Real-time borehole trajectory steering control system and method for wireless measurement-while-drilling instrument

Abstract

The application discloses a borehole trajectory real-time guiding control system and method of a wireless measurement-while-drilling instrument, and belongs to the technical field of drilling engineering. The system comprises the following steps: collecting borehole posture data and environmental factor data through the wireless measurement-while-drilling instrument, and generating fusion data; generating adaptive control parameters through an adaptive learning mechanism based on the fusion data and historical guiding effect data; performing channel adaptation processing and dynamic compensation on the adaptive control parameters to generate a compensation control signal; controlling the operating posture of a guiding tool by using the compensation control signal, calculating the deviation between actual borehole trajectory data and preset target borehole trajectory data, and generating trajectory deviation data; and updating the learning parameters of the adaptive learning mechanism based on the trajectory deviation data. The application adopts multi-source data fusion perception, intelligent decision based on an adaptive learning mechanism, dynamic compensation communication based on channel adaptation, and closed-loop feedback updating based on trajectory deviation, and can realize real-time guiding control of a borehole trajectory.

Description

Technical Field

[0001] This invention relates to the field of drilling engineering technology, and in particular to a real-time wellbore trajectory guidance control system and method for wireless measurement while drilling instruments. Background Technology

[0002] Wireless measurement-while-drilling (MWD) instruments are core equipment in modern directional and horizontal well drilling engineering. They can measure wellbore inclination, azimuth, and toolface attitude parameters in real time during drilling and transmit the data to the surface via wireless methods such as mud pulses. The surface system uses this data to determine the deviation between the current wellbore trajectory and the designed trajectory, formulates a directional control strategy, and then transmits control commands to the downhole directional tools to adjust the drilling direction, ultimately achieving precise control of the wellbore trajectory.

[0003] Existing wellbore trajectory steering control methods typically rely on simplified geological models and fixed control algorithms. The general operational process involves surface engineers manually calculating or generating control commands for the steering tool, such as the tool face angle and drilling pressure for the next step, based on wellbore attitude data uploaded by the measurement-while-drilling instrument and combined with experience or pre-set control logic. These commands are then encoded and transmitted wirelessly to the downhole system. Upon receiving the commands, the downhole steering tool mechanically executes the operation, completing one steering adjustment.

[0004] However, existing technologies have significant shortcomings in dealing with complex downhole environments. First, their control decisions rely primarily on isolated wellbore attitude data, failing to fully integrate information on environmental factors such as downhole vibration and temperature. This results in an incomplete perception of the actual downhole conditions and a lack of targeted control commands. Second, the generation of control commands largely depends on static models or human experience, lacking the ability to adapt to dynamic changes in formation conditions. Furthermore, the transmission quality of downhole wireless channels is affected by multiple factors, and existing communication methods often lack specific compensation processing for control signals, making it difficult to guarantee the reliability of command transmission. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides a real-time wellbore trajectory guidance control system and method for wireless measurement while drilling instruments. It employs multi-source data fusion sensing, intelligent decision-making based on an adaptive learning mechanism, dynamic compensation communication with channel adaptation, and closed-loop feedback updates based on trajectory deviation to achieve real-time wellbore trajectory guidance control.

[0006] The above objectives can be achieved through the following approach:

[0007] A real-time wellbore trajectory guidance control method for a wireless measurement-while-drilling (MWD) instrument includes: acquiring wellbore attitude data and environmental factor data through the wireless MWD instrument, and performing fusion processing to generate fused data; generating adaptive control parameters based on the fused data and historical guidance effect data obtained from a drilling database through an adaptive learning mechanism; performing channel adaptation processing on the adaptive control parameters to generate an initial control signal; dynamically compensating the initial control signal in conjunction with the environmental factor data to generate a compensated control signal; controlling the working attitude of the guidance tool using the compensated control signal, acquiring actual wellbore trajectory data generated by the working attitude, and calculating the deviation between the actual wellbore trajectory data and the preset target wellbore trajectory data to generate trajectory deviation data; and updating the learning parameters of the adaptive learning mechanism based on the trajectory deviation data.

[0008] Optionally, generating fused data includes: performing weighted averaging on the wellbore attitude data to generate preliminary fused data; extracting environmental error information from the environmental factor data; using the environmental error information to correct the preliminary fused data; and generating fused data.

[0009] Optionally, generating adaptive control parameters includes: training a wellbore trajectory prediction model based on historical steering effect data to obtain a trained prediction model; inputting the fused data into the trained prediction model to generate preliminary control parameters; extracting operating condition features from the environmental factor data, optimizing the preliminary control parameters, and generating adaptive control parameters.

[0010] Optionally, optimizing the preliminary control parameters includes: performing a correlation analysis between the operating condition characteristics and the preliminary control parameters to obtain an optimization direction vector; and adjusting the values ​​of the preliminary control parameters according to the optimization direction vector to generate adaptive control parameters.

[0011] Optionally, generating the initial control signal includes: parsing the adaptive control parameters into hardware operation instructions for the guidance tool; performing channel coding on the hardware operation instructions, adding error correction codes, and generating a coded signal; and modulating the coded signal to generate an initial control signal adapted to downhole channel transmission.

[0012] Optionally, generating the compensation control signal includes: quantifying the degree of interference of the environmental factor data on signal transmission and generating an environmental impact index; generating an adjustment function based on the environmental impact index; and applying the adjustment function to the initial control signal to generate a compensation control signal.

[0013] Optionally, the generation of environmental impact indicators includes: extracting vibration data and temperature data from the environmental factor data; performing dimensionless processing on the vibration data and temperature data to obtain normalized environmental data; and performing weighted fusion on the normalized environmental data to generate environmental impact indicators.

[0014] Optionally, generating trajectory deviation data includes: using the compensation control signal to control the guiding tool to adjust the working posture, generating an adjusted working posture; based on the adjusted working posture, collecting actual drilling data through the wireless measurement-while-drilling instrument, extracting the wellbore azimuth and inclination angle, and generating actual wellbore trajectory data; calling preset target wellbore trajectory data, calculating the Euclidean distance between the actual wellbore trajectory data and the target wellbore trajectory data, and generating trajectory deviation data.

[0015] Optionally, updating the learning parameters of the adaptive learning mechanism based on the trajectory deviation data includes: using the trajectory deviation data as error input to calculate the parameter gradient of the adaptive learning mechanism; dynamically adjusting the learning rate based on the magnitude of the trajectory deviation data to generate a dynamic learning rate parameter; and using the dynamic learning rate parameter and the parameter gradient to update the learning parameters of the adaptive learning mechanism.

[0016] Based on the same inventive concept, this invention also provides a real-time wellbore trajectory guidance control system for a wireless measurement-while-drilling instrument. The system includes: a multi-source data fusion module for acquiring wellbore attitude data and environmental factor data through the wireless measurement-while-drilling instrument, and performing fusion processing to generate fused data; an adaptive control decision module for generating adaptive control parameters based on the fused data and historical guidance effect data obtained from a drilling database through an adaptive learning mechanism; a channel adaptation processing module for performing channel adaptation processing on the adaptive control parameters to generate an initial control signal; a dynamic compensation module for dynamically compensating the initial control signal in conjunction with the environmental factor data to generate a compensated control signal; a deviation calculation module for controlling the working attitude of the guidance tool using the compensated control signal, acquiring actual wellbore trajectory data generated by the working attitude, and calculating the deviation between the actual wellbore trajectory data and preset target wellbore trajectory data to generate trajectory deviation data; and a self-learning feedback update module for updating the learning parameters of the adaptive learning mechanism based on the trajectory deviation data.

[0017] Compared with the prior art, the present invention has the following advantages:

[0018] This invention improves the intelligence and accuracy of guidance control by constructing a closed-loop control system integrating data fusion, adaptive decision-making, robust communication, and online learning. The system can proactively integrate information from multiple downhole sensors and correct for environmental errors, providing a data foundation for decision-making. Simultaneously, by combining historical big data with an adaptive learning mechanism based on real-time operating conditions, control commands can adapt to complex and changing formation conditions, avoiding the response lag or inaccuracy problems caused by relying on fixed models in traditional methods.

[0019] This invention enhances the reliability of downhole control command transmission by designing a channel adaptation and dynamic compensation mechanism. Addressing the issue of signal attenuation and distortion in downhole high-temperature and high-vibration environments, such as mud pulse wireless channels, this invention predicts channel quality and performs feedforward compensation before signal transmission, ensuring that critical guidance commands are delivered completely and accurately to downhole tools, thus guaranteeing the stability and real-time performance of the entire closed-loop control link.

[0020] This invention introduces a self-learning feedback update mechanism based on actual trajectory deviations, endowing the system with the ability to continuously self-optimize. The system is no longer a tool that statically executes preset programs, but rather capable of reflecting on and dynamically adjusting its internal decision-making model parameters by comparing the differences between actual drilling results and the target. This online learning capability allows the system to continuously accumulate experience from practice, autonomously improving its guidance performance in unknown or sudden geological environments, and achieving continuous evolution of the control strategy.

[0021] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating the real-time wellbore trajectory guidance control method of a wireless measurement-while-drilling instrument according to an embodiment of the present invention.

[0024] Figure 2 This is a schematic diagram of the adaptive control parameter generation and optimization process according to an embodiment of the present invention.

[0025] Figure 3This is a schematic diagram of the spatial distribution of the three-dimensional wellbore trajectory according to an embodiment of the present invention.

[0026] Figure 4 This is a schematic diagram of the real-time wellbore trajectory guidance control system of the wireless measurement while drilling instrument according to an embodiment of the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0028] Reference Figure 1 One embodiment of the present invention proposes a real-time wellbore trajectory guidance control system and method for wireless measurement while drilling instruments. It adopts multi-source data fusion sensing, intelligent decision-making based on adaptive learning mechanism, dynamic compensation communication with channel adaptation, and closed-loop feedback update based on trajectory deviation, which can realize real-time guidance control of wellbore trajectory.

[0029] The method described in this embodiment specifically includes:

[0030] S1. Wellbore attitude data and environmental factor data are collected through wireless measurement while drilling instruments, and fused to generate fused data;

[0031] Optionally, the generation of fused data includes:

[0032] The wellbore attitude data is weighted and averaged to generate preliminary fused data;

[0033] Environmental error information is extracted from the environmental factor data, and the environmental error information is used to correct the preliminary fused data to generate fused data.

[0034] Specifically, the method first performs a weighted average of the wellbore attitude data to generate preliminary fused data. The wellbore attitude data is acquired at different times or under different operating modes by multiple sensors integrated within the wireless measurement-while-drilling instrument, such as a triaxial magnetometer and a triaxial accelerometer. Due to the differences in accuracy of each sensor and the influence of instantaneous measurement noise, the original wellbore attitude data contains uncertainties. To reduce random errors, a weighted average algorithm is used for preliminary fusion. This process assigns different weights to wellbore attitude data from different measurement sources, such as well inclination angle or wellbore azimuth angle, based on their reliability. The weights are set based on the real-time performance evaluation of the sensors; for example, by analyzing the variance of the sensor output signals. Sensors with smaller signal variances are considered to have more reliable current measurement results and are therefore given higher weights. The resulting preliminary fused data has higher stability and reliability compared to a single data source. After obtaining the preliminary fused data, environmental factor data is further used to correct the data to eliminate measurement errors caused by the harsh downhole environment, ultimately generating the final fused data. The downhole environment is complex and variable. High temperatures can cause thermal drift in sensor elements, while severe vibrations during drilling can introduce dynamic errors into measurements. Environmental factor data is a quantitative description of these physical conditions, acquired by temperature sensors, vibration sensors, and other data collected by wireless measurement-while-drilling (MWD) instruments. The correction process first requires extracting environmental error information from the environmental factor data to characterize the impact of errors. This environmental error information is a quantified correction value, the magnitude of which is directly related to the deviation of specific environmental factors, such as temperature and vibration amplitude. This correction value can be calculated using a pre-calibrated error model in the laboratory, which describes the functional relationship between wellbore attitude measurement deviations and environmental factors. For example, errors caused by temperature can be determined by comparing the real-time temperature with the reference temperature during sensor calibration, while errors caused by vibration are related to vibration intensity. The calculated environmental error information is then subtracted from or compensated for in the initial fused data to complete the correction. The corrected data is the fused data, and its calculation process can be expressed as follows:

[0035] ,

[0036] in, This represents the final fused data; This represents the preliminary fused data obtained after weighted averaging. This represents the environmental error information extracted and calculated from environmental factor data. This correction step eliminates the influence of environmental interference on the measurement results.

[0037] For example, in a drilling measurement, accelerometer A of the wireless measurement-while-drilling instrument measured a well inclination angle of 45.2 degrees, while accelerometer B measured a well inclination angle of 45.4 degrees. The system analyzed the stability of the recent output data of both instruments, determining that accelerometer A had a smaller signal variance and assigning it a weight of 0.7, while accelerometer B had a weight of 0.3. Through weighted averaging, the preliminary fused well inclination angle data was obtained. The temperature was 45.26 degrees Celsius. Simultaneously, the instrument's temperature sensor detected a current downhole temperature of 150 degrees Celsius. The system consulted a preset temperature error model, which indicated that at 150 degrees Celsius, the accelerometer would experience a positive thermal drift error of 0.26 degrees Celsius, representing environmental error information. The degree is 0.26. Finally, the fused data is calculated and generated. The fused data output by this method is not only more accurate but also more robust, providing a more accurate and reliable data foundation for subsequent real-time wellbore trajectory guidance control decisions. This improves the response accuracy of the entire guidance control system and the final drilling operation quality.

[0038] S2. Based on the fused data and historical guidance effect data obtained from the drilling database, adaptive control parameters are generated through an adaptive learning mechanism.

[0039] Optionally, the generation of adaptive control parameters includes:

[0040] Based on historical guidance effect data, a wellbore trajectory prediction model is trained to obtain the trained prediction model.

[0041] The fused data is input into the trained prediction model to generate preliminary control parameters;

[0042] The operating condition characteristics are extracted from the environmental factor data, and the preliminary control parameters are optimized to generate adaptive control parameters.

[0043] Specifically, this method first constructs and trains a wellbore trajectory prediction model based on historical steering effect data through offline learning. Historical steering effect data is a large number of past drilling operation records extracted from drilling databases, containing steering tool control commands and their resulting actual wellbore trajectory changes under specific geological conditions and engineering parameters. The wellbore trajectory prediction model typically employs complex nonlinear models such as deep neural networks, and its training objective is to learn and reproduce the inherent mapping pattern from "control commands" to "trajectory results" contained in historical data. After sufficient training, the model can predict the control strategy required to achieve the expected trajectory target based on given wellbore state information, at which point the trained prediction model is obtained. After obtaining the trained prediction model, the real-time control stage begins. In this stage, fused data is input into the trained prediction model. The fused data is high-precision wellbore attitude data after multi-source information fusion and environmental error correction, accurately representing the current drill bit's real-time position and attitude. The trained prediction model receives this fused data and, combined with preset target wellbore trajectory data, performs forward inference calculations to generate a set of preliminary control parameters. This process can be understood as the model answering the question, "Given the current wellbore condition, what directional operation should be performed to approach the target trajectory most quickly and stably?" Its calculation can be expressed as:

[0044] ,

[0045] in, These represent the initial control parameters generated, such as the tool face angle setting or thrust percentage of the guide tool; This represents the prediction model after training; It is the input real-time fused data; This refers to the preset target wellbore trajectory data. This set of initial control parameters is a theoretically optimal solution derived from historical big data experience. However, real-time downhole operating conditions vary greatly, and relying solely on models based on historical data may not fully address current specific situations. Therefore, this method introduces an optimization step. This step extracts key operating condition features from a parallel environmental factor data stream. Subsequently, these operating condition features are used to dynamically optimize the initial control parameters, generating the final adaptive control parameters.

[0046] For example, the system integrates the well inclination angle of 45.00 degrees from the fused data and the pre-set target well trajectory data at the current depth with an inclination angle of 46.00 degrees, and inputs them into the trained prediction model. After model inference, a set of preliminary control parameters is output, such as a parameter combination {tool face angle: 90 degrees, push force percentage: 60%}. Simultaneously, the system extracts high-frequency axial vibration signals from environmental factor data and, by analyzing their spectrum, extracts the working condition characteristics of high formation hardness. Based on these working condition characteristics, the optimization logic determines that a 90-degree tool face angle may be insufficient for directional drilling in hard formations, and that drilling resistance will also increase. Therefore, the preliminary control parameters are co-optimized: the tool face angle is modified to 95 degrees, and the push force percentage is modified to 70%. This method combines the predictive capabilities of machine learning based on historical big data with the dynamic adaptive capabilities based on real-time environmental perception, achieving intelligent and precise guidance control decisions.

[0047] Optionally, optimizing the preliminary control parameters includes:

[0048] The correlation analysis between the operating condition characteristics and the preliminary control parameters is performed to obtain the optimization direction vector;

[0049] Based on the optimized direction vector, the values ​​of the initial control parameters are adjusted to generate adaptive control parameters.

[0050] Specifically, the process first performs correlation analysis, which deeply correlates the real-time extracted operating condition features with the preliminary control parameters output by the wellbore trajectory prediction model. Operating condition features are quantified real-time environmental information, such as feature values ​​reflecting formation hardness extracted by analyzing the drill string vibration spectrum, or indicators reflecting wellbore stability obtained by monitoring torque change rate. Preliminary control parameters are a set of theoretically guided instructions provided by the model, such as recommended tool face angles and push force magnitudes. The essence of correlation analysis is to establish an evaluation model or knowledge base to determine the potential deviations in actual effects when executing given preliminary control parameters under specific operating conditions. The result of this analysis is quantified into an optimization direction vector. This optimization direction vector is not a single numerical value, but a vector matching the dimension of the preliminary control parameters. Each component indicates the direction and relative magnitude of adjustment for the corresponding control parameter to counteract the adverse effects of the current operating conditions or better utilize favorable conditions. After obtaining the optimization direction vector, the values ​​of the preliminary control parameters are adjusted based on this vector to generate the final adaptive control parameters. This adjustment is a controlled update process implemented through a mathematical operation, which can be expressed as the following formula:

[0051] ,

[0052] in, It is the final generated adaptive control parameter, which is a vector containing all the adjusted guidance instructions; It is the initial control parameter vector that serves as the starting point for optimization, generated by the wellbore trajectory prediction model; It is an optimization direction vector obtained through correlation analysis, and its direction indicates the optimal path for parameter adjustment; This is an optimization step size factor, a dimensionless scalar used to control the magnitude of the adjustment. This factor can be a preset constant or a dynamically adjusted variable based on the severity of the operating conditions or the confidence level of the correlation analysis, ensuring the stability and convergence of the adjustment process and avoiding over-correction. For example... Figure 3 As shown, this process demonstrates how to dynamically optimize initial control parameters based on real-time downhole operating conditions. This vectorized adjustment operation ensures that the synergy and coupling between multiple control parameters are maintained during the optimization process, and the final output adaptive control parameters are the optimal instructions that balance historical data model predictions with real-time operating condition adaptability.

[0053] For example, the initial control parameters output by the wellbore trajectory prediction model are a two-dimensional vector, with a tool face angle of 90 degrees and a push force percentage of 60%. The system extracts the working condition feature as "high-hardness interlayer," and the correlation analysis module determines that this working condition will weaken the directional drilling effect and increase drilling resistance. Therefore, an optimized direction vector is generated. Its tool face angle component is +5 degrees, and its pushing force component is +10%. Simultaneously, based on the significance of this working condition characteristic, the system sets an optimization step size factor. The value is 1.0. Finally, the calculated adaptive control parameters are... In the middle, the tool face angle is adjusted to The degree and the percentage of pushing force are adjusted to This proactive, feedforward optimization can pre-compensate for directional deviations that may be caused by factors such as formation changes and downhole vibrations, improving the success rate and accuracy of control commands in complex and ever-changing downhole environments, thereby making the entire directional control process more robust, efficient, and precise.

[0054] S3. Perform channel adaptation processing on the adaptive control parameters to generate an initial control signal;

[0055] Optionally, generating the initial control signal includes:

[0056] The adaptive control parameters are parsed into hardware operation instructions for the guiding tool;

[0057] The hardware operation instructions are channel encoded, error correction codes are added, and encoded signals are generated.

[0058] The encoded signal is modulated to generate an initial control signal suitable for underground channel transmission.

[0059] Specifically, the adaptive control parameters are first parsed into hardware operation instructions for the guidance tool. These adaptive control parameters are numerical, high-level instructions generated by the adaptive control decision module. These parameters need to be translated into low-level instructions that the internal actuators of the guidance tool can directly recognize and execute. The parsing process is precisely this translation process. By consulting a preset mapping table or executing a transformation function, it converts the numerical adaptive control parameters into a series of specific, binary hardware operation instructions, such as controlling a stepper motor to rotate a specific number of steps or controlling a solenoid valve to open for a specific duration. Based on this, to ensure the integrity of these critical hardware operation instructions transmitted in noisy downhole channels, channel coding is required, along with the addition of error correction codes, to generate the encoded signal. Downhole channels, especially mud pulse channels, are highly susceptible to interference from drill string vibration, pump noise, and signal attenuation, leading to bit errors in transmitted data. Channel coding introduces redundant information, enabling the receiver not only to detect errors but also, to a certain extent, to correct them. This method employs error-correcting codes, such as Reed-Solomon codes, which possess strong resistance to sudden errors. The original hardware operation instruction sequence is taken as input, processed by an encoding algorithm, and then appended with a checksum, forming a data sequence longer than the original instruction—the encoded signal. Finally, the encoded signal is modulated to generate an initial control signal suitable for downhole channel transmission. The encoded signal is essentially still a digitized binary sequence and cannot propagate directly in the physical medium. Modulation converts these binary "0" and "1" symbols into physical waveforms suitable for propagation in downhole channels, such as drilling fluid. In wireless measurement-while-drilling systems, a common modulation method is mud pulse modulation, such as pulse position modulation (PPM) or on / off keying (OOK). This process maps the "1" and "0" in the encoded signal to the generation or non-generation of a positive pressure pulse in the drilling fluid. These sequentially generated pressure pulses propagate in the drilling fluid, forming a physical signal that can be received and demodulated by downhole directional tools; this signal is the initial control signal.

[0060] For example, the system parses the two adaptive control parameters—a 95-degree tool face angle and a 70% push force percentage—into a 16-bit binary hardware operation instruction. Subsequently, this 16-bit instruction is channel-coded using Reed-Solomon error correction code, with an additional 8-bit checksum, generating a 24-bit encoded signal. Finally, using on-off keying modulation, each "1" in this 24-bit encoded signal sequence is mapped to a 0.5-second positive pressure pulse with an amplitude of 10 MPa, while each "0" corresponds to a 0.5-second pulse-free period. The pressure states for these 24 periods are sequentially generated by a pulse generator, forming the initial control signal propagating in the drilling fluid. This method, through a complete and rigorous signal processing chain, solves the core problem of reliably transmitting upper-level intelligent decision-making results to downhole actuators, ensuring smooth and high-fidelity information flow between control decisions and physical execution. This is the fundamental guarantee for the stable operation of a closed-loop real-time guidance control system.

[0061] S4. Dynamically compensate the initial control signal by combining the environmental factor data to generate a compensated control signal;

[0062] Optionally, the generation of the compensation control signal includes:

[0063] The degree of interference of the environmental factor data on signal transmission is quantified to generate environmental impact indicators;

[0064] Based on the aforementioned environmental impact indicators, an adjustment function is generated;

[0065] The adjustment function is applied to the initial control signal to generate a compensation control signal.

[0066] Specifically, this method first quantifies real-time environmental factor data to assess its potential interference with signal transmission and generates a comprehensive environmental impact index. The environmental factor data includes information such as downhole temperature, pressure, and multidimensional vibrations generated by drill string movement. When signals are transmitted in downhole channels, such as mud pulse channels, they are affected by the combined effects of these factors. For example, high temperatures change mud viscosity, affecting signal attenuation, while severe vibrations generate noise at frequencies close to the signal, reducing the signal-to-noise ratio. The generation process of the environmental impact index involves comprehensively calculating these multi-source, heterogeneous environmental factor data into a single, dimensionless value. After obtaining the environmental impact index, an adjustment function is dynamically generated based on it. This adjustment function is essentially a mapping from the environmental impact index to a signal adjustment strategy. It clarifies the degree to which certain physical properties of the initial control signal should be adjusted when the environmental impact index is at a certain value. For example, the adjustment function can be defined as linearly or non-linearly increasing the signal transmission power or pulse width as the environmental impact index increases. Increasing the transmit power directly boosts the signal energy to combat channel attenuation and noise; while increasing the pulse width extends the signal in the time domain, making it easier to detect from noise at the receiver through integration. Finally, this method applies the generated adjustment function to the initial control signal generated by the channel adaptation processing module, thereby generating a compensation control signal. This application process does not change the control command content carried by the signal, but rather adjusts its physical waveform characteristics. For example, if the adjustment function specifies energy enhancement, then when generating mud pulses, the opening amplitude or opening time of the pulse valve will be increased to produce a pressure wave with a higher amplitude or wider width. This enhanced signal is the compensation control signal. Its generation process can be described as follows:

[0067] ,

[0068] in, Key energy-related parameters representing the compensation control signal, such as signal amplitude; This represents the original parameter value corresponding to the initial control signal; These are environmental impact indicators calculated in real time. This is an adjustment function generated based on environmental impact indicators. Here, it manifests as a gain function, and its output value is typically greater than or equal to 1. Through this operation, the compensation control signal carries the exact same instruction information as the initial control signal, but its physical form has been optimized, giving it stronger anti-interference capabilities.

[0069] For example, the system calculates the current environmental impact index as 0.8 based on real-time environmental factor data. This is a relatively high value, indicating poor channel conditions. The system generates an adjustment function based on this index, which specifies that when the environmental impact index is 0.8, the output value of the gain function G(0.8) is 1.5. The system applies this gain to the amplitude parameter of the initial control signal. If the mud pulse amplitude of the initial control signal... The original setting was 10 MPa. After adjustment, the pulse amplitude of the compensation control signal was... The amplitude is increased to 10 MPa multiplied by 1.5, which is 15 MPa. This amplified signal is the final compensated control signal sent. This feedforward control method can overcome the time-varying and uncertain characteristics of the downhole channel, ensuring that even under the worst operating conditions, the sent control signal can reach the downhole steering tool with sufficient energy and clarity. This guarantees the stability of the control link and the real-time transmission of commands, providing communication assurance for the operation of the entire wellbore trajectory real-time closed-loop control system.

[0070] Optionally, the generated environmental impact indicators include:

[0071] Vibration and temperature data are extracted from the environmental factor data;

[0072] The vibration data and temperature data are dimensionless to obtain normalized environmental data;

[0073] The normalized environmental data are weighted and fused to generate environmental impact indicators.

[0074] Specifically, the process first selectively extracts vibration and temperature data, which have the most significant impact on the transmission quality of downlink control signals, from multi-source environmental factor data. Vibration data, typically collected by the accelerometer built into the wireless measurement-while-drilling instrument, reflects the intensity and frequency characteristics of mechanical vibrations generated by the drill string during drilling due to collisions with the wellbore and interaction with the formation, and is the main source of downhole physical noise. Temperature data is measured by temperature sensors, which directly affects the viscosity and other fluid properties of the drilling fluid, thereby altering the attenuation characteristics of the mud pulse signal during propagation. After extracting the vibration and temperature data, to address the problem of these two types of data having different physical dimensions and vastly different numerical ranges, making direct comparison or fusion impossible, they must be dimensionless processed to generate normalized environmental data. Dimensionless processing employs a standardized mathematical transformation, such as the max-min normalization method, to map the original physical measurement values ​​to a unified, preset interval, typically from 0 to 1. This process can be expressed by the following formula:

[0075] ,

[0076] in, It is the normalized environmental data obtained through calculation; it is a dimensionless numerical value. Represents the raw vibration or temperature data currently being collected in real time; and These are the minimum and maximum values ​​that are preset in the instrument design or historical experience for this type of data. Through this transformation, both the acceleration value representing vibration intensity and the Celsius value representing ambient temperature are converted into a relative degree value between 0 and 1, making them comparable and additive. Finally, the multiple normalized environmental data obtained after dimensionless processing are weighted and fused to generate the final environmental impact index. Since the degree of influence of vibration and temperature on signal transmission may not be the same, weighted fusion allows assigning different importance to them based on prior knowledge or experimental calibration results. This process involves multiplying each normalized environmental data point by its corresponding weight coefficient and then summing them. The calculation formula is:

[0077] ,

[0078] in, These are the final environmental impact indicators; and These are the normalized vibration data and normalized temperature data after dimensionless processing, respectively. and These are preset weighting coefficients, representing the contributions of vibration and temperature to signal transmission interference, respectively, and their sum is usually set to 1. This fused environmental impact index comprehensively quantifies the overall adverse impact of the current downhole environment on signal transmission with a single value.

[0079] For example, the system collects real-time vibration data of 3g and temperature data of 150 degrees Celsius. The preset range for vibration data is known to be 0g to 5g, and the range for temperature data is 20 degrees Celsius to 200 degrees Celsius. The system first performs dimensionless processing on both to obtain normalized vibration data. The value is 0.6, representing the normalized temperature data. The value is approximately 0.72. Based on prior knowledge, vibration has a greater impact on mud pulse signals; therefore, a vibration weight is set. The temperature weight is 0.7. The value is 0.3. Finally, environmental impact indicators are generated through weighted fusion calculation. The environmental impact indicators obtained by this method provide accurate and reliable quantitative basis for subsequent signal compensation strategies, enabling the system to adjust signal transmission parameters in a targeted manner, thereby achieving intelligent perception and proactive adaptation to channel changes.

[0080] S5. Use the compensation control signal to control the working posture of the guiding tool, obtain the actual wellbore trajectory data generated by the working posture, calculate the deviation between the actual wellbore trajectory data and the preset target wellbore trajectory data, and generate trajectory deviation data.

[0081] Optionally, the generated trajectory deviation data includes:

[0082] The compensation control signal is used to control the guiding tool to adjust the working posture, thereby generating the adjusted working posture;

[0083] Based on the adjusted working posture, the actual drilling data is collected by the wireless measurement-while-drilling instrument, the wellbore azimuth and inclination angle are extracted, and the actual wellbore trajectory data is generated.

[0084] The preset target wellbore trajectory data is called, the Euclidean distance between the actual wellbore trajectory data and the target wellbore trajectory data is calculated, and trajectory deviation data is generated.

[0085] Specifically, this method first utilizes a dynamically compensated control signal to control the downhole steerable tool to adjust its operating posture. The compensation control signal is a robust signal adapted for channel and environmental compensation. It drives the actuators of the steerable tool, such as hydraulic pushers or variable-diameter stabilizers, changing their state relative to the drill string axis, thereby generating an adjusted operating posture. This adjusted operating posture directly determines the force state and direction of the drill bit during the next short drilling segment. During drilling with the steerable tool in the adjusted operating posture, real-time drilling data is continuously acquired via wireless measurement-while-drilling (MWD) instruments. This real-time drilling data is a direct measurement result of the wellbore trajectory formation process. From this rich real-time drilling data, key parameters that accurately describe the spatial orientation of the wellbore are extracted: the wellbore azimuth and inclination angle. The inclination angle defines the angle between the wellbore trajectory and the direction of gravity, while the wellbore azimuth defines the projection direction of the wellbore onto the horizontal plane. These two angles together constitute a set of spherical coordinates, which can uniquely determine the tangential direction of the wellbore in three-dimensional space. By extracting these two angle values ​​at each measurement point and combining them with the known drilling depth, a new segment of actual wellbore trajectory data can be reconstructed. Finally, the preset target wellbore trajectory data, pre-stored in the drilling computer, is retrieved and compared with the newly generated actual wellbore trajectory data to calculate the deviation, thus generating trajectory deviation data. The preset target wellbore trajectory data is an ideal drilling path planned by geological steering experts based on reservoir models and engineering requirements. Figure 3 As shown in the figure, this diagram visually illustrates the spatial deviation between the actual wellbore trajectory and the target wellbore trajectory. The deviation is calculated using Euclidean distance as a metric, which comprehensively evaluates the overall distance of the actual trajectory point from the corresponding target point in three-dimensional space. The calculation formula is as follows:

[0086] ,

[0087] in, The generated trajectory deviation data represents the magnitude of the deviation at a certain depth point; It is the three-dimensional rectangular coordinate of the actual trajectory point obtained by converting the well inclination angle, well azimuth angle and depth from the actual well trajectory data; These are the coordinates of the corresponding depth point in the preset target wellbore trajectory data. This calculated trajectory deviation data is a scalar that intuitively quantifies the degree to which the current drilling results conform to the target, and is the ultimate manifestation of the overall control effect.

[0088] For example, the directional tool drills a distance under the control of a compensation control signal. The wireless measurement-while-drilling instrument collects actual drilling data at the new measuring point, extracting the actual wellbore azimuth angle as 60 degrees and the actual well inclination angle as 46.2 degrees. Based on these angles and the current depth, the system calculates the three-dimensional coordinates of the actual wellbore trajectory data at that measuring point as (100.5, 174.2, 2000.0). Subsequently, the system calls the preset target wellbore trajectory data and finds that at the same measuring depth, the target coordinates should be (100.0, 175.0, 2000.0). Finally, the system calculates the Euclidean distance between these two coordinate points, obtaining a trajectory deviation data D of approximately 0.94 meters. This method provides an error signal for the system's self-learning feedback update module, driving the entire adaptive learning mechanism to continuously optimize its internal model and parameters, thereby achieving continuous improvement in wellbore trajectory control accuracy and autonomous adaptation to unknown geological conditions.

[0089] S6. Update the learning parameters of the adaptive learning mechanism based on the trajectory deviation data.

[0090] Optionally, updating the learning parameters of the adaptive learning mechanism based on the trajectory deviation data includes:

[0091] The trajectory deviation data is used as error input to calculate the parameter gradient of the adaptive learning mechanism;

[0092] Based on the magnitude of the trajectory deviation data, the learning rate is dynamically adjusted to generate dynamic learning rate parameters.

[0093] The learning parameters of the adaptive learning mechanism are updated using the dynamic learning rate parameter and the parameter gradient.

[0094] Specifically, this method first uses trajectory deviation data as a quantized error input to calculate the parameter gradient of the internal model of the adaptive learning mechanism with respect to its learning parameters. An adaptive learning mechanism, such as a well trajectory prediction model, contains a large number of learning parameters, such as the weights and biases of a neural network. The parameter gradient is obtained by backpropagating the error represented by the trajectory deviation data within the model using algorithms such as backpropagation. It reveals the degree and direction of contribution of each learning parameter to the final generated trajectory deviation data. In other words, the parameter gradient indicates the fastest direction in which each learning parameter should be adjusted to reduce the current deviation. Simultaneously with calculating the parameter gradient, the learning rate is dynamically adjusted based on the magnitude of the current trajectory deviation data to generate a dynamic learning rate parameter. The learning rate is a key hyperparameter controlling the step size of parameter updates. A fixed learning rate may converge too slowly when the deviation is large, or cause oscillations and make convergence difficult when the deviation is small. This method establishes a functional relationship: when the trajectory deviation data is large, a larger learning rate is used to accelerate convergence and quickly correct significant errors; conversely, when the trajectory deviation data is small, it automatically switches to a smaller learning rate for fine-tuning, ensuring the stability of the learning process and the final convergence accuracy. This learning rate, dynamically generated based on real-time error, is the dynamic learning rate parameter. Finally, the calculated parameter gradient and the generated dynamic learning rate parameter are used together to update the learning parameters of the adaptive learning mechanism. This update process follows the basic principles of gradient descent or its variants, and its core mathematical expression can be represented as:

[0095] ,

[0096] in, This represents the updated learning parameters; Represents the current learning parameters; It is a dynamic learning rate parameter determined by the magnitude of the trajectory deviation data, and is a dimensionless scalar. It is an error function quantized from trajectory deviation data; That is, the error function with respect to the learning parameters The calculated gradient of the parameters has the same dimension as the learned parameters. The same applies. The physical meaning of this formula is that the current learning parameters are moved a small step along the direction that minimizes error most quickly—the negative gradient direction. The step size is determined by both the dynamic learning rate and the magnitude of the gradient itself. By continuously iterating this update process, the internal model of the adaptive learning mechanism becomes increasingly accurate.

[0097] For example, the system uses a trajectory deviation of 0.94 meters as the error input. Due to the large deviation, the system dynamically generates a large dynamic learning rate parameter with a value of 0.01. Simultaneously, using the backpropagation algorithm, the gradient of a key learning parameter in the wellbore trajectory prediction model is calculated to be +2.5. The system updates this learning parameter using gradient descent; the new parameter value equals the old value minus 0.01 multiplied by 2.5. This update process is performed synchronously on all learning parameters in the model, ensuring that the model, in the next prediction under similar operating conditions, tends to generate a control command that results in a smaller deviation. This closed-loop self-learning feedback update mechanism gives the entire control system excellent adaptability, enabling it to autonomously adapt to unforeseen geological changes encountered during drilling and continuously optimize its guidance strategy, ultimately achieving continuous improvement in the accuracy and robustness of wellbore trajectory control.

[0098] Based on the same inventive concept, such as Figure 4 As shown, the present invention also provides a real-time wellbore trajectory guidance control system for a wireless measurement-while-drilling instrument, the system comprising:

[0099] The multi-source data fusion module is used to collect wellbore attitude data and environmental factor data through wireless measurement while drilling instruments, and to perform fusion processing to generate fused data;

[0100] An adaptive control decision module is used to generate adaptive control parameters based on the fused data and historical guidance effect data obtained from the drilling database through an adaptive learning mechanism.

[0101] The channel adaptation processing module is used to perform channel adaptation processing on the adaptive control parameters and generate an initial control signal;

[0102] The dynamic compensation module is used to dynamically compensate the initial control signal by combining the environmental factor data to generate a compensated control signal;

[0103] The deviation calculation module is used to control the working posture of the guiding tool using the compensation control signal, obtain the actual wellbore trajectory data generated by the working posture, calculate the deviation between the actual wellbore trajectory data and the preset target wellbore trajectory data, and generate trajectory deviation data.

[0104] The self-learning feedback update module is used to update the learning parameters of the adaptive learning mechanism based on the trajectory deviation data.

[0105] It should be noted that the electrical connections between the various units described above do not necessarily represent direct or indirect connections. Any indirect connection method can be applied to the embodiments of the present invention as long as it achieves the purpose of the present invention. The above descriptions are merely exemplary embodiments of the present invention and should not be construed as limiting the scope of the present invention.

[0106] All equivalent changes and modifications made in accordance with the teachings of this invention are still within the scope of this invention. Those skilled in the art will readily conceive of other embodiments of this invention upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this invention that follow the general principles of this invention and include common knowledge or conventional techniques in the art not described herein.

Claims

1. A real-time wellbore trajectory guidance control method for wireless measurement-while-drilling instruments, characterized in that, The method includes: Wellbore attitude data and environmental factor data are acquired through wireless measurement-while-drilling instruments and fused to generate fused data, including: The wellbore attitude data is weighted and averaged to generate preliminary fused data; Environmental error information is extracted from the environmental factor data, and the environmental error information is used to correct the preliminary fused data to generate fused data; Based on the fused data and historical steering effect data obtained from the drilling database, adaptive control parameters are generated through an adaptive learning mechanism, including: Based on historical guidance effect data, a wellbore trajectory prediction model is trained to obtain the trained prediction model. The fused data is input into the trained prediction model to generate preliminary control parameters; Extract the operating condition characteristics from the environmental factor data, optimize the preliminary control parameters, and generate adaptive control parameters; The adaptive control parameters are subjected to channel adaptation processing to generate an initial control signal, including: The adaptive control parameters are parsed into hardware operation instructions for the guiding tool; The hardware operation instructions are channel encoded, error correction codes are added, and encoded signals are generated. The encoded signal is modulated to generate an initial control signal adapted for transmission in the underground channel; The initial control signal is dynamically compensated based on the environmental factor data to generate a compensated control signal, including: The degree of interference of the environmental factor data on signal transmission is quantified to generate environmental impact indicators; Based on the aforementioned environmental impact indicators, an adjustment function is generated; The adjustment function is applied to the initial control signal to generate a compensation control signal; The working attitude of the directional tool is controlled by the compensation control signal, the actual wellbore trajectory data generated by the working attitude is acquired, and the deviation between the actual wellbore trajectory data and the preset target wellbore trajectory data is calculated to generate trajectory deviation data, including: The compensation control signal is used to control the guiding tool to adjust the working posture, thereby generating the adjusted working posture; Based on the adjusted working posture, the actual drilling data is collected by the wireless measurement-while-drilling instrument, the wellbore azimuth and inclination angle are extracted, and the actual wellbore trajectory data is generated. Call the preset target wellbore trajectory data, calculate the Euclidean distance between the actual wellbore trajectory data and the target wellbore trajectory data, and generate trajectory deviation data; The learning parameters of the adaptive learning mechanism are updated based on the trajectory deviation data.

2. The real-time wellbore trajectory guidance control method for wireless measurement-while-drilling instruments according to claim 1, characterized in that, The optimization of the preliminary control parameters includes: The correlation analysis between the operating condition characteristics and the preliminary control parameters is performed to obtain the optimization direction vector; Based on the optimized direction vector, the values ​​of the initial control parameters are adjusted to generate adaptive control parameters.

3. The real-time wellbore trajectory guidance control method for wireless measurement-while-drilling instruments according to claim 1, characterized in that, The environmental impact indicators generated include: Vibration and temperature data are extracted from the environmental factor data; The vibration data and temperature data are dimensionless to obtain normalized environmental data; The normalized environmental data are weighted and fused to generate environmental impact indicators.

4. The real-time wellbore trajectory guidance control method for wireless measurement-while-drilling instruments according to claim 1, characterized in that, The process of updating the learning parameters of the adaptive learning mechanism based on the trajectory deviation data includes: The trajectory deviation data is used as error input to calculate the parameter gradient of the adaptive learning mechanism; Based on the magnitude of the trajectory deviation data, the learning rate is dynamically adjusted to generate dynamic learning rate parameters. The learning parameters of the adaptive learning mechanism are updated using the dynamic learning rate parameter and the parameter gradient.

5. A real-time wellbore trajectory guidance control system for a wireless measurement-while-drilling instrument, applied to the real-time wellbore trajectory guidance control method for a wireless measurement-while-drilling instrument as described in any one of claims 1-4, characterized in that, The system includes: The multi-source data fusion module is used to collect wellbore attitude data and environmental factor data through wireless measurement while drilling instruments, and to perform fusion processing to generate fused data; An adaptive control decision module is used to generate adaptive control parameters based on the fused data and historical guidance effect data obtained from the drilling database through an adaptive learning mechanism. The channel adaptation processing module is used to perform channel adaptation processing on the adaptive control parameters and generate an initial control signal; The dynamic compensation module is used to dynamically compensate the initial control signal by combining the environmental factor data to generate a compensated control signal; The deviation calculation module is used to control the working posture of the guiding tool using the compensation control signal, obtain the actual wellbore trajectory data generated by the working posture, calculate the deviation between the actual wellbore trajectory data and the preset target wellbore trajectory data, and generate trajectory deviation data. The self-learning feedback update module is used to update the learning parameters of the adaptive learning mechanism based on the trajectory deviation data.

Images