different manufacturers vary greatly, resulting in different measurement lag distances. Existing control systems are difficult to adapt uniformly, leading to long integration and debugging cycles and high costs.

[0008] Furthermore, although some trajectory prediction and correction technologies exist in the field of oil drilling, oil drilling is a long-bore, slow-formation-change operation, which is fundamentally different from the short-distance, frequent formation lithology changes, limited construction space, and diverse equipment types in coal mines. Therefore, the relevant technologies cannot be directly transferred to the directional drilling scenario in coal mines. Summary of the Invention

[0009] In view of this, the purpose of this invention is to provide an adaptive learning measurement and control method and system for intelligent directional drilling rigs in coal mines. It constructs an adaptive control mechanism of "feedforward prediction - dimensionality reduction calculation - attribution mapping - closed-loop iteration" to solve the technical problems of control delay caused by measurement lag, scattered correction of multiple disturbances, lack of self-learning ability, and poor equipment compatibility, so as to achieve high-precision directional drilling control with low correction frequency in complex strata.

[0010] To achieve the above objectives, the technical solution of the present invention is as follows: In a first aspect, the present invention provides an adaptive learning measurement and control method for intelligent directional drilling rigs in coal mines, comprising the following steps: S1. Obtain the drilling measurement data and trajectory deviation data for each measurement interval, and calculate the actual build-up rate at the current measurement point based on the drilling measurement data and footage of the continuous measurement intervals. S2. Based on the actual build-up rate, the distance between the measurement unit and the drill bit, and the formation build-up gain coefficient, perform feedforward compensation calculation and output the predicted value of the drill bit's true attitude to eliminate measurement lag. S3. Based on the predicted value of the drill bit's true attitude, extrapolate the actual spatial coordinates of the drill bit, correct the vertical trajectory deviation and the horizontal trajectory deviation, and obtain the corrected trajectory deviation to eliminate measurement lag; combine the deviation gain coefficient, comprehensive compensation angle, and single comprehensive correction coefficient corresponding to the corrected trajectory deviation to calculate the target tool facing angle, and generate control commands to be sent to the drilling rig for execution. S4. After the drilling rig completes the drilling of the current measurement interval, it obtains the current trajectory deviation and the actual tool face angle, and calculates the comprehensive control error that combines the trajectory deviation and the tool face error. S5. Based on the dominant characteristics of trajectory deviation and tool surface error in the comprehensive control error, as well as the abnormal slope rate, the error attribution is mapped to the corresponding set of parameters to be corrected. S6. Using the gradient descent algorithm, with the goal of minimizing the overall control error, the correction amount of the parameter to be corrected is calculated in reverse and the parameter is updated. The updated parameter is then fed back to the next measurement interval to form a closed-loop control.

[0011] Secondly, the present invention provides an adaptive learning measurement and control system for intelligent directional drilling rigs in coal mines. The system is deployed in an explosion-proof computing terminal in a coal mine and is electrically connected to the measurement-while-drilling probe and the intelligent drilling rig actuator. The system includes: a data and inclination rate acquisition module, an attitude feedforward compensation module, a tool face calculation module, an error assessment module, an error attribution module, and a self-learning control module. Each module works together to implement the above-mentioned measurement and control method.

[0012] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the steps of the above-described measurement and control method.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. Eliminating measurement lag significantly improves trajectory accuracy. By using a near-bit trajectory feedforward compensation model, the actual bit attitude is extrapolated based on the build-up rate and formation gain coefficient. After correcting the trajectory deviation, control calculation is performed, which solves the control delay problem caused by 6-meter-level measurement lag. Compared with the uncompensated algorithm, the trajectory accuracy is improved by more than 50%, and the trajectory deviation in complex formations is controlled within ±0.5°.

[0014] 2. Single-coefficient unified correction, simple model with strong adaptability. A single comprehensive correction coefficient is used to integrate all drilling interference factors, replacing the scattered sub-correction system. The model is easy to deploy, the fluctuation range of directional efficiency is reduced by 30%, and it is suitable for various working conditions such as soft coal seams, hard rock strata, and fractured zones.

[0015] 3. Attribution-based self-learning ensures convergence and stability with minimal corrections. Based on error-dominated characterization and targeted correction of parameters in cases of abnormal pitch rates, it avoids oscillations caused by blind iteration of all parameters. Combined with gradient descent algorithm, it achieves online optimization at measurement intervals, reducing correction times by over 60% and improving drilling efficiency by over 20%.

[0016] 4. Multi-device compatibility and short deployment cycle. Supports flexible configuration of measurement distance L, compatible with a full range of measurement equipment such as magnetic probes and gyro probes, and comes with a standardized drilling rig control interface, reducing the integration and debugging cycle by 50%; combined with the working condition adaptive library, only 1-2 boreholes are needed for adaptation to new mining areas. Attached Figure Description

[0017] The above and other objects, features, and advantages of exemplary embodiments of the present invention will become readily apparent upon reading the following detailed description with reference to the accompanying drawings. In the drawings, several embodiments of the invention are illustrated by way of example and not limitation, with the same or corresponding reference numerals denoteing the same or corresponding parts, wherein: Figure 1 A schematic diagram illustrating the overall process of an adaptive learning measurement and control method for intelligent directional drilling rigs in underground coal mines is shown. Figure 2 A schematic diagram of the three-layer architecture of an adaptive learning measurement and control system for intelligent directional drilling rigs in underground coal mines is shown. Figure 3 A schematic diagram of the near-drill bit trajectory feedforward compensation formula is shown. Figure 4 The diagram illustrates the data flow and control logic of the tool-oriented angle solution model. Figure 5 The input-output logic diagram of the slope rate comprehensive correction model is shown schematically. Figure 6 The overall closed-loop control flowchart of the adaptive learning measurement and control method for intelligent directional drilling rigs in underground coal mines is schematically shown. Figure 7 The schematic diagram illustrates the modular structure of an adaptive learning measurement and control system for intelligent directional drilling rigs in underground coal mines. Detailed Implementation

[0018] Exemplary embodiments of the present invention are described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the invention and to fully convey the scope of the invention to those skilled in the art.

[0019] It should be noted that, unless otherwise stated, the technical or scientific terms used in this invention should have the ordinary meaning as understood by one of ordinary skill in the art.

[0020] The methods described in the embodiments of the present invention will be explained in detail below.

[0021] This embodiment provides an adaptive learning measurement and control method and system for intelligent directional drilling rigs in coal mines. This method and system can be widely applied to drilling measurement and control and automatic trajectory correction tasks in various directional drilling construction scenarios, such as gas extraction holes, water exploration holes, and holes for investigating hidden geological factors causing disasters in coal mines. It is compatible with various types of drilling measurement equipment, including magnetic probes and gyro probes, and is adapted to mainstream intelligent directional drilling rig control systems.

[0022] In this embodiment, the algorithm and data processing tasks can be deployed in the orifice control equipment, explosion-proof edge computing terminal, or underground explosion-proof control host of the intelligent directional drilling rig, according to the actual construction needs in the coal mine. The system as a whole adopts a three-layer closed-loop control architecture of "data acquisition layer - core processing layer - execution output layer": the drilling measurement instrument and drilling rig sensors collect dip angle, azimuth angle, drilling footage, and drilling rig operating parameters in real time; the near-bit feedforward compensation module, tool face angle calculation module, and self-learning control unit in the core processing layer complete attitude feedforward prediction, target tool face calculation, and back-end optimization of control parameters; finally, the generated target tool face angle control command is sent to the execution unit of the intelligent drilling rig through a standardized control interface, and the execution unit drives the drill string to complete the trajectory correction action.

[0023] See Figure 2 The measurement and control system running the method of this invention adopts a three-layer closed-loop architecture, including a data acquisition layer, a core processing layer, and an execution output layer, and is equipped with a working condition adaptive library to provide data support. The data acquisition layer is used to acquire drilling measurement data (dip angle, azimuth angle), trajectory deviation data (vertical deviation, horizontal deviation), and drilling rig sensor data (drill pressure, torque, feed displacement, etc.); the core processing layer includes a near-bit attitude compensation module, a tool face angle calculation module, a comprehensive working condition correction module, and a self-learning control unit; the execution output layer sends control commands to the intelligent drilling rig execution unit to drive the drill string deflection. The working condition adaptive library and the self-learning control unit interact bidirectionally, storing and iterating the optimal parameter set under different formations and working conditions. Before executing step 110, the method first performs system initialization and working condition adaptive library parameter retrieval, the specific steps of which are as follows: During the preparation stage of drilling or when entering a new formation, the control system retrieves an initial parameter set matching the current formation from the adaptive working condition library. This initial parameter set is used to initialize the formation build-up gain coefficient, deviation gain coefficient, comprehensive compensation angle, and single comprehensive correction coefficient. The adaptive working condition library stores the optimal parameter sets corresponding to different lithological formations and drill string combinations, and is iteratively updated based on over 2,500 sets of field data. During initialization, the formation build-up gain coefficient is set to 0.3 by default, and the single comprehensive correction coefficient is set to 1.0 by default. The vertical deviation gain coefficient, horizontal deviation gain coefficient, and comprehensive compensation angle are taken from the preset initial values ​​in the adaptive working condition library that match the current formation. Therefore, drilling in new mining areas or new formations can directly inherit parameters from similar historical formations, shortening the trajectory control adaptation time in the initial stage of drilling.

[0024] Step 110: Obtain the measurement-while-drilling data and trajectory deviation data including vertical and horizontal trajectory deviations for each measurement interval, and calculate the actual build-up rate of the current measurement point based on the measurement-while-drilling data and footage obtained in the continuous measurement intervals.

[0025] This step begins with acquiring drilling data and calculating trajectory deviations. The measurement interval can be flexibly configured according to the construction conditions and trajectory control accuracy requirements, for example, configured as one drill pipe length, two drill pipe lengths (conventional directional drilling conditions), or one meter of advance distance (high-precision directional drilling conditions). At the end of each measurement interval, real-time drilling measurement data is acquired using a drilling measurement instrument, including the dip angle, azimuth angle, and actual tool face angle of the current measurement point.

[0026] When the measurement-while-drilling instrument uses magnetic sensor-type measuring instruments (such as magnetic probes), data is acquired through its integrated triaxial accelerometer and triaxial magnetic sensor. The distance between the measuring unit and the drill bit is then measured. Typically 6 to 7 meters, the control system will default to... The setting is 6 meters. When the measurement-while-drilling instrument uses a gyro sensor-type measuring instrument (such as a gyro probe), data is acquired through its integrated triaxial accelerometer and triaxial gyro sensor. The distance between the measuring unit and the drill bit is then recorded. Approximately 3.5 to 4 meters. This refers to the distance between the measuring unit and the drill bit in actual working conditions. In this embodiment, the actual distance between the measuring unit and the drill bit can be manually input via the orifice control device. The model automatically performs parameter adaptation calculations. The drilling measurement data acquired by the aforementioned drilling measurement instruments specifically includes at least one of the dip angle, azimuth angle, and actual tool face angle of the current measurement point.

[0027] Specifically, the distance L between the measuring unit and the drill bit supports manual configuration and automatic adaptation: the default value for magnetic sensor measuring instruments is L=6m, and the default value for gyroscope sensor measuring instruments is L=3.5m; when the actual distance does not match the default value, the actual distance is input through the borehole control device, and the model automatically adapts and calculates.

[0028] Meanwhile, the trajectory calculation software of the borehole control equipment calculates and obtains trajectory deviation data in real time based on the drilling measurement data. This trajectory deviation data includes both vertical and horizontal borehole deviations. Specifically, the vertical borehole deviation... Defined as: ; in, The design vertical position (i.e., design elevation or design vertical coordinate) corresponding to the current measurement point is directly extracted from the preset borehole engineering design trajectory model and is stored in the control system in advance. The actual vertical position (i.e., actual elevation or actual vertical coordinates) of the current measurement point is derived and calculated by the trajectory calculation software of the wellhead control equipment based on drilling measurement data (dip angle, azimuth angle, and cumulative footage) using a logging trajectory extrapolation algorithm (such as the minimum curvature method or arc length method). The above formula is used to represent the deviation between the actual vertical position of the borehole and the designed vertical position, where... This indicates that the actual position of the borehole is too low and needs to be corrected upwards. This indicates that the actual position of the borehole is too high and needs to be corrected downwards; the borehole is deviated to the left or right. Defined as: ; in, The design horizontal position (i.e., design yaw or design lateral coordinate) corresponding to the current measurement point is directly extracted from the preset borehole engineering design trajectory model and is stored in the control system in advance. The actual horizontal position of the current measurement point (i.e., actual yaw or actual lateral coordinate) is also derived and calculated by the trajectory calculation software based on the drilling measurement data using a logging trajectory extrapolation algorithm. The above formula is used to represent the deviation between the actual horizontal position of the borehole and the designed horizontal position, where... This indicates that the actual position of the borehole is off to the left and needs to be corrected to the right. This indicates that the actual position of the borehole is off to the right and needs to be corrected to the left.

[0029] Based on the acquired dip angle, azimuth angle, and trajectory deviation data, the control system further incorporates the drilling footage collected by the drilling rig's sensors. The actual build-up rate reflecting the anisotropic characteristics of the formation at the current measurement point is calculated. Specifically, the actual build-up rate is calculated by dividing the difference between the dip angle and azimuth angle over two to three or more consecutive measurement intervals by the footage length. Calculations show that, in a physical sense, this is characterized by the rate of change of angle per unit advance, specifically including the rate of change of tilt angle per unit advance. (Unit: ° / m) and rate of change of azimuth angle per unit advance (Unit: ° / m). Among them, This represents the total difference in tilt angle over consecutive measurement intervals. This represents the total difference in azimuth angles within a continuous measurement interval. This represents the total drilling length corresponding to the continuous measurement interval. The calculated actual build-up rate (including components in both dip and azimuth directions) serves as the objective basis for subsequent drill bit attitude feedforward compensation and formation gain calculations.

[0030] Step 120: Based on the actual build-up rate, the distance between the measurement unit and the drill bit, and the formation build-up gain coefficient, perform feedforward compensation calculation and output the predicted value of the drill bit's true attitude after eliminating measurement lag.

[0031] This step eliminates measurement lag caused by the physical distance between the measurement-while-drilling (MWD) probe and the drill bit, outputting the predicted true attitude of the drill bit's cutting point. In directional drilling, the attitude data acquired by the MWD probe reflects the historical trajectory information of the probe's location, not the drill bit's current true attitude. This step uses the actual build-up rate (rate of change of dip angle and azimuth angle per unit footage) calculated at the measurement point as a basis, takes the structural distance from the measurement unit to the drill bit as the feedforward extrapolation length, and introduces a formation build-up gain coefficient to correct for build-up rate deviations caused by formation anisotropy. Geometric extrapolation is performed along the probe-to-drill-bit direction on the trajectory of the unmeasured section to calculate the predicted dip angle and azimuth angle at the drill bit. This feedforward compensation ensures that subsequent toolface angle calculations and trajectory corrections are based on the drill bit's true attitude, avoiding overcorrection or response lag caused by directly using lagging measurement data.

[0032] See Figure 3 In step 120, the system inputs the tilt angle of the measurement point. Azimuth and the unit advance rate, combined with the formation directional gain coefficient Along the lag distance Perform feedforward extrapolation to output the actual predicted inclination angle of the drill bit. With predicted azimuth When the inclination rate at a measurement point exceeds a preset anomaly threshold (e.g., 2° / m), it is determined to be a sudden change in formation or measurement noise, triggering [a certain event / action]. Correction range limit, single correction amount This helps prevent attitude prediction distortion and oscillation.

[0033] Specifically, the method for performing feedforward compensation calculation in step 120 is as follows: The predicted inclination angle in the true attitude prediction value of the drill bit is calculated based on the inclination angle at the measurement point, the rate of change of the inclination angle per unit advance, the distance between the measurement unit and the drill bit, and the formation directional drilling gain coefficient. : ; The predicted azimuth angle in the true attitude prediction value of the drill bit is calculated based on the azimuth angle of the measurement point, the rate of change of the azimuth angle per unit advance, the distance between the measurement unit and the drill bit, and the formation build-up gain coefficient. : ; in, To measure the inclination angle of the point, To measure the azimuth of the point, The rate of change of tilt angle per unit advance. The rate of change of azimuth angle per unit advance. To measure the distance between the unit and the drill bit, This is the stratigraphic skew gain coefficient.

[0034] Stratigraphic gain coefficient This measurement unit is used to quantify the amplification or attenuation effect of formation anisotropy (such as formation dip angle, joint development, and hardness differences) on the build-up rate. The initial default value is 0.3, derived from statistical fitting of historical field data from conventional coal seams. It also supports dynamic reverse correction during drilling by a self-learning unit based on actual correction errors. (Distance of measurement unit from drill bit) This refers to the spatial distance from the center of the probe sensor to the cutting surface of the drill bit, defaulting to 6m; magnetic sensor instruments. Typically, the range is 6 to 7 meters, while for gyroscope sensors it is 3.5 to 4 meters, depending on the actual situation. The data can be manually input via orifice control equipment or explosion-proof terminal, and the model will automatically adapt. After feedforward extrapolation, the output shows the actual predicted inclination and azimuth angles of the drill bit, eliminating overcorrection and slow response caused by measurement lag.

[0035] In a preferred embodiment, during the feedforward compensation calculation, to prevent abrupt changes in formation (such as the drill bit suddenly crossing the interface between soft and hard formations or a fault) or abnormal noise in sensor measurements from affecting the formation skew gain coefficient... In the event of severe oscillations and overfitting, the system integrates a coefficient update limiting protection mechanism to address abnormal tilting conditions. The specific execution logic is as follows: When the rate of change of dip angle per unit advance exceeds a preset tilting threshold, it is determined that the current tilting condition is abnormal, and the tilting gain coefficient of the formation is triggered. The correction range is limited, thus limiting the formation orthographic gain coefficient. The update magnitude in a single iteration is less than or equal to the preset correction upper limit. The preset tilting threshold is 2° / m, and the preset correction upper limit is 0.05.

[0036] Specifically, the self-learning control unit updates the stratigraphic tilting gain coefficient each time. Previously, the rate of change of the unit advance angle was monitored in real time. And compare it with the preset directional drilling threshold. Under the conventional operating conditions of screw drills in coal mines, this threshold is taken as... .when At this point, it is determined to be an abnormal directional drilling condition, possibly caused by measurement noise or impact at the formation interface. If the full update is directly performed using gradient descent in this situation... This can easily lead to drastic changes or divergence in parameters, thus triggering a limiting mechanism: limiting... The magnitude of a single update shall not exceed 0.05. If the absolute value of the original correction is greater than 0.05, it shall be truncated to 0.05. Update according to this upper limit step size. This mechanism can prevent abnormal operating conditions. Over-correction ensures the smooth convergence of the feedforward compensation model in complex formations.

[0037] Step 130: Based on the predicted true attitude of the drill bit, extrapolate the actual spatial coordinates of the drill bit, correct the vertical trajectory deviation and the horizontal trajectory deviation to obtain the corrected trajectory deviation that eliminates measurement lag; combine the deviation gain coefficients corresponding to the corrected trajectory deviations, the comprehensive compensation angle, and the single comprehensive correction coefficient covering drilling interference factors to calculate the target tool facing angle, and generate control commands based on the target tool facing angle and send them to the drilling rig for execution.

[0038] The original trajectory deviation was calculated based on the lagging coordinates of the measurement points, and directly using it for control would cause phase deviation. This step uses the actual coordinates of the measurement points as a reference, combines the drill bit's predicted attitude with the lag distance L, extrapolates the actual spatial coordinates of the drill bit's cutting position, and compares them with the coordinates corresponding to the designed trajectory to obtain the corrected deviation. 1. Calculate the vertical depth increment and horizontal displacement increment from the measurement point to the drill bit: ; 2. Decompose the horizontal displacement into the orthogonal coordinate system of the design trajectory according to the predicted azimuth angle to obtain the horizontal offset increment and vertical depth addition: 3. Calculate the difference between the actual coordinates and the design coordinates at the drill bit location, and obtain: ; The calculation benchmark for the corrected trajectory deviation is moved forward to the current cutting position of the drill bit, completely eliminating the phase difference caused by measurement lag, so that the subsequent correction control accurately corresponds to the true position of the drill bit, and avoids overcorrection and response lag from the root.

[0039] This step calculates the target tool facing angle based on the corrected trajectory deviation and issues control commands. Using the corrected vertical and horizontal trajectory deviations as inputs, they are multiplied by the corresponding deviation gain coefficients and then superimposed with a comprehensive compensation angle to obtain the basic tool facing angle. A single comprehensive correction coefficient is then introduced to correct the theoretical ramp rate, yielding the final target tool facing angle. This method takes into account both trajectory deviation and operating condition disturbances, providing a control output benchmark for self-learning.

[0040] Specifically, the method for calculating the target tool facing angle in step 130 is as follows: Based on the corrected vertical trajectory deviation and its corresponding vertical deviation gain coefficient, the corrected horizontal trajectory deviation and its corresponding horizontal deviation gain coefficient, and superimposed with the comprehensive compensation angle, the basic tool facing angle is calculated. : ; Based on the preset theoretical ramp rate and the single comprehensive correction coefficient, the actual ramp rate used to generate the control command is calculated. : ; Based on the aforementioned basic tools facing angle With the actual slope rate The target tool facing angle is calculated. in, This is due to the deviation in the vertical trajectory. This refers to the upper and lower deviation gain coefficients. For the left and right trajectory deviation, This refers to the left and right deviation gain coefficients. For comprehensive compensation angle, To construct the theoretical slope, It is a single comprehensive correction factor used to uniformly compensate for the combined effects of screw bend angle, drill pipe friction, formation characteristics, and drill bit size on directional drilling characteristics.

[0041] Specifically, the formula introduces a single comprehensive correction coefficient. It has a clear engineering physics significance, representing the ratio of the actual slope rate to the theoretical slope rate (i.e., This single coefficient comprehensively covers and calculates the combined effects of multiple complex drilling interference factors on directional drilling characteristics, including screw bend angle, drill pipe-hole wall friction resistance, formation characteristics (such as formation hardness and dip angle), and drill bit geometry. During the system initialization phase, this single comprehensive correction coefficient... The initial default value is set to 1.0 (representing an ideal, interference-free operating condition baseline). Its reasonable physical value range is limited to 0.7 to 1.2. The initial default value is 1.0. When there is no valid field data to support correction, the default value is set to 1.0. This maintains the basic solution logic. Furthermore, this embodiment features a robust fault-tolerant protection design: when downhole data transmission is interrupted or there is no effective field data to support real-time parameter correction, the system automatically locks. The default baseline value is used to ensure the smooth execution of the basic calculation logic and to ensure the stable and safe operation of the control system under extreme conditions of data deficiency.

[0042] The comprehensive compensation angle introduced in the formula This is used to centrally absorb and eliminate systematic angular deviations during dynamic drilling, specifically including the counter-torsion angle generated by the screw motor startup, the tool face assembly angle difference between the measurement-while-drilling probe and the screw bend, and the elastic torsional deformation of the kilometer-long drill string within the borehole. During the system initialization phase, the comprehensive compensation angle... The initial values ​​are taken from preset empirical values ​​or a pre-established adaptive parameter library for working conditions; during subsequent dynamic drilling, Instead of remaining fixed, it is one of the core parameters in the set of parameters to be corrected. The subsequent self-learning control unit updates it in real time based on the gradient descent algorithm according to the feedback control error, thereby dynamically adapting to the change in torsion angle caused by the increase in drill string length.

[0043] Based on the aforementioned calculated basic tool facing angle Compared with the actual slope rate The system calculates the target tool facing angle to be sent to the intelligent drilling rig and generates standardized control commands accordingly. These commands are then sent to the intelligent drilling rig execution unit via a closed-loop control interface to drive the drill string to deflect, thereby driving the downhole opening / direction mechanism to rotate the screw bend and adjust it to the target angle phase.

[0044] Please refer to Figure 4 , Figure 4 The diagram visually illustrates the data flow and control logic of the tool-oriented angle-based solution model. For example... Figure 4 As shown, the model receives the measured upper and lower deviations. Left and right deviation Combined with the gain coefficient input in real time by the self-learning unit and comprehensive compensation angle Solve and output the optimal tool face angle for the current measurement interval. .

[0045] Please refer to Figure 5 , Figure 5 The input-output logic diagram of the slope rate comprehensive correction model is shown. For example... Figure 5 As shown, the model will use the preset theoretical slope. Multiplied by a single comprehensive correction factor updated in reverse by the self-learning unit. The product is used to output the corrected actual slope. This is used to generate the final control commands issued.

[0046] Step 140: After the drilling rig completes drilling for the current measurement interval based on the control command, the current trajectory deviation and the actual tool face angle are obtained; based on the current trajectory deviation and the tool face error between the calculated target tool face angle and the actual tool face angle, the comprehensive control error is calculated.

[0047] This step is used to calculate the comprehensive control error. After the drilling rig completes drilling in the current measurement interval, it acquires the current trajectory deviation calculated by the wellhead control equipment and the actual tool face angle fed back by the drilling rig. The difference between the actual tool face angle and the target tool face angle is used to obtain the tool face error. The trajectory deviation and tool face error are weighted and fused to obtain the comprehensive control error, where the trajectory deviation has a higher weight than the tool face error. This error reflects both the spatial position control result and the tool face execution accuracy, providing a feedback benchmark for subsequent self-learning. After the drilling rig completes drilling in the current measurement interval based on the control command issued in step 130, the measurement-while-drilling instrument and the wellhead trajectory calculation software execute the measurement-while-drilling status acquisition and calculation action again (i.e., repeat the data acquisition and calculation mechanism similar to step 110), and re-acquire the latest current trajectory deviation (including the latest vertical trajectory deviation) generated after the drilling rig performs the correction action. Deviation from left and right trajectories (and read the actual tool face angle of the current measurement interval measured by the measurement-while-drilling sensor). .

[0048] Specifically, the method for calculating the comprehensive control error in step 140 is as follows: Calculate the tool face error based on the target tool face angle and the actual tool face angle; Based on preset trajectory deviation weights and tool surface error weights, the trajectory deviation and the tool surface error are weighted and summed to calculate the comprehensive control error. : ; in, For trajectory deviation weights, For tool surface error weights, This is due to the deviation in the vertical trajectory. For the left and right trajectory deviation, For the target tool facing angle, This refers to the actual tool face angle. The trajectory deviation weight and tool face error weight can be dynamically configured based on the drilling type; the default weight is [value missing]. The weighting configuration under the condition of exploring and releasing water holes is as follows: , .

[0049] Specifically, regarding the weights in the formula and The configuration is based on the actual control priorities in underground directional drilling projects in coal mines, namely, "trajectory spatial position accuracy is more critical than tool face control response accuracy." Under the system's default baseline conditions, the trajectory deviation weight is set to... The tool surface error weight is set to .

[0050] Preferably, the weights are dynamically configured and fine-tuned according to different drilling construction purposes and safety risk levels through orifice control equipment or explosion-proof terminals to accurately adapt to different drilling operations.

[0051] When the current drilling is a high-risk exploration and drainage hole, to strictly prevent the borehole from deviating from the target area and penetrating the aquifer or causing a water inrush accident, the system prioritizes trajectory position accuracy to the highest level. In this case, the control system or operator can increase the trajectory deviation weight. The toolface error weight will be reduced accordingly. This allows the control algorithm to concentrate computing power and response amplitude to prioritize reducing spatial geometric position deviations.

[0052] When constructing conventional directional gas extraction boreholes, the system automatically resumes operation due to the relatively stable tolerance of the trajectory space. The default baseline weight ratio ensures both trajectory accuracy and the smoothness of tool surface movements.

[0053] Step 150: Based on the dominant characteristics of the current trajectory deviation and tool face error in the comprehensive control error, and the abnormal situation of the actual pitching rate, the error attribution is mapped to the corresponding set of parameters to be corrected. The set of parameters to be corrected includes at least one of the formation pitching gain coefficient, the deviation gain coefficient, the comprehensive compensation angle, and the single comprehensive correction coefficient.

[0054] This step determines the parameters to be corrected based on the dominant component of the comprehensive control error, avoiding oscillations caused by simultaneous correction of all parameters. When trajectory deviation is dominant, the focus is on correcting the deviation gain coefficient; when tool surface error is dominant, the focus is on correcting the single comprehensive correction coefficient or the comprehensive compensation angle. Accordingly, parameters directly related to the current error are selected from the parameter set, which includes the formation orthographic gain coefficient, deviation gain coefficient, and single comprehensive correction coefficient, thus narrowing the parameter optimization search space.

[0055] Based on the above principles, the specific method for mapping error attribution to the corresponding set of parameters to be corrected in step 150 is as follows: Step 151: When the current trajectory deviation dominates the overall control error, select the deviation gain coefficients corresponding to the vertical trajectory deviation, the horizontal trajectory deviation, and the overall compensation angle as the set of parameters to be corrected.

[0056] The system first calculates the weighted contribution ratio of the normalized trajectory deviation within the current measurement interval to the overall control error. When the current trajectory deviation dominates the overall control error, i.e., when the weighted contribution ratio is greater than a preset first threshold (e.g., the first threshold is set to 60%), the control system determines that the spatial geometric position of the current borehole has deviated macroscopically and locks the set of parameters to be corrected. (in This refers to the upper and lower deviation gain coefficients. This refers to the left and right deviation gain coefficients. (For comprehensive compensation angle). In terms of physical mechanism, when spatial position deviation is dominant, the gain coefficient responsible for the sensitivity of the response in the orthogonal spatial dimension is directly adjusted. and Combined with geometric phase offset angle The translational correction can drive the drill string to pull back its trajectory with the most direct angular correction, quickly reducing spatial geometric position drift, while avoiding blind intervention in other parameters such as formation directional drilling.

[0057] Geometric deviations in spatial location (such as wellbore misalignment) have the greatest impact on engineering safety. Setting a high threshold (60%) ensures that the deviation gain coefficient is only significantly adjusted when a substantial macroscopic deviation in location actually occurs, preventing excessive parameter adjustments caused by small positional fluctuations.

[0058] Step 152: When the deviation between the actual build-up rate and the theoretical build-up rate exceeds a preset threshold, the formation build-up gain coefficient and the single comprehensive correction coefficient are selected as the set of parameters to be corrected.

[0059] The system compares in real time the deviation between the actual build-up rate calculated from the measured attitude and the preset build-up rate of the current model, and calculates the weighted contribution ratio of abnormal build-up rate deviation to the overall control error. When this weighted contribution ratio is greater than a preset second threshold (e.g., the second threshold is set to 50%), the system determines that the current control error mainly originates from abrupt changes in formation conditions (such as the drill bit crossing the coal-rock interface or soft-hard interlayers) or a mismatch in build-up efficiency caused by physical wear of the drill bit's cutting teeth. At this time, the system locks the set of parameters to be corrected as follows: (in This is the stratigraphic skew gain coefficient. (This is a single comprehensive correction coefficient). Physically, the build-up rate anomaly directly reflects changes in "formation anisotropy" and "actual build-up output capacity of the drill string assembly." The correction target is set as... and It can compensate for the deviation in the build-up response caused by changes in formation hardness gradient and tool mechanical efficiency without destroying the macroscopic trajectory deviation response weights, ensuring that the model can quickly adapt to the build-up characteristics of new formations downhole.

[0060] An abnormal build-up rate reflects a sudden change in formation hardness or drill bit wear. When it accounts for more than half (50%), it indicates that the current formation or drill string characteristics have undergone substantial changes, and fine-tuning must be intervened. and .

[0061] Step 153: When the tool surface error dominates the overall control error, select the single overall correction coefficient and the overall compensation angle as the set of parameters to be corrected; the single correction magnitude of each parameter shall not exceed 10% of its initial value.

[0062] The system calculates the weighted contribution ratio of the normalized tool face angle deviation error to the overall control error. When this weighted contribution ratio exceeds a preset third threshold (e.g., the third threshold is set to 40%), the control system determines that the current control error mainly originates from the angle response lag of the downhole mechanical steering actuator, the violent fluctuation of the screw motor's counter-torque, or the elastic torsional deformation of the kilometer-long drill string. At this point, the system locks the set of parameters to be corrected as follows: (in For a single comprehensive correction coefficient, (For comprehensive compensation angle). Physically, the dominance of tool surface error implies a significant phase offset or dynamic lag between the "command issued angle" and the "actual deflection angle." This is achieved by preferentially absorbing the anti-torsional damping during bundling. With respect to the translation of the angular reference Reverse iterative fine-tuning can quickly compensate for drill string anti-torsion angle and dynamic angle lag, eliminate overshoot and jitter in face angle control, and effectively suppress the occurrence of reciprocating angle oscillation of downhole azimuth mechanism.

[0063] The tool face angle is extremely sensitive to screw counter-torque and kilometer-long drill string torsion. Even a slight increase in angle deviation (reaching 40%) requires immediate intervention and correction. To prevent high-frequency oscillations in the steering mechanism, its trigger sensitivity is relatively higher.

[0064] Step 160: Using the gradient descent algorithm, with the goal of minimizing the integrated control error, calculate the correction amount of each parameter in the parameter set to be corrected in reverse along the error gradient and update the parameter values. Then, feed back the updated parameter set to be corrected to the next measurement interval to replace the original parameters and perform the attitude feedforward compensation calculation and the target tool facing angle calculation respectively.

[0065] This step uses the comprehensive control error as the objective function and employs a gradient descent algorithm to update each parameter to be corrected along the negative direction of the error gradient. The updated parameter set is fed back to the next measurement interval, replacing the original parameters to participate in the feedforward compensation in step 120 and the tool face angle calculation in step 130, forming a closed-loop adaptive control.

[0066] Specifically, the method for calculating the correction amount of each parameter in the parameter set to be corrected and updating the parameter values ​​in step 160 is as follows: Step 161: Calculate the error gradient of the integrated control error relative to the parameter to be corrected based on the numerical differentiation method, and update the parameter to be corrected in reverse according to the product of the preset learning rate and the error gradient.

[0067] First, the local partial derivative of the current integrated control error with respect to the selected parameter to be corrected, i.e., the error gradient, is obtained using numerical differentiation. Specifically, a small perturbation is applied to the current value of the parameter to be corrected, and the difference in the integrated control error before and after the perturbation is calculated to obtain the error gradient, which characterizes the parameter's sensitivity to error. Then, the parameter to be corrected is updated in reverse using the following gradient descent update formula: ; in, The current value of the parameter to be corrected selected in the aforementioned steps before updating, the parameter to be corrected includes the formation tilting gain coefficient. Upper and lower deviation gain coefficients Left and right deviation gain coefficients Comprehensive compensation angle and single comprehensive correction coefficient At least one of them; The error gradient is calculated using numerical differentiation. The learning rate hyperparameter is preset and its default value is set to 0.02 during the system initialization phase. Its engineering function is to regulate the iteration step size of a single backward update and prevent parameter adjustment overshoot or oscillation caused by excessive gradient. This refers to the new value of the parameter to be corrected calculated after the reverse update. This new value will be output in real time and written to the system storage unit to replace the original parameter value and feed it back to the next measurement interval, serving as the adaptive reference for performing attitude feedforward compensation calculation and target tool facing angle calculation in the next measurement interval.

[0068] Step 162: After the update, determine whether the parameter value exceeds the preset value range. If the parameter value exceeds the preset value range, forcibly truncate the parameter value to the boundary value of the preset value range.

[0069] This step performs boundary checks and truncation on the parameters updated in step 161. Among them, the formation eccentricity gain coefficient... The magnitude of a single correction shall not exceed 10% of its initial value of 0.3, i.e., 0.27 to 0.33; single comprehensive correction coefficient Limited to 0.7–1.2; comprehensive compensation angle Limited to -30° to 30°; upper and lower deviation gain coefficients Gain coefficient with left and right deviations The parameters are limited to a reasonable range of positive real numbers. If the updated parameters exceed the corresponding range, they are forcibly truncated to the nearest boundary value; otherwise, they are retained. The validated parameter set is written to the storage unit for use in the next measurement interval.

[0070] In a preferred embodiment, before updating the parameter set to be corrected using the gradient descent algorithm, a pre-emptive control error dead zone gating trigger mechanism is introduced to prevent high-frequency oscillations of control parameters caused by downhole measurement-while-drilling noise and drill string vibration interference. Specifically, this triggering control can be performed in the following three steps: Determine whether the overall control error is greater than a preset error threshold, wherein the preset error threshold is 0.05; If the overall control error is greater than the preset error threshold, then the gradient descent algorithm is started to update the set of parameters to be corrected in reverse. If the overall control error is less than or equal to the preset error threshold, the value of the current set of parameters to be corrected remains unchanged and is used in the next measurement interval.

[0071] Specifically, a dead-zone gate is set before starting gradient descent: when the overall control error... When the default value is 0.05, the residual is considered measurement noise, and all parameters are kept unchanged and used in the next measurement interval; when In this process, the set of parameters to be corrected is first locked through error attribution, and then gradient descent is initiated for updating. This mechanism avoids overfitting to small noise points and suppresses high-frequency parameter oscillations and inefficient computational overhead.

[0072] See Figure 6 , Figure 6 The overall closed-loop control process of the method of the present invention is shown, including the following steps: Step 1: Load initial parameters and start self-learning initialization; Step 2: Collect drilling attitude and position status data in real time during each measurement interval; Step 3: Calculate the comprehensive control error of the current measurement interval. Step 4 will Error threshold with respect to preset dead zone (For example, 0.05) Perform gating judgment: If Keep the current parameters unchanged and use them in the next measurement interval; if If the error is detected, self-learning correction is triggered, proceeding to step 5; step 5 performs error attribution allocation based on the error-dominant characterization, locking the parameter set to be corrected; step 6 uses numerical differentiation to perform gradient descent self-correction along the error gradient; step 7 recalculates the target tool face angle based on the updated parameters; step 8 issues instructions to the intelligent drilling rig for execution, and then loops into the next measurement interval. Step 160 corresponds to... Figure 6In step 6, the comprehensive control error is used as the target loss function. The gradient descent algorithm is used to update the parameter set to be corrected, and the updated parameters are fed back to the next measurement interval. The original parameters are used to participate in the attitude feedforward compensation calculation in step 120 and the target tool face angle calculation in step 130, respectively.

[0073] This invention constructs a "feedforward prediction—dimensionality reduction solution—attribution mapping—closed-loop iteration" adaptive control mechanism while drilling. Addressing the measurement lag caused by the physical distance between the measuring tube and the drill bit during drilling, the actual build-up rate and the formation build-up gain coefficient are used to spatially feedforward extrapolate the drill bit attitude, outputting a predicted value of the drill bit's true attitude that eliminates lag, thus avoiding control delay and overcorrection based on lag data. To address the problem that traditional solutions do not integrate multi-source physical interference and lack self-learning capabilities, a single comprehensive correction coefficient is used to uniformly compensate for the combined effects of screw bend angle, drill pipe friction, formation characteristics, and drill bit size on the build-up rate. Based on a comprehensive control error attribution mapping and gradient descent algorithm for each measurement interval, the deviation gain coefficient, formation build-up gain coefficient, and single comprehensive correction coefficient are updated in a targeted manner. This avoids numerical oscillations caused by indiscriminate correction of all parameters, suppresses accumulated trajectory deviation, reduces the frequency of correction, and improves trajectory control accuracy and adaptability under complex and abrupt formation conditions.

[0074] Based on the same inventive concept, as an implementation of the above-mentioned adaptive learning measurement and control method for intelligent directional drilling rigs in coal mines, this embodiment of the invention also provides an adaptive learning measurement and control system for intelligent directional drilling rigs in coal mines. Figure 7 This is a structural diagram of the system in an embodiment of the present invention. See also: Figure 7 As shown, the system is deployed in an explosion-proof computing terminal in a coal mine and is electrically connected to the measurement-while-drilling probe and the intelligent drilling rig actuator. The system includes: The data and build-up rate acquisition module is used to acquire the drilling measurement data and trajectory deviation data including vertical and horizontal trajectory deviations for each measurement interval, and to calculate the actual build-up rate of the current measurement point based on the drilling measurement data and footage of the continuous measurement interval. The attitude feedforward compensation module is used to perform feedforward compensation calculations based on the actual build-up rate, the distance between the measurement unit and the drill bit, and the formation build-up gain coefficient, and output the true attitude prediction value of the drill bit that eliminates measurement lag. The tool face calculation module is used to correct the vertical trajectory deviation and the horizontal trajectory deviation based on the actual attitude prediction value of the drill bit to obtain the corrected trajectory deviation after eliminating measurement lag; combined with the deviation gain coefficient corresponding to each of the corrected trajectory deviations, the comprehensive compensation angle, and the single comprehensive correction coefficient covering drilling interference factors, the target tool face angle is calculated, and a control command is generated based on the target tool face angle and sent to the drilling rig for execution. The error assessment module is used to acquire the current trajectory deviation and the actual tool face angle after the drilling rig completes drilling at the current measurement interval based on the control command; and to calculate the comprehensive control error based on the current trajectory deviation and the tool face error between the calculated target tool face angle and the actual tool face angle. The error attribution module is used to map the error attribution to the corresponding set of parameters to be corrected based on the dominant characteristics of the current trajectory deviation and tool face error in the comprehensive control error and the abnormal situation of the actual inclination rate. The set of parameters to be corrected includes at least one of the formation inclination gain coefficient, the deviation gain coefficient, the comprehensive compensation angle and the single comprehensive correction coefficient. The self-learning control module is used to employ a gradient descent algorithm to calculate the correction amount of each parameter in the parameter set to be corrected along the error gradient and update the parameter values, with the goal of minimizing the comprehensive control error. The updated parameter set to be corrected is then fed back to the next measurement interval to replace the original parameters for the attitude feedforward compensation calculation and the target tool facing angle calculation.

[0075] The attitude feedforward compensation module is also used to: determine that the current situation is an abnormal directional drilling condition when the rate of change of the dip angle per unit advance is greater than the preset directional drilling threshold, and limit the update amplitude of the formation directional drilling gain coefficient in a single iteration to be less than or equal to the preset correction upper limit value.

[0076] The self-learning control module is also used to: determine whether the comprehensive control error is greater than a preset error threshold; if it is greater than the threshold, start the gradient descent algorithm to update the parameters; if it is less than or equal to the threshold, keep the current parameters unchanged.

[0077] It also includes a working condition adaptive library module, which stores the optimal parameter sets corresponding to different lithological formations and different drill bit combinations, and retrieves the initial parameter set that matches the current formation during the drilling initialization stage.

[0078] This invention constructs a "feedforward prediction—dimensionality reduction solution—attribution mapping—closed-loop iteration" adaptive control mechanism while drilling. Addressing the measurement lag caused by the physical distance between the measuring tube and the drill bit during drilling, the actual build-up rate and the formation build-up gain coefficient are used to spatially feedforward extrapolate the drill bit attitude, outputting a predicted value of the drill bit's true attitude that eliminates lag, thus avoiding control delay and overcorrection based on lag data. To address the problem that traditional solutions do not integrate multi-source physical interference and lack self-learning capabilities, a single comprehensive correction coefficient is used to uniformly compensate for the combined effects of screw bend angle, drill pipe friction, formation characteristics, and drill bit size on the build-up rate. Based on a comprehensive control error attribution mapping and gradient descent algorithm for each measurement interval, the deviation gain coefficient, formation build-up gain coefficient, and single comprehensive correction coefficient are updated in a targeted manner. This avoids numerical oscillations caused by indiscriminate correction of all parameters, suppresses accumulated trajectory deviation, reduces the frequency of correction, and improves trajectory control accuracy and adaptability under complex and abrupt formation conditions.

[0079] This invention also provides an electronic device comprising: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a program for an adaptive learning measurement and control method for intelligent directional drilling rigs in coal mines. When the processor executes the computer program, it implements all the steps in the aforementioned adaptive learning measurement and control method for intelligent directional drilling rigs in coal mines, such as S110 to S160 and their sub-steps. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the aforementioned system embodiments.

[0080] For example, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the electronic device.

[0081] The electronic device may be a computing device such as a server or cloud server deployed in an adaptive learning measurement and control system for intelligent directional drilling rigs in coal mines. The electronic device may include, but is not limited to, processors and memory. Those skilled in the art will understand that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. It may include more or fewer components than described above, or a combination of certain components, or different components. For example, the electronic device may also include data acquisition cards and output cards for connecting sensors and actuators, network communication equipment, buses, etc.

[0082] The processor can be a central processing unit, or other general-purpose processors, digital signal processors, application-specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (OPGs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the electronic device, connecting various parts of the entire electronic device via various interfaces and lines.

[0083] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function, etc.; the data storage area may store data created according to system usage, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, RAM, plug-in hard disk, smart memory card, secure digital card, flash memory card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0084] Wherein, if the modules / units integrated in the electronic device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory, a random access memory, an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0085] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. An adaptive learning measurement and control method for intelligent directional drilling rigs in coal mines, characterized in that, The method includes: Acquire the drilling measurement data and trajectory deviation data including vertical and horizontal trajectory deviations for each measurement interval, and calculate the actual build-up rate at the current measurement point based on the drilling measurement data and footage from the continuous measurement intervals. Based on the actual build-up rate, the distance between the measurement unit and the drill bit, and the formation build-up gain coefficient, feedforward compensation calculation is performed to output the predicted value of the drill bit's true attitude that eliminates measurement lag. Based on the predicted true attitude of the drill bit, the actual spatial coordinates at the drill bit are extrapolated, and the vertical and horizontal trajectory deviations at the original measurement point are corrected to obtain the corrected trajectory deviations that eliminate measurement lag. Combining the deviation gain coefficients corresponding to the corrected trajectory deviations, the comprehensive compensation angle, and the single comprehensive correction coefficient covering drilling interference factors, the target tool facing angle is calculated, and a control command is generated based on the target tool facing angle and sent to the drilling rig for execution. After the drilling rig completes drilling for the current measurement interval based on the control command, the current trajectory deviation and the actual tool face angle are obtained; based on the current trajectory deviation and the tool face error between the calculated target tool face angle and the actual tool face angle, the comprehensive control error is calculated; Based on the dominant characteristics of the current trajectory deviation and tool face error in the comprehensive control error, as well as the abnormal situation of the actual inclination rate, the error attribution is mapped to the corresponding set of parameters to be corrected. The set of parameters to be corrected includes at least one of the formation inclination gain coefficient, the deviation gain coefficient, the comprehensive compensation angle, and the single comprehensive correction coefficient. Using the gradient descent algorithm, with the goal of minimizing the integrated control error, the correction amount of each parameter in the parameter set to be corrected is calculated in reverse along the error gradient and the parameter values ​​are updated. The updated parameter set to be corrected is then fed back to the next measurement interval to replace the original parameters for the attitude feedforward compensation calculation and the target tool facing angle calculation.

2. The method according to claim 1, characterized in that, The feedforward compensation calculation based on the actual build-up rate, the distance between the measurement unit and the drill bit, and the formation build-up gain coefficient, outputs a predicted value of the drill bit's true attitude that eliminates measurement lag, including: The predicted inclination angle in the true attitude prediction value of the drill bit is calculated based on the inclination angle at the measurement point, the rate of change of inclination angle per unit advance, the distance between the measurement unit and the drill bit, and the formation directional drilling gain coefficient. : ; The predicted azimuth angle in the true attitude prediction value of the drill bit is calculated based on the azimuth angle of the measurement point, the rate of change of azimuth angle per unit advance, the distance between the measurement unit and the drill bit, and the formation build-up gain coefficient. : ; in, To measure the inclination angle of the point, To measure the azimuth of the point, The rate of change of tilt angle per unit advance. The rate of change of azimuth angle per unit advance. To measure the distance between the unit and the drill bit, This is the stratigraphic skew gain coefficient.

3. The method according to claim 2, characterized in that, The feedforward compensation calculation also includes an anomaly protection step: when the rate of change of dip angle per unit advance exceeds a preset directional drilling threshold, it is determined that the current directional drilling condition is abnormal, and the directional drilling gain coefficient of the formation is triggered. The correction range is limited, thus limiting the formation orthographic gain coefficient. The update magnitude in a single iteration is less than or equal to the preset correction upper limit; The preset tilting threshold is 2° / m, and the preset correction upper limit is 0.

05.

4. The method according to claim 1, characterized in that, The target tool facing angle is calculated by combining the deviation gain coefficients corresponding to the corrected trajectory deviations, the comprehensive compensation angle, and the single comprehensive correction coefficient covering drilling interference factors, including: Based on the corrected vertical trajectory deviation and its corresponding vertical deviation gain coefficient, the corrected horizontal trajectory deviation and its corresponding horizontal deviation gain coefficient, and superimposed with the comprehensive compensation angle, the basic tool facing angle is calculated. : ； Based on the preset theoretical ramp rate and the single comprehensive correction coefficient, the actual ramp rate used to generate the control command is calculated. : ； Based on the aforementioned basic tool facing angle With the actual slope rate The target tool facing angle is calculated. in, To correct the deviation in the upper and lower trajectories. This refers to the upper and lower deviation gain coefficients. To correct the left and right trajectory deviation, The left and right deviation gain coefficients. For comprehensive compensation angle, To construct the slope for theoretical purposes, It is a single comprehensive correction factor used to uniformly compensate for the combined effects of screw bend angle, drill pipe friction, formation characteristics, and drill bit size on directional drilling characteristics.

5. The method according to claim 4, characterized in that, The single comprehensive correction coefficient The value range is 0.7 to 1.2, and the initial default value is 1.0; When there is no valid field data to support the correction, the default is... To maintain the operation of the basic solution logic.

6. The method according to claim 1, characterized in that, The calculation of the comprehensive control error based on the current trajectory deviation and the calculated tool face error between the target tool face angle and the actual tool face angle includes: Calculate the tool face error based on the target tool face angle and the actual tool face angle; Based on preset trajectory deviation weights and tool surface error weights, the trajectory deviation and the tool surface error are weighted and summed to calculate the comprehensive control error. : ； in, For trajectory deviation weights, For tool surface error weights, This is due to the deviation in the vertical trajectory. For the deviation of the left and right trajectories, For the target tool facing angle, For actual tool orientation angle; The trajectory deviation weight and tool face error weight can be dynamically configured according to the drilling type, with a default weight of [value missing]. The weighting configuration under the condition of exploring and releasing water holes is as follows: .

7. The method according to claim 1, characterized in that, The step of mapping the error attribution to the corresponding set of parameters to be corrected based on the dominant characteristics of the current trajectory deviation and tool surface error in the comprehensive control error includes: When the weighted contribution of trajectory deviation is greater than 60%, it is determined that the current trajectory deviation is dominant, and the vertical deviation gain coefficient, the horizontal deviation gain coefficient, and the comprehensive compensation angle are selected as the set of parameters to be corrected. When the deviation between the actual build-up rate and the theoretical build-up rate exceeds 20%, it is determined that the build-up rate anomaly is dominant, and the formation build-up gain coefficient and the single comprehensive correction coefficient are selected as the set of parameters to be corrected. When the weighted contribution of tool surface error is greater than 40%, it is determined that tool surface error is dominant, and the single comprehensive correction coefficient and comprehensive compensation angle are selected as the set of parameters to be corrected.

8. The method according to claim 1, characterized in that, The step of calculating the correction amount of each parameter in the set of parameters to be corrected in reverse along the error gradient and updating the parameter values ​​includes: The error gradient of the integrated control error relative to the parameter to be corrected is calculated based on the numerical differentiation method, and the parameter to be corrected is updated in reverse according to the product of the preset learning rate and the error gradient. After the update, it checks whether the parameter value exceeds the preset value range. When the parameter value exceeds the preset value range, the parameter value is forcibly truncated and limited to the boundary value of the preset value range. The single adjustment range of each parameter does not exceed 10% of its initial value.

9. The method according to claim 1, characterized in that, Before employing the gradient descent algorithm, the method further includes: Determine whether the overall control error is greater than a preset error threshold; If the overall control error is greater than the preset error threshold, then the gradient descent algorithm is started to update the set of parameters to be corrected in reverse. If the overall control error is less than or equal to the preset error threshold, the value of the current set of parameters to be corrected remains unchanged and is used until the next measurement interval. The preset error threshold is 0.

05.

10. The method according to claim 1, characterized in that, Before acquiring the measurement-while-drilling data for each measurement interval and the trajectory deviation data including vertical and horizontal trajectory deviations, the method further includes: Retrieve and retrieve the initial parameter set matching the current drilling formation from the preset adaptive working condition library, and initialize the formation build-up gain coefficient, the deviation gain coefficient, the comprehensive compensation angle, and the single comprehensive correction coefficient to the corresponding preset initial values ​​in the initial parameter set; The adaptive working condition library stores the optimal parameter sets corresponding to different lithological formations and different drilling tool combinations, and is iteratively updated based on more than 2,500 sets of field drilling data.

11. The method according to claim 2, characterized in that, The distance L between the measuring unit and the drill bit can be manually configured and automatically adapted: the default value for magnetic sensor measuring instruments is L=6m, and the default value for gyroscope sensor measuring instruments is L=3.5m; when the actual distance does not match the default value, the actual distance can be input through the borehole control device, and the model will automatically adapt and calculate.

12. An adaptive learning measurement and control system for intelligent directional drilling rigs in coal mines, characterized in that, The system is deployed in an explosion-proof computing terminal in a coal mine and is electrically connected to the measurement-while-drilling probe and the intelligent drilling rig actuator. The system includes: The data and build-up rate acquisition module is used to acquire the drilling measurement data and trajectory deviation data including vertical and horizontal trajectory deviations for each measurement interval, and to calculate the actual build-up rate of the current measurement point based on the drilling measurement data and footage of the continuous measurement interval. The attitude feedforward compensation module is used to perform feedforward compensation calculations based on the actual build-up rate, the distance between the measurement unit and the drill bit, and the formation build-up gain coefficient, and output the true attitude prediction value of the drill bit that eliminates measurement lag. The tool face calculation module is used to correct the vertical trajectory deviation and the horizontal trajectory deviation based on the actual attitude prediction value of the drill bit to obtain the corrected trajectory deviation after eliminating measurement lag; combined with the deviation gain coefficient corresponding to each of the corrected trajectory deviations, the comprehensive compensation angle, and the single comprehensive correction coefficient covering drilling interference factors, the target tool face angle is calculated, and a control command is generated based on the target tool face angle and sent to the drilling rig for execution. The error assessment module is used to acquire the current trajectory deviation and the actual tool face angle after the drilling rig completes drilling at the current measurement interval based on the control command; and to calculate the comprehensive control error based on the current trajectory deviation and the tool face error between the calculated target tool face angle and the actual tool face angle. The error attribution module is used to map the error attribution to the corresponding set of parameters to be corrected based on the dominant characteristics of the current trajectory deviation and tool face error in the comprehensive control error and the abnormal situation of the actual inclination rate. The set of parameters to be corrected includes at least one of the formation inclination gain coefficient, the deviation gain coefficient, the comprehensive compensation angle and the single comprehensive correction coefficient. The self-learning control module is used to employ a gradient descent algorithm to calculate the correction amount of each parameter in the parameter set to be corrected along the error gradient and update the parameter values, with the goal of minimizing the comprehensive control error. The updated parameter set to be corrected is then fed back to the next measurement interval to replace the original parameters for the attitude feedforward compensation calculation and the target tool facing angle calculation.

13. The system according to claim 12, characterized in that, The attitude feedforward compensation module is also used to: determine that the current situation is an abnormal directional drilling condition when the rate of change of the unit advance dip angle is greater than the preset directional drilling threshold, and limit the update amplitude of the formation directional drilling gain coefficient in a single iteration to be less than or equal to the preset correction upper limit value.

14. The system according to claim 12, characterized in that, The self-learning control module is also used to: determine whether the comprehensive control error is greater than a preset error threshold; if it is greater than the threshold, start the gradient descent algorithm to update the parameters; if it is less than or equal to the threshold, keep the current parameters unchanged.

15. The system according to claim 12, characterized in that, It also includes a working condition adaptive library module, which stores the optimal parameter sets corresponding to different lithological formations and different drill bit combinations, and retrieves the initial parameter set that matches the current formation during the drilling initialization stage.

16. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the adaptive learning measurement and control method for intelligent directional drilling rigs in coal mines as described in any one of claims 1 to 11.

Images