Control method and device for push-the-bit rotary steerable system

Abstract

The application discloses a control method and device for a push-the-bit rotary steerable system, and the method comprises the following steps: inputting formation parameter data into a multi-dimensional data matching model to obtain predicted rib plate adjustment parameters; searching in a constructed formation database to obtain rib plate adjustment parameters matched with the formation characteristics corresponding to the formation parameter data; when the similarity between the predicted rib plate adjustment parameters and the searched rib plate adjustment parameters is greater than a similarity threshold, generating control instructions for each rib plate of the push-the-bit rotary steerable system according to the predicted rib plate adjustment parameters, a design trajectory and an actual trajectory of a well hole in which drilling operation is performed; and sending the control instructions for each rib plate to a driving device of the push-the-bit rotary steerable system to drive the driving device to drive the movement of each rib plate. The application can deeply integrate formation characteristics, realize self-adaptive adjustment of rib plates and keep the resultant force stable, thereby improving the efficiency and quality of drilling operation and reducing operation cost and risk.

Description

Technical Field

[0001] This invention relates to the field of oil drilling technology, and in particular to a control method and device for a push-type rotary steering system. Background Technology

[0002] As shallow, easily exploitable oil resources are continuously depleted, oil exploration and development are gradually expanding into deeper, ultra-deeper formations and areas with complex geological structures. In this process, push-type rotary steerable systems, with their unique advantages in precise wellbore trajectory control, have become an indispensable core equipment in oil drilling operations, and their performance directly affects the success and efficiency of drilling operations.

[0003] The working mechanism of the push-type rotary steerable system resembles a sophisticated mechanical symphony. Centered on the extension and retraction of the ribs and thrust adjustment, it applies a specific directional force to the drill bit during drilling through a carefully designed mechanical structure and control system. Like an invisible hand, it guides the drill bit along a pre-planned trajectory through the underground rock formations, ensuring the wellbore trajectory accurately hits the target oil layer. Under ideal drilling conditions, this system can effectively achieve initial control of the wellbore trajectory, laying the foundation for oil extraction.

[0004] However, the stratigraphic environment is highly complex. From a geological structural perspective, strata have undergone long geological epochs of change, influenced by various geological forces such as plate tectonics and sedimentation, resulting in intricate structures. Key parameters of strata, such as lithology, hardness, and dip angle, exhibit significant spatial differences in different regions.

[0005] In terms of lithology, the underground strata encompass a variety of rock types, including sandstone, shale, limestone, and conglomerate. Sandstone has a granular structure with high porosity and permeability, but its cementation varies considerably, potentially leading to wellbore collapse during drilling. Shale has a fine texture and well-developed bedding, making it prone to swelling and spalling when exposed to water, placing stringent requirements on drilling fluid performance and tool stability. Limestone is primarily composed of calcium carbonate, which is easily dissolved in acidic environments, altering wellbore shape and formation mechanical properties. The vast differences in compressive strength, elastic modulus, and other mechanical properties among different rock types necessitate that push-type rotary steerable systems possess exceptional adaptability when facing diverse lithologies; otherwise, severe problems such as accelerated tool wear, reduced drilling efficiency, and even stuck drill bits can easily occur.

[0006] Variations in formation hardness also pose a significant challenge to drilling operations. Formation hardness ranges extremely widely, from soft, easily drillable mudstone to extremely hard granite, which is as hard as steel – differences in hardness can be tens or even hundreds of times. When drilling through soft mudstone, the system is prone to problems such as drill bit slippage and loss of drilling direction control; while in hard granite formations, the system bears enormous loads, power consumption increases dramatically, and drill bit life is significantly shortened. To maintain stable and efficient drilling performance in formations of varying hardness, push-type rotary steerable systems need to adjust operating parameters and steerable strategies in real time according to formation hardness, which places extremely high demands on their control systems and mechanical structures.

[0007] Variations in formation dip are also a key factor affecting drilling operations. Formation dip ranges greatly from nearly horizontal, gently sloping formations to steep, highly dipping formations. In gently sloping formations, while drilling is relatively smooth, precise control of the tool's guidance is still necessary to avoid wellbore trajectory deviations. In highly dipping formations, the system must not only overcome the effects of gravity but also cope with the variations in drilling resistance caused by formation anisotropy. Even slight errors can lead to wellbore trajectory deviations from the designed path, preventing accurate drilling to the target oil layer.

[0008] Currently, the traditional push-type rotary steerable systems widely used in the oil drilling industry mostly rely on fixed control parameters and simple feedback mechanisms. These traditional methods are often designed based on idealized formation models and assumptions, lacking sufficient consideration of the complexity of actual formations. When faced with the complex and variable lithology, hardness, and dip angles of real formations, traditional control methods struggle to achieve rapid and accurate responses. The system cannot adjust the extension and retraction of the ribs and the magnitude of the thrust in a timely manner according to formation changes, leading to a decrease in wellbore trajectory control accuracy and failing to meet the stringent trajectory accuracy requirements of modern oil drilling.

[0009] In actual drilling operations, wellbore dogleg accuracy is a key indicator for measuring drilling quality and the degree of alignment between the drilling trajectory and the target oil layer. To ensure that the wellbore trajectory accurately penetrates the target oil layer and achieves efficient production, the ribs of the push-type rotary steerable system must be dynamically adjusted to control the degree of wellbore curvature. However, traditional rib adjustment methods have fundamental flaws. In the process of achieving independent adjustment of the three ribs, the lack of an effective coordination and control mechanism makes it impossible to maintain a constant resultant force. When the rib opening angle is too large, the contact area between the system and the wellbore increases significantly, leading to a sharp rise in frictional resistance. This not only directly causes a significant decrease in drilling speed, severely affecting the progress of drilling operations, but also increases the wear of the drill pipe and drill bit, shortening their service life.

[0010] According to extensive statistics from actual engineering projects, in certain complex formations, the drilling efficiency of traditional push-type rotary steerable systems can decrease by 40%-60% compared to expectations. This significant drop in drilling efficiency directly leads to a substantial increase in drilling cycles. Extended drilling cycles mean a significant increase in direct costs such as labor and equipment rental costs. Simultaneously, prolonged operation increases equipment wear and tear, raises the probability of equipment failure, and may even trigger downhole accidents such as lost circulation, well collapse, and stuck pipe. These potential risks not only further increase operating costs but also threaten the lives of oil drilling personnel, severely restricting the development efficiency and economic benefits of the oil drilling industry.

[0011] In conclusion, as oil exploration and development advances into deeper and more complex formations, developing a control method that can deeply integrate with formation characteristics, achieve adaptive adjustment of the ribs, and maintain stable combined forces has become crucial for overcoming technological bottlenecks and enhancing the industry's competitiveness in oil drilling. This technological breakthrough will not only improve the efficiency and quality of drilling operations and reduce operating costs and risks, but will also provide strong technical support for the efficient exploration and development of oil resources, thus driving the development of the oil drilling industry. Summary of the Invention

[0012] In a first aspect, embodiments of the present invention provide a control method for a push-type rotary steering system, which can deeply integrate formation characteristics, achieve adaptive adjustment of the ribs and maintain stable resultant force, improve the efficiency and quality of drilling operations, and reduce operating costs and risks. The method includes:

[0013] When the push-type rotary steering system is performing drilling operations, it acquires formation parameter data.

[0014] The stratigraphic parameter data is input into a multi-dimensional data matching model to obtain the predicted rib adjustment parameters. The multi-dimensional data matching model is obtained by training a neural network based on historical stratigraphic parameter data under different stratigraphic conditions.

[0015] Search the constructed stratigraphic database to obtain the rib adjustment parameters that match the stratigraphic features corresponding to the stratigraphic parameter data;

[0016] When the similarity between the predicted rib adjustment parameters and the retrieved rib adjustment parameters is greater than the similarity threshold, control commands for each rib of the push-type rotary steering system are generated based on the predicted rib adjustment parameters, the design trajectory of the wellbore performing the drilling operation, and the actual trajectory.

[0017] Control commands for each rib are sent to the drive unit of the push-type rotary guide system, so that the drive unit drives the movement of each rib.

[0018] Secondly, embodiments of the present invention also provide a control device for a push-type rotary steering system, which can deeply integrate formation characteristics, achieve adaptive adjustment of the ribs and maintain stable resultant force, improve the efficiency and quality of drilling operations, and reduce operating costs and risks. The device includes:

[0019] The data acquisition module is used to acquire formation parameter data when the push-type rotary steering system is performing drilling operations.

[0020] The rib adjustment parameter prediction module is used to input formation parameter data into a multi-dimensional data matching model to obtain the predicted rib adjustment parameters. The multi-dimensional data matching model is obtained by training a neural network based on historical formation parameter data under different formation conditions.

[0021] The retrieval module is used to search the constructed stratigraphic database to obtain rib adjustment parameters that match the stratigraphic features corresponding to the stratigraphic parameter data.

[0022] The control command generation module is used to generate control commands for each rib of the push-type rotary steering system when the similarity between the predicted rib adjustment parameters and the retrieved rib adjustment parameters is greater than the similarity threshold, based on the predicted rib adjustment parameters, the design trajectory of the wellbore performing the drilling operation, and the actual trajectory.

[0023] The control command sending module is used to send control commands for each rib to the drive device of the push-type rotary guide system, so that the drive device drives the movement of each rib.

[0024] Thirdly, embodiments of the present invention also provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described control method for a push-type rotary guide system.

[0025] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described control method for a push-type rotary guide system.

[0026] Fifthly, embodiments of the present invention also provide a computer program product, the computer program product including a computer program, which, when executed by a processor, implements the above-described control method for a push-type rotary guide system.

[0027] In this embodiment of the invention, a multi-dimensional data matching model is trained based on a large amount of historical formation parameter data, which can quickly output predicted rib adjustment parameters to provide a preliminary reference for drilling operations. The formation database stores the optimal adjustment parameters that have been verified in practice and match formation characteristics. Combining these two methods and generating control commands when the similarity meets the standard utilizes both the model's rapid prediction capability and the database's practical experience data for verification, reducing the potential biases that may arise from single predictions or retrievals, and making the adjustment parameters more closely aligned with actual drilling needs. Different formation conditions vary greatly, making it difficult for traditional methods to adapt quickly and accurately. In this technology, the multi-dimensional data matching model can learn the parameter mapping relationships under different formation conditions, and the formation database also covers diverse formation characteristics and corresponding adjustment parameters. When encountering complex or special formations, the system can find suitable rib adjustment parameters through a dual approach of model prediction and database retrieval, enabling the push-type rotary steering system to adapt well to different formation environments and ensuring the stability of drilling operations. The technical process combines the wellbore's designed trajectory and actual trajectory to generate control commands, allowing for targeted adjustments to the movement of each rib based on trajectory deviations. This process fully utilizes predicted and retrieved reliable adjustment parameters to ensure precise rib movement, effectively reducing the deviation between the actual and designed trajectories, improving wellbore trajectory control accuracy, and meeting the stringent trajectory requirements of drilling operations. The multi-dimensional data matching model can quickly output predicted parameters, and formation database retrieval can efficiently obtain matching parameters; their synergistic application shortens parameter determination time. Simultaneously, precise control commands make rib adjustment more efficient, reducing the number of adjustments and drilling downtime caused by inappropriate parameters, thereby improving overall drilling efficiency. From formation parameter acquisition, adjustment parameter prediction and retrieval, to control command generation and transmission, the entire process forms a coherent automated workflow, reducing manual intervention. The application of the multi-dimensional data matching model demonstrates the system's intelligent learning and predictive capabilities, enabling the push-type rotary steering system to autonomously adapt to different formation conditions, improving the system's intelligence level and degree of automation. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0029] Figure 1 This is a flowchart of a control method for a push-type rotary guide system in an embodiment of the present invention;

[0030] Figure 2This is a flowchart illustrating the control process for the push-type rotary guide system in an embodiment of the present invention.

[0031] Figure 3 This is a schematic diagram of the control device for a push-type rotary guide system in an embodiment of the present invention;

[0032] Figure 4 This is another structural schematic diagram of the control device for the push-type rotary guide system in an embodiment of the present invention;

[0033] Figure 5 This is a schematic diagram of a computer device in an embodiment of the present invention. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Here, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0035] The proposed solution in this invention aims to develop an innovative adaptive control method for a push-type rotary steering system. This method automatically and dynamically adjusts the working state of the three ribs by real-time and precise detection of formation parameters. While ensuring that the resultant force acting on the wellbore remains constant, it precisely meets the stringent design requirements for wellbore trajectory dogleg accuracy. Simultaneously, by optimizing the separation of the ribs, it effectively reduces resistance during drilling, significantly increases drilling speed, and thus comprehensively improves the overall efficiency of oil drilling operations, substantially reduces production costs, and provides a more competitive technical solution for the oil and gas exploration and development industry.

[0036] Figure 1 This is a flowchart of a control method for a push-type rotary guide system according to an embodiment of the present invention. The method includes:

[0037] Step 101: When the push-type rotary steering system is performing drilling operations, the collected formation parameter data is obtained;

[0038] Step 102: Input the formation parameter data into the multi-dimensional data matching model to obtain the predicted rib adjustment parameters. The multi-dimensional data matching model is obtained by training a neural network based on historical formation parameter data under different formation conditions.

[0039] Step 103: Search the constructed stratigraphic database to obtain the rib adjustment parameters that match the stratigraphic features corresponding to the stratigraphic parameter data;

[0040] Step 104: When the similarity between the predicted rib adjustment parameters and the retrieved rib adjustment parameters is greater than the similarity threshold, control commands for each rib of the push-type rotary steering system are generated based on the predicted rib adjustment parameters, the design trajectory of the wellbore performing the drilling operation, and the actual trajectory.

[0041] Step 105: Send the control command for each rib to the drive unit of the push-type rotary guide system so that the drive unit drives the movement of each rib.

[0042] In this embodiment of the invention, a multi-dimensional data matching model is trained based on a large amount of historical formation parameter data, which can quickly output predicted rib adjustment parameters to provide a preliminary reference for drilling operations. The formation database stores the optimal adjustment parameters that have been verified in practice and match formation characteristics. The combination of the two, and the generation of control commands when the similarity meets the standard, utilizes both the rapid prediction capability of the model and the actual experience data of the database for verification, reducing the deviation that may occur in single prediction or retrieval, and making the adjustment parameters more in line with actual drilling needs. Different formation conditions vary greatly, and traditional methods are difficult to adapt quickly and accurately. In this technology, the multi-dimensional data matching model can learn the parameter mapping relationship under different formation conditions, and the formation database also covers a variety of formation characteristics and corresponding adjustment parameters. When encountering complex or special formations, the system can find suitable rib adjustment parameters through a dual approach of model prediction and database retrieval, enabling the push-type rotary steering system to adapt better to different formation environments and ensuring the stability of drilling operations. The technical process combines the wellbore design trajectory and the actual trajectory to generate control commands, and can adjust the movement of each rib in a targeted manner according to trajectory deviations. This process fully utilizes predicted and retrieved reliable adjustment parameters to ensure precise rib movement, effectively reducing the deviation between the actual and designed trajectories, improving wellbore trajectory control accuracy, and meeting the stringent trajectory requirements of drilling operations. The multi-dimensional data matching model can quickly output predicted parameters, and formation database retrieval can efficiently obtain matching parameters; their synergistic application shortens parameter determination time. Simultaneously, precise control commands make rib adjustment more efficient, reducing the number of adjustments and drilling downtime caused by inappropriate parameters, thereby improving overall drilling efficiency. From formation parameter acquisition, adjustment parameter prediction and retrieval, to control command generation and transmission, the entire process forms a coherent automated workflow, reducing manual intervention. The application of the multi-dimensional data matching model demonstrates the system's intelligent learning and predictive capabilities, enabling the push-type rotary steering system to autonomously adapt to different formation conditions, improving the system's intelligence level and degree of automation.

[0043] Each step is described in detail below.

[0044] In step 101, when the push-type rotary steering system is performing drilling operations, the collected formation parameter data is obtained;

[0045] The push-type rotary guide system includes a main body, a drill bit connection part, three ribs, and an adjustment mechanism. The adjustment mechanism includes a drive device, a displacement sensor, and a control module. The method proposed in this embodiment of the invention is applied to the control module of the push-type rotary guide system.

[0046] Once the push-type rotary steerable system is lowered into the well with the drill string and drilling operations commence, the formation detection device installed at the front end of the main unit immediately enters operational mode. This device integrates multiple advanced detection technologies, such as acoustic wave detection, gamma-ray measurement, and resistivity logging. Acoustic wave detection emits sound waves into the formation and receives the reflected waves. Based on parameters such as wave propagation time, velocity, and amplitude, it accurately analyzes the formation's mechanical properties, such as elastic modulus and Poisson's ratio, thereby determining the formation's hardness and lithological characteristics. Gamma-ray measurement utilizes the differences in radioactive elements contained in different rocks within the formation. By detecting gamma-ray intensity, it identifies the lithological type of the formation, such as mudstone, sandstone, and limestone. Resistivity logging measures the formation's resistivity to understand its porosity, fluid properties, and other information, providing abundant data for a comprehensive assessment of formation characteristics. These technologies work together to collect key parameters of the formation in real time and comprehensively, such as lithology, hardness, dip angle, porosity, and fluid properties. The collected analog signals are converted into high-speed electrical signals through a high-precision analog-to-digital converter module and transmitted quickly to the control module via fiber optic transmission or a dedicated data bus, ensuring the accuracy and real-time performance of the formation parameters.

[0047] In step 102, the formation parameter data is input into the multi-dimensional data matching model to obtain the predicted rib adjustment parameters. The multi-dimensional data matching model is obtained by training a neural network based on historical formation parameter data under different formation conditions.

[0048] The control module establishes a connection with the formation monitoring device via wired (e.g., dedicated cable) or wireless (e.g., electromagnetic waves, acoustic waves, etc.) downhole communication technologies. It receives formation parameter data collected and transmitted by the monitoring device in real time. This data includes, but is not limited to, lithology, rock hardness, formation dip angle, porosity, and permeability. Simultaneously, it acquires current wellbore trajectory design requirements (e.g., coordinate data of the target wellbore trajectory curve, dogleg tolerance range, etc.) and real-time tool operating status information (e.g., current drilling depth, rotational speed, bit pressure, drilling fluid properties, etc.) from other sensors or systems.

[0049] In one embodiment, the method further includes:

[0050] Historical stratigraphic parameter data for different geological conditions, as well as the corresponding rib adjustment parameters and drilling performance data of the push-type rotary steerable system, are obtained to form a dataset.

[0051] After standardizing the dataset, the neural network is iteratively trained using the dataset with the error between the predicted rib adjustment parameters and the actual effective rib adjustment parameters as the target, to obtain a trained multi-dimensional data matching model. The actual effective rib adjustment parameters are determined based on the drilling effect data.

[0052] In this embodiment of the invention, the multi-dimensional data matching model is based on a neural network architecture, typically constructed using a multi-layer neural network (such as a fully connected neural network, a convolutional neural network, or, considering the multi-dimensional characteristics of the data, a hybrid structure). The input layer of the multi-dimensional data matching model receives multi-dimensional formation parameter data from a formation detection device, such as lithology (which can be encoded as vectors of different categories), hardness (numerical data), formation dip angle (angular value), porosity, etc. The hidden layer of the multi-dimensional data matching model uses a large number of neurons and activation functions (such as ReLU, Sigmoid, etc.) to extract and transform features from the input formation parameter data, uncovering the complex relationships behind the data. The output layer of the multi-dimensional data matching model outputs rib adjustment parameters corresponding to the formation characteristics. These rib adjustment parameters are typically also multi-dimensional vectors, including but not limited to the rib's movement direction (which can be represented by discrete values, such as 1 representing inward, 2 representing outward, etc.), movement distance, and movement speed.

[0053] During training, the multi-dimensional data matching model uses historical formation parameter data under different formation conditions, rib adjustment parameters of the push-type rotary steering system, and corresponding drilling effect data to form a dataset. These datasets come from sources such as actual drilling operation records and simulation experimental data.

[0054] The dataset undergoes standardization, including cleaning and removing outliers and missing values. Numerical data is normalized using methods such as min-max normalization (mapping data to the 0-1 range) or Z-score normalization (making the data conform to a normal distribution with a mean of 0 and a standard deviation of 1) to eliminate the impact of dimensional differences on model training. For lithological and other categorical data, one-hot encoding or label encoding are used for numerical transformation.

[0055] The preprocessed data can be divided into training, validation, and test sets, typically in a 7:1:2 ratio. The training set is used for learning model parameters, the validation set is used to adjust model hyperparameters (such as learning rate, number of hidden layer neurons, etc.), and the test set is used to evaluate the model's generalization ability on unknown data.

[0056] Stochastic gradient descent (SGD), Adam, and other optimization algorithms can be used to target the error between the predicted rib adjustment parameters and the actual effective rib adjustment parameters. The error can be expressed as mean squared error (MSE) or mean absolute error (MAE). During training, the weights and biases of the neural network are continuously adjusted to ensure that the multi-dimensional data matching model can accurately map appropriate rib adjustment parameters from the formation parameter data.

[0057] Drilling performance data is an indispensable key element in the construction and optimization of the entire multidimensional data matching model. The core objective of the multidimensional data matching model is to learn matching rib adjustment parameters from formation parameters. However, not all adjustment parameters in historical records are valid. Drilling performance data (such as drilling speed, wellbore trajectory accuracy, tool wear, drilling efficiency, and stability) is an important basis for judging whether adjustment parameters are reasonable. If the drilling performance corresponding to a certain set of rib adjustment parameters is good (such as small wellbore trajectory deviation and stable drilling speed), then that set of rib adjustment parameters will be considered valid rib adjustment parameters and used as the target value (i.e., label) for training the multidimensional data matching model. If the drilling performance is poor (such as stuck pipe or excessive trajectory deviation), the corresponding rib adjustment parameters will be marked as invalid or low-quality data and may be removed during the data preprocessing stage to avoid the multidimensional data matching model learning incorrect mapping relationships.

[0058] In step 103, a search is performed in the constructed stratigraphic database to obtain rib adjustment parameters that match the stratigraphic features corresponding to the stratigraphic parameter data;

[0059] The stratigraphic database is structured and indexed according to stratigraphic characteristics (such as lithological classification, hardness range, and dip angle range). When stratigraphic characteristics are input, the database first determines the approximate data partitions based on lithology. Then, within the corresponding partitions, it performs precise filtering based on parameters such as hardness and dip angle to narrow the search range and quickly locate data records with similar stratigraphic characteristics.

[0060] In one embodiment, the method further includes:

[0061] Based on stratigraphic characteristics, the dataset is classified and aggregated to obtain multiple sets of stratigraphic characteristics;

[0062] Based on all rib adjustment parameters in each group of stratigraphic features, determine the optimal rib adjustment parameters corresponding to each group of stratigraphic features.

[0063] Each set of stratigraphic features is associated with the corresponding optimal rib adjustment parameters and stored in the stratigraphic database.

[0064] In the above embodiments, the dataset is classified and aggregated according to the definition of stratigraphic characteristics in actual business (such as a combination of lithology, hardness range, and dip angle range):

[0065] Determine the grouping dimensions: For example, classify lithology into categories such as sandstone, shale, and limestone; classify hardness into ranges such as low (<50MPa), medium (50-150MPa), and high (>150MPa); classify dip angle into ranges such as 0-10°, 10-30°, and >30°; and combine them to form specific stratigraphic characteristics such as sandstone + medium hardness + 10-30° dip angle.

[0066] Precise matching grouping: Each group of stratigraphic parameter data is assigned to the corresponding combination group (for example, if a sample has sandstone as its lithology, hardness of 80MPa (medium hardness), and dip angle of 20° (10-30°), it is assigned to the sandstone + medium hardness + 10-30° dip angle group).

[0067] Handling marginal or rare combinations: For stratigraphic features with very few samples (such as a stratigraphic feature group with only 1-2 samples), they can be temporarily marked as to be supplemented and updated after new data is accumulated.

[0068] For each group of formation characteristics, the optimal rib adjustment parameters are selected based on drilling performance indicators (efficiency, trajectory accuracy):

[0069] Define the optimal criterion:

[0070] It is necessary to clearly define the quantitative indicators that achieve the highest drilling efficiency and the best trajectory accuracy, for example:

[0071] Drilling efficiency: Take the sample with the largest advance per unit time;

[0072] Trajectory accuracy: The sample with the smallest deviation between the actual trajectory and the designed trajectory is selected;

[0073] If there are conflicting samples with high efficiency but low accuracy or high accuracy but low efficiency, a comprehensive score can be calculated by setting weights (e.g., efficiency accounts for 60% and accuracy accounts for 40%), and the sample with the highest comprehensive score can be selected; or the Pareto optimality principle can be used to screen parameters that are better than other samples in the group in both indicators.

[0074] If multiple optimal rib adjustment parameters exist for each group of formation characteristics (e.g., multiple rib adjustment parameters exhibit excellent performance in both efficiency and accuracy), further statistical analysis can be performed:

[0075] Take the average / median of the rib adjustment parameters (applicable to numerical parameters, such as travel distance and speed);

[0076] Select the rib adjustment parameter that appears most frequently (applicable to discrete parameters, such as the direction of movement);

[0077] All candidate optimal parameters are retained as parameter range suggestions (for flexible selection in actual operations).

[0078] In step 104, when the similarity between the predicted rib adjustment parameters and the retrieved rib adjustment parameters is greater than the similarity threshold, control commands for each rib of the push-type rotary steering system are generated based on the predicted rib adjustment parameters, the design trajectory of the wellbore performing the drilling operation, and the actual trajectory, and are sent to the drive device of the push-type rotary steering system.

[0079] In practice, cosine similarity, Euclidean distance and other measurement methods can be used to calculate the similarity between the predicted rib adjustment parameters and the retrieved rib adjustment parameters.

[0080] In one embodiment, control commands for each rib of the push-type rotary steering system are generated based on predicted rib adjustment parameters, the designed trajectory of the wellbore for drilling operations, and the actual trajectory, including:

[0081] Based on the designed trajectory and the actual trajectory, a trajectory deviation model is generated. The deviation parameters for the current well depth are calculated using the deviation model. The deviation parameters include spatial position deviation, directional deviation, and deviation change rate.

[0082] Determine the physical constraints, technological constraints, and formation constraints at the current well depth;

[0083] Based on the deviation parameters and drilling resistance, construct the objective function;

[0084] The control command elements are determined based on the number of ribs;

[0085] Search for the solution that minimizes the objective function in the parameter space formed by the control instruction elements;

[0086] The control commands for each rib are extracted from the obtained solution.

[0087] In this embodiment of the invention, a design trajectory model T_target(D) is constructed using cubic spline curves or parametric equations (such as (well depth D, azimuth θ, dip angle α) in cylindrical coordinates) based on the design trajectory parameters (e.g., three-dimensional coordinates corresponding to the well depth, azimuth sequence, and dip angle sequence) to ensure that the target position (x0, y0, z0) and direction vector (θ0, α0) can be accurately calculated at any well depth D. An actual trajectory model is constructed based on the actual trajectory measured while drilling (actual coordinates (x1, y1, z1), azimuth θ1, and dip angle α1 corresponding to the well depth D).

[0088] Based on the designed trajectory model and the actual trajectory model, a trajectory deviation model is formed.

[0089] The deviation parameters for the current well depth are calculated using the deviation model:

[0090] Spatial position deviation (straight-line distance deviation) ΔS;

[0091] Directional deviations: azimuth deviation Δθ, tilt deviation Δα;

[0092] Deviation change rate: Calculates the deviation increment within a 3-meter well section (used to determine the trajectory trend, such as whether the deviation continues).

[0093] Physical constraints:

[0094] For each rib (e.g., 3 ribs A / B / C), set independent parameter boundaries:

[0095] Movement direction: dir∈{-1,+1} (-1 means inward contraction, +1 means outward push; when it cannot be 0, the minimum action threshold must be specified);

[0096] Movement distance: d∈[0,d_max] (d_max is the maximum stroke of a single rib, such as 8mm, which is determined by the hardware design);

[0097] Movement speed: v∈[v_min,v_max] (v_min avoids slow movement affecting response, v_max prevents mechanical shock, such as 0.5~5mm / s).

[0098] Process constraints:

[0099] Dogleg constraint: K≤K_max (dogleg K is the directional change angle per 30-meter well section, K_max is usually 3~10° / 30m to avoid drill string fatigue);

[0100] Coordination constraint: The direction of the resultant force of the multi-rib plate push must be consistent with the direction of trajectory adjustment (deviation ≤ 5°) to avoid tool eccentricity.

[0101] Formation constraints:

[0102] Based on the current formation hardness H (e.g., MPa) and lithology, an upper limit for the pushing force is set (positively correlated with the displacement distance):

[0103] Soft formations (H<50MPa): d≤3mm (to prevent wellbore collapse);

[0104] Medium-hard formations (50≤H≤150MPa): d≤6mm;

[0105] Hard strata (H>150MPa): d≤8mm (requires greater thrust to adjust trajectory).

[0106] Based on trajectory deviation and drilling resistance, a calculable objective function F is constructed to clarify the optimization direction:

[0107] Trajectory deviation term E_trajectory:

[0108] Taking into account both position and orientation deviations, a weighted mean square error is used:

[0109] E_trajectory=w1×(ΔS / S_max) 2 +w2×(Δθ / θ_max) 2 +w3×(Δα / α_max) 2

[0110] Where S_max, θ_max, and α_max are the maximum allowable deviations (e.g., S_max = 10cm, θ_max = 5°), and w1 / w2 / w3 are the weights (e.g., if w2 increases in the horizontal section, w3 increases in the vertical section).

[0111] Drilling resistance term E_resistance:

[0112] Based on the relationship between the rib pushing force and the formation friction coefficient, the resistance is quantified:

[0113] E_resistance=Σ(μ_i×d_i×H)(μ_i is the friction coefficient between the i-th rib and the well wall, d_i is its moving distance, and H is the formation hardness), reflects the additional energy consumption and drill bit wear caused by pushing.

[0114] Objective function:

[0115] F = α × E_trajectory + β × E_resistance (α + β = 1, such as α = 0.7 when trajectory accuracy is prioritized, and β = 0.6 when efficiency is prioritized, calibrated using historical data).

[0116] Based on the number of ribs (taking 3 as an example), define the parameter vector to be optimized, that is, the control command elements of each rib:

[0117] X=[dir_A,d_A,v_A,dir_B,d_B,v_B,dir_C,d_C,v_C]

[0118] in:

[0119] dir_A / B / C∈{-1,+1}(discrete variable, direction of push / contraction);

[0120] d_A / B / C∈[0,d_max_i](continuous variable, distance moved under formation constraints);

[0121] v_A / B / C∈[v_min,v_max](continuous variable, movement speed).

[0122] The genetic algorithm (or particle swarm optimization algorithm) is used to search for the solution that minimizes the objective function in the parameter space formed by the control command elements. The steps are as follows:

[0123] Initialize the population:

[0124] Randomly generate N candidate solutions (e.g., N = 50), each solution must satisfy all constraints in step 2 (e.g., d_A ≤ d_max, K ≤ K_max), and filter out invalid solutions.

[0125] Fitness assessment:

[0126] For each candidate solution X:

[0127] Calculate the effect of the push-in force on the trajectory adjustment: Simulate the change in wellbore direction corresponding to X using a mechanical model, update the deviations Δθ, Δα, and ΔS, and obtain E_trajectory;

[0128] Calculate E_resistance;

[0129] The fitness value is F (the smaller the value, the better).

[0130] Iterative optimization:

[0131] Selection: Retain the top 30% of candidate solutions in terms of fitness (elite retention strategy);

[0132] Cross: Perform parameter cross on the selected solution (e.g., swap the d values ​​of ribs A and B) to generate a new solution;

[0133] Mutation: Randomly change some parameters (such as fine-tuning the v value or flipping the dir) to increase diversity;

[0134] Constraint verification: The new solution must satisfy the constraint conditions again, otherwise it will be eliminated or modified (e.g., if it exceeds d_max, it will be truncated to d_max).

[0135] Convergence criterion: When the optimal fitness value changes by ≤1% for 5 consecutive generations, or when the maximum number of iterations (e.g., 100 generations) is reached, the iteration stops and the optimal parameter vector X_opt is output.

[0136] Extract the instructions for each rib from X_opt:

[0137] Floor A: (dir_A_opt,d_A_opt,v_A_opt);

[0138] Rib plate B: (dir_B_opt, d_B_opt, v_B_opt);

[0139] Rib C: (dir_C_opt, d_C_opt, v_C_opt).

[0140] The system can simulate the execution of commands to verify whether the resultant force direction is consistent with the trajectory adjustment requirements (e.g., if Δθ needs to be reduced, the resultant force direction should point towards the target azimuth); whether the dogleg degree K is within the allowable range (calculated by simulating the direction change of a 30-meter well section); and whether the objective function F meets the threshold (e.g., F≤F0, where F0 is the maximum allowable optimization value in engineering). If not, the algorithm is returned to iterate again (increasing the number of iterations or adjusting the crossover / mutation probabilities).

[0141] The control commands for each rib can be converted into signals that the actuator can recognize:

[0142] Direction dir: Represented by binary signals (1 for pushing, 0 for contraction);

[0143] Distance d: Converted to pulse count (e.g., 1mm corresponds to 1000 pulses, d_A_opt = 3mm corresponds to 3000 pulses);

[0144] Speed ​​v: controlled by pulse frequency (e.g., v = 2 mm / s corresponds to 2000 pulses / second).

[0145] The encoded instructions are synchronously sent to the drive devices (motors / hydraulic devices) of each rib via a control bus (such as a CAN bus), and the timestamp of the instruction transmission is recorded for subsequent execution feedback verification.

[0146] In step 105, control commands for each rib are sent to the drive unit of the push-type rotary guide system so that the drive unit drives the movement of each rib.

[0147] In one embodiment, the method further includes:

[0148] The system receives displacement data of the rib plate collected by the drive device during the movement of the rib plate according to the control command, and adjusts the control parameters of the drive device according to the control command and the displacement data of the rib plate so that the rib plate moves according to the control command.

[0149] In this embodiment of the invention, the drive device, acting as the actuator for the rib plate movement, responds rapidly to and drives the corresponding rib plate to move radially after receiving a movement command from the control module. The drive device employs a high-precision hydraulic drive system equipped with a high-performance servo motor, a precision hydraulic pump, and a control valve assembly. The servo motor precisely adjusts the speed and output flow of the hydraulic pump according to the control command, while the control valve assembly precisely controls the flow direction and pressure of the hydraulic oil, thereby achieving precise displacement control of the rib plate. The displacement sensor, a key component for rib plate position monitoring, employs a non-contact laser displacement sensor or a high-precision linear variable differential transformer (LVDT) sensor, installed at a critical position between the rib plate and the main body. The sensor monitors the rib plate displacement in real time at an extremely high sampling frequency (up to the kilohertz level) and feeds back the displacement data to the control module in real time as a digital signal. Upon receiving the displacement data, the control module adjusts and optimizes the movement of the rib plate in real time.

[0150] In one embodiment, the control parameter is a PID control parameter;

[0151] Based on the control commands and the displacement data of the ribs, the control parameters of the drive device are adjusted, including:

[0152] Calculate the displacement deviation between the rib's moving distance and the rib's displacement data in the control command.

[0153] The displacement deviation and the current PID adjustment parameters are input into the trained deviation evolution model to obtain the predicted deviation sequence. The deviation evolution model is obtained by training the LSTM network based on historical deviations and historical PID adjustment parameters.

[0154] Using the rib's moving distance, rib's displacement data, displacement deviation, rate of change of displacement deviation, PID adjustment parameters, and predicted deviation sequence in the control command as the state space, and the increment of the PID adjustment parameters as the action, reinforcement learning is performed to obtain the optimal action space of the output.

[0155] Based on the optimal action space output and the current PID tuning parameters, generate new PID tuning parameters;

[0156] The new PID control parameters are input into the trained deviation evolution model to obtain the predicted deviation sequence;

[0157] If the two predicted deviation sequences meet the unpreset accuracy requirements, repeat the above steps until the two predicted deviation sequences meet the preset accuracy requirements, and then use the new PID adjustment parameters as the control parameters of the drive device.

[0158] First, determine the core mapping relationships:

[0159] PID control parameters → drive signal: u_pid = f(Kp, Ki, Kd, ​​e) (e is the displacement deviation, i.e., e = the moving distance of the rib in the control command d_ref - the displacement data of the rib d_act);

[0160] Drive signal → Rib displacement: Based on the hydraulic / motor drive characteristics, establish d_act=g(u_pid,H,F) (H is the formation hardness, F is the wellbore reaction force);

[0161] Deviation Evolution Model: The deviation evolution model is obtained by training an LSTM network. This deviation evolution model fits e(t+1)=h(e(t),Kp,Ki,Kd,H) to predict the future deviation trend. The prediction time domain Hp=5 (the next 5 control cycles, each cycle is 0.1s). Based on the current deviation e(t) and PID control parameters, the predicted deviation sequence [e(t+1),e(t+2),...,e(t+Hp)] within the future Hp is predicted.

[0162] State space S = [e, de / dt, d_act, d_ref, H, Kp, Ki, Kd, ​​prediction bias sequence]

[0163] (de / dt is the rate of change of displacement deviation, reflecting the adjustment trend).

[0164] Action space A: The increment of the PID control parameter, discretized using continuous space.

[0165] A = [ΔKp, ΔKi, ΔKd], where ΔKp ∈ [-0.1, 0.1] (step size 0.02), and ΔKi and ΔKd are similarly adjusted to ensure the smoothness of parameter fine-tuning.

[0166] Reward function R: Integrates MPC predictions with actual deviations to enhance accurate tracking.

[0167] R = α·[1-max(prediction bias sequence)] + β·exp(-e 2 )-γ·(|ΔKp|+|ΔKi|+|ΔKd|)

[0168] The first item: the smaller the future bias predicted by MPC, the higher the reward (α is the forward weight);

[0169] The second factor is: the smaller the current actual deviation, the higher the reward (β is the weight of real-time accuracy);

[0170] The third item: suppressing frequent adjustments of parameters (γ is the smoothing weight).

[0171] Action generation: The agent (using Deep Deterministic Policy Gradient (DDPG)) outputs the increment of the PID adjustment parameter A_t = [ΔKp, ΔKi, ΔKd] based on the current state S_t.

[0172] Calculate the new PID control parameters:

[0173] Kp' = clamp(Kp + ΔKp, Kp_min, Kp_max) (clamp is the truncation function to ensure it is within a safe range), and Ki' and Kd' are similar.

[0174] Kp' is used as the control parameter of the drive device, which causes the rib to move according to the control command. The above process is a cyclic process. Subsequently, the new PID adjustment parameters (Kp', Ki', Kd') are input into the deviation evolution model to re-predict the prediction deviation sequence of the future Hp cycle and verify whether it meets the accuracy requirements (e.g., maximum prediction deviation < 0.005 mm).

[0175] If the verification passes, output the new PID control parameters (Kp', Ki', Kd') for the current cycle.

[0176] In one embodiment, the method further includes:

[0177] Calculate the resultant force of all ribs based on their displacement data;

[0178] When the resultant force does not meet the preset requirements, the control parameters of the drive device are adjusted.

[0179] Based on the displacement data of the three ribs fed back by the displacement sensors, and adhering to the rigorous principle of force composition, the control module calculates the resultant force using the following formula, taking into account the fixed spatial distribution angles of the three ribs (the angles between adjacent ribs and the main axis are 0 degrees, 120 degrees, and 240 degrees, respectively):

[0180]

[0181] Where F1, F2, and F3 are the forces exerted by the three ribs on the well wall, and they are linearly positively correlated with the displacement data of each rib, i.e., F... i =kd i (k is a proportionality coefficient rigorously calibrated through experiments, taking into account various factors such as the material properties of the rib, structural dimensions, contact area, and formation friction coefficient), d i For the displacement data of the rib plate, θ i The angle between each rib and the main axis.

[0182] There are two situations where the resultant force does not meet the preset requirements. First, the magnitude of the resultant force deviates from the preset resultant force value (with an error range of ±2%), or the direction of the resultant force does not meet the wellbore trajectory dogleg requirements (with an error angle range of ±0.5°). In this case, the control module immediately activates the adaptive adjustment algorithm to adjust the control parameters of the drive device. Through this closed-loop control mechanism, it continuously iterates until the magnitude of the resultant force is constant and the direction meets the design requirements, ensuring the stability of the tool and the accuracy of trajectory control during the drilling process.

[0183] Throughout the drilling operation, the formation monitoring device continuously collects formation parameters, while the control module maintains real-time monitoring and analysis of formation changes and wellbore trajectory requirements. As drilling depth increases and formation conditions dynamically change, the control module repeatedly performs the aforementioned steps of formation parameter acquisition, parameter matching and command generation, precise rib adjustment, and resultant force calculation and closed-loop control, achieving dynamic and real-time adjustment of the tool rib status. For example, when drilling into a hard formation, the control module quickly matches the corresponding adjustment parameters based on the hardness parameters obtained from the formation monitoring device, generating commands to reduce the individual size of the three ribs. The drive unit responds rapidly, reducing the contact area between the ribs and the wellbore, effectively reducing drilling resistance and increasing drilling speed. Simultaneously, by precisely adjusting the rib movement, the module ensures that the resultant force remains constant and its direction conforms to the wellbore trajectory requirements, achieving efficient and precise drilling operations. When encountering sudden changes in formation dip angle, the control module can promptly adjust the thrust direction and magnitude of the ribs, allowing the tool to quickly adapt to formation changes, maintaining the accuracy of the wellbore trajectory, and ensuring the smooth progress of the drilling operation.

[0184] The following specific embodiment illustrates the effectiveness of the method proposed in this invention.

[0185] Taking the horizontal well drilling operation of shale gas in a large oilfield as an example, Figure 2 This is a flowchart illustrating the control process for a push-type rotary guide system in an embodiment of the present invention. Figure 2 During the drilling preparation phase, a large amount of formation parameter data related to shale gas formation characteristics was obtained, including lithological characteristics of shale at different depths (such as clay mineral content, quartz content, etc.), hardness distribution, formation dip angle variation, porosity, and permeability data. This resulted in a dataset, which was then classified and aggregated according to formation characteristics to obtain multiple sets of formation features. Based on all rib adjustment parameters in each set of formation features, the optimal rib adjustment parameters for each set were determined, constructing a formation database specific to this oilfield. According to the design requirements of this drilling operation, the curve of the wellbore design trajectory was defined, the maximum allowable dogleg angle of the wellbore trajectory was set to no more than 5° per 30-meter well section, and the preset constant value of the resultant force during drilling was determined to be 80 kN, with an error range controlled within ±1.6 kN.

[0186] Once the push-type rotary steerable system is lowered into the well with the drill string and drilling begins, the formation detection device at the front end of the main unit is quickly activated. The sonic detector emits acoustic signals into the formation at a frequency of 10 times per second and receives reflected wave data in real time. Through complex signal processing algorithms, it calculates mechanical parameters such as the formation's elastic modulus and Poisson's ratio. The gamma-ray detector continuously monitors the formation's radioactivity intensity and transmits the data to the control module for lithological identification. The resistivity logging tool simultaneously measures the formation's resistivity, obtaining information on formation porosity and fluid properties. At a well depth of 500 meters, the formation detection device's comprehensive analysis indicates that the current formation is a black shale rich in organic matter, with low hardness, a dip angle of 8°, and a porosity of 12%. These formation parameter data are transmitted to the control module in the form of high-speed electrical signals.

[0187] After receiving formation parameter data, the control module inputs the data into a multi-dimensional data matching model to obtain predicted rib adjustment parameters. It then searches the constructed formation database to obtain rib adjustment parameters that match the formation features corresponding to the formation parameter data. When the similarity between the predicted and retrieved rib adjustment parameters exceeds a similarity threshold, control commands for each rib of the push-type rotary steerable system are generated based on the predicted rib adjustment parameters, the wellbore's design trajectory, and the actual trajectory. These commands are then sent to the drive unit of the push-type rotary steerable system. Combining the current wellbore trajectory requirements and the tool's initial operating state (drill string speed 60 rpm, drilling pressure 120 kN), a trajectory control optimization algorithm is used to generate control commands for three ribs: rib one moves radially inward by 3 mm, rib two moves radially inward by 2.5 mm, and rib three moves radially inward by 2.8 mm, all at a speed of 1.5 mm / s.

[0188] Upon receiving the control command, the hydraulic servo system responds immediately. The servo motor precisely adjusts the output of the hydraulic pump, and the control valve group accurately controls the flow and pressure of the hydraulic oil, driving the three ribs to move radially according to the control command. Displacement sensors monitor the rib displacement data in real time at a sampling frequency of 1000Hz and feed the data back to the control module. Based on the displacement data, the control module uses a closed-loop control algorithm to fine-tune the rib movement in real time, ensuring the accuracy of the rib movement.

[0189] During rib adjustment, the control module calculates the resultant force exerted by the three ribs on the wellbore using the force composition formula based on displacement data from the displacement sensors. The calculation shows the current resultant force is 79.2 kN, with a 0.3° deviation from the designed wellbore trajectory direction. The control module determines that the resultant force meets the preset value requirement, but the directional deviation needs adjustment. Therefore, it regenerates a control command, increasing the movement distance of rib one to 3.2 mm, and sends it to the drive unit. After executing the control command, the drive unit recalculates the resultant force; this time, the resultant force is 80.1 kN, and the directional deviation has decreased to 0.1°, meeting the preset requirement.

[0190] As drilling continued, the formation monitoring system continuously collected formation parameters. When drilling reached a depth of 800 meters, the system detected that the formation lithology was gradually changing to silty shale with increased hardness, and the formation dip angle became 12°. Upon receiving the new formation parameter data, the control module quickly repeated the above control process, re-matched the rib adjustment parameters, generated new control commands, and drove the system to adjust the rib state according to the new commands, ensuring the system maintained efficient and precise drilling at all times. Throughout the drilling operation, the control module continuously and dynamically adjusted the rib state based on formation changes and wellbore trajectory requirements, ultimately successfully completing the drilling of this shale gas horizontal well. The wellbore trajectory error was controlled within a minimal range, and drilling efficiency was improved by approximately 40% compared to traditional control methods, achieving significant technical and economic benefits.

[0191] This invention also proposes a control device for a push-type rotary guide system, the principle of which is similar to the control method for a push-type rotary guide system, and will not be described in detail here.

[0192] Figure 3 This is a schematic diagram of the control device for a push-type rotary guide system in an embodiment of the present invention, including:

[0193] The data acquisition module 301 is used to acquire formation parameter data when the push-type rotary steering system is performing drilling operations.

[0194] The rib adjustment parameter prediction module 302 is used to input formation parameter data into a multi-dimensional data matching model to obtain the predicted rib adjustment parameters. The multi-dimensional data matching model is obtained by training a neural network based on historical formation parameter data under different formation conditions.

[0195] The retrieval module 303 is used to perform retrieval in the constructed stratigraphic database to obtain rib adjustment parameters that match the stratigraphic features corresponding to the stratigraphic parameter data.

[0196] The control command generation module 304 is used to generate control commands for each rib of the push-type rotary steering system based on the predicted rib adjustment parameters, the design trajectory of the wellbore performing the drilling operation, and the actual trajectory when the similarity between the predicted rib adjustment parameters and the retrieved rib adjustment parameters is greater than the similarity threshold.

[0197] The control command sending module 305 is used to send control commands for each rib to the drive device of the push-type rotary guide system, so that the drive device drives the movement of each rib.

[0198] Figure 4 This is another schematic diagram of a control device for a push-type rotary guide system according to an embodiment of the present invention. In one embodiment, the device further includes a model training module 401, used for:

[0199] Historical stratigraphic parameter data for different geological conditions, as well as the corresponding rib adjustment parameters and drilling performance data of the push-type rotary steerable system, are obtained to form a dataset.

[0200] After standardizing the dataset, the neural network is iteratively trained using the dataset with the error between the predicted rib adjustment parameters and the actual effective rib adjustment parameters as the target, to obtain a trained multi-dimensional data matching model. The actual effective rib adjustment parameters are determined based on the drilling effect data.

[0201] In one embodiment, the apparatus further includes a formation database construction module 402, for:

[0202] Based on stratigraphic characteristics, the dataset is classified and aggregated to obtain multiple sets of stratigraphic characteristics;

[0203] Based on all rib adjustment parameters in each group of stratigraphic features, determine the optimal rib adjustment parameters corresponding to each group of stratigraphic features.

[0204] Each set of stratigraphic features is associated with the corresponding optimal rib adjustment parameters and stored in the stratigraphic database.

[0205] In one embodiment, the control command generation module is used to:

[0206] Based on the designed trajectory and the actual trajectory, a trajectory deviation model is generated. The deviation parameters for the current well depth are calculated using the deviation model. The deviation parameters include spatial position deviation, directional deviation, and deviation change rate.

[0207] Determine the physical constraints, technological constraints, and formation constraints at the current well depth;

[0208] Based on the deviation parameters and drilling resistance, construct the objective function;

[0209] The control command elements are determined based on the number of ribs;

[0210] Search for the solution that minimizes the objective function in the parameter space formed by the control instruction elements;

[0211] The control commands for each rib are extracted from the obtained solution.

[0212] In one embodiment, the device further includes a control parameter adjustment module 403, used for:

[0213] The system receives displacement data of the rib plate collected by the drive device during the movement of the rib plate according to the control command, and adjusts the control parameters of the drive device according to the control command and the displacement data of the rib plate so that the rib plate moves according to the control command.

[0214] In one embodiment, the control parameter is a PID control parameter;

[0215] The control parameter adjustment module 403 is used for:

[0216] Calculate the displacement deviation between the rib's moving distance and the rib's displacement data in the control command.

[0217] The displacement deviation and the current PID adjustment parameters are input into the trained deviation evolution model to obtain the predicted deviation sequence. The deviation evolution model is obtained by training the LSTM network based on historical deviations and historical PID adjustment parameters.

[0218] Using the rib's moving distance, rib's displacement data, displacement deviation, rate of change of displacement deviation, PID adjustment parameters, and predicted deviation sequence in the control command as the state space, and the increment of the PID adjustment parameters as the action, reinforcement learning is performed to obtain the optimal action space of the output.

[0219] Based on the optimal action space output and the current PID tuning parameters, generate new PID tuning parameters;

[0220] The new PID control parameters are input into the trained deviation evolution model to obtain the predicted deviation sequence;

[0221] If the two predicted deviation sequences meet the unpreset accuracy requirements, repeat the above steps until the two predicted deviation sequences meet the preset accuracy requirements, and then use the new PID adjustment parameters as the control parameters of the drive device.

[0222] In one embodiment, the control command generation module is further configured to:

[0223] Calculate the resultant force of all ribs based on their displacement data;

[0224] When the resultant force does not meet the preset requirements, the control parameters of the drive device are adjusted.

[0225] The beneficial effects achieved by the method and apparatus proposed in the embodiments of the present invention are as follows:

[0226] Exceptional adaptive adjustment performance: Capable of quickly and accurately sensing subtle changes in formation characteristics, and automatically generating optimal rib adjustment commands based on preset adjustment strategies and algorithms. This enables the push-type rotary steering system to respond rapidly in complex and ever-changing formation environments, achieving highly adaptive adjustment. Whether facing abrupt changes in lithology, differences in hardness, or variations in formation dip angle, the system can quickly adjust to its optimal operating state, significantly improving its environmental adaptability and operational reliability, and effectively reducing drilling accidents and inefficiencies caused by formation incompatibility.

[0227] High-precision trajectory and resultant force control: The application of an innovative force synthesis calculation model and closed-loop control algorithm ensures a high degree of constancy in the magnitude and precise control of the resultant force during the adjustment of the three ribs. Through precise displacement monitoring and real-time resultant force calculation, minute and precise adjustments can be made to the ribs, controlling the fluctuation range of the resultant force magnitude within a very small error range (±2%), while ensuring a high degree of conformity between the resultant force direction and the wellbore trajectory dogleg accuracy requirements (error angle ±0.5°). This high-precision control capability enables the wellbore trajectory to be drilled strictly according to design requirements, greatly improving the control accuracy of the wellbore trajectory, providing a solid foundation for subsequent oil and gas extraction operations, and effectively reducing the increase in drilling costs and resource waste caused by trajectory deviations.

[0228] Significantly Improved Drilling Efficiency: By optimizing the discrete adjustment strategy of the ribs, this invention effectively reduces the frictional resistance between the system and the wellbore. While maintaining trajectory control accuracy, it significantly increases drilling speed. Practical engineering applications have verified that, compared to traditional control methods, the push-type rotary steering system using this invention can improve drilling efficiency by 30%-50% in complex formations. This means that under the same drilling task, it can significantly shorten the drilling cycle, reduce costs such as equipment rental time, manpower input, and energy consumption, bringing significant economic benefits to oil drilling companies. Simultaneously, the high drilling performance also helps reduce the occurrence of complex downhole conditions, improving the safety and reliability of drilling operations.

[0229] Wide versatility and scalability: Employing a modular and standardized design concept, it can be adapted to various models and specifications of push-type rotary steering systems. Whether it's a new high-end system or an existing conventional system, only appropriate adjustments to the hardware interface and software parameters are needed to implement this control method. Furthermore, by continuously updating the formation database and optimizing the control algorithm, this method can be easily extended to more complex drilling scenarios, such as deep-sea drilling and high-temperature, high-pressure formation drilling. This excellent versatility and scalability give this invention broad market application prospects and can provide strong support for the technological upgrading and development of the oil drilling industry.

[0230] This invention also provides a computer device. Figure 5 This is a schematic diagram of a computer device in an embodiment of the present invention. The computer device 500 includes a memory 510, a processor 520, and a computer program 530 stored in the memory 510 and executable on the processor 520. When the processor 520 executes the computer program 530, it implements the above-described control method for a push-type rotary guide system.

[0231] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described control method for a push-type rotary guide system.

[0232] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described control method for a push-type rotary guide system.

[0233] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0234] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0235] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0236] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0237] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A control method for a push-the-bit rotary steerable system, characterized in that, include: When the push-type rotary steering system is performing drilling operations, it acquires formation parameter data. The stratigraphic parameter data is input into a multi-dimensional data matching model to obtain the predicted rib adjustment parameters. The multi-dimensional data matching model is obtained by training a neural network based on historical stratigraphic parameter data under different stratigraphic conditions. Search the constructed stratigraphic database to obtain the rib adjustment parameters that match the stratigraphic features corresponding to the stratigraphic parameter data; When the similarity between the predicted rib adjustment parameters and the retrieved rib adjustment parameters is greater than the similarity threshold, control commands for each rib of the push-type rotary steering system are generated based on the predicted rib adjustment parameters, the design trajectory of the wellbore performing the drilling operation, and the actual trajectory. Control commands for each rib are sent to the drive unit of the push-type rotary guide system, so that the drive unit drives the movement of each rib.

2. The method according to claim 1, characterized in that, Also includes: Historical stratigraphic parameter data for different geological conditions, as well as the corresponding rib adjustment parameters and drilling performance data of the push-type rotary steerable system, are obtained to form a dataset. After standardizing the dataset, the neural network is iteratively trained using the dataset with the error between the predicted rib adjustment parameters and the actual effective rib adjustment parameters as the target, to obtain a trained multi-dimensional data matching model. The actual effective rib adjustment parameters are determined based on the drilling effect data.

3. The method according to claim 1, characterized in that, Also includes: Based on stratigraphic characteristics, the dataset is classified and aggregated to obtain multiple sets of stratigraphic characteristics; Based on all rib adjustment parameters in each group of stratigraphic features, determine the optimal rib adjustment parameters corresponding to each group of stratigraphic features. Each set of stratigraphic features is associated with the corresponding optimal rib adjustment parameters and stored in the stratigraphic database.

4. The method according to claim 1, characterized in that, Based on the predicted rib adjustment parameters, the designed trajectory of the wellbore for drilling operations, and the actual trajectory, control commands for each rib of the push-type rotary steering system are generated, including: Based on the designed trajectory and the actual trajectory, a trajectory deviation model is generated. The deviation parameters for the current well depth are calculated using the deviation model. The deviation parameters include spatial position deviation, directional deviation, and deviation change rate. Determine the physical constraints, technological constraints, and formation constraints at the current well depth; Based on the deviation parameters and drilling resistance, construct the objective function; The control command elements are determined based on the number of ribs; Search for the solution that minimizes the objective function in the parameter space formed by the control instruction elements; The control commands for each rib are extracted from the obtained solution.

5. The method according to claim 1, characterized in that, Also includes: The system receives displacement data of the rib plate collected by the drive device during the movement of the rib plate according to the control command, and adjusts the control parameters of the drive device according to the control command and the displacement data of the rib plate so that the rib plate moves according to the control command.

6. The method according to claim 5, characterized in that, The control parameters are PID adjustment parameters; Based on the control commands and the displacement data of the ribs, the control parameters of the drive device are adjusted, including: Calculate the displacement deviation between the rib's moving distance and the rib's displacement data in the control command. The displacement deviation and the current PID adjustment parameters are input into the trained deviation evolution model to obtain the predicted deviation sequence. The deviation evolution model is obtained by training the LSTM network based on historical deviations and historical PID adjustment parameters. Using the rib's moving distance, rib's displacement data, displacement deviation, rate of change of displacement deviation, PID adjustment parameters, and predicted deviation sequence in the control command as the state space, and the increment of the PID adjustment parameters as the action, reinforcement learning is performed to obtain the optimal action space of the output. Based on the optimal action space output and the current PID tuning parameters, generate new PID tuning parameters; The new PID control parameters are input into the trained deviation evolution model to obtain the predicted deviation sequence; If the two predicted deviation sequences meet the unpreset accuracy requirements, repeat the above steps until the two predicted deviation sequences meet the preset accuracy requirements, and then use the new PID adjustment parameters as the control parameters of the drive device.

7. The method according to claim 1, characterized in that, Also includes: Calculate the resultant force of all ribs based on their displacement data; When the resultant force does not meet the preset requirements, the control parameters of the drive device are adjusted.

8. A control device for a push-type rotary guide system, characterized in that, include: The data acquisition module is used to acquire formation parameter data when the push-type rotary steering system is performing drilling operations. The rib adjustment parameter prediction module is used to input formation parameter data into a multi-dimensional data matching model to obtain the predicted rib adjustment parameters. The multi-dimensional data matching model is obtained by training a neural network based on historical formation parameter data under different formation conditions. The retrieval module is used to search the constructed stratigraphic database to obtain rib adjustment parameters that match the stratigraphic features corresponding to the stratigraphic parameter data. The control command generation module is used to generate control commands for each rib of the push-type rotary steering system when the similarity between the predicted rib adjustment parameters and the retrieved rib adjustment parameters is greater than the similarity threshold, based on the predicted rib adjustment parameters, the design trajectory of the wellbore performing the drilling operation, and the actual trajectory. The control command sending module is used to send control commands for each rib to the drive device of the push-type rotary guide system, so that the drive device drives the movement of each rib.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1 to 7.

11. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method of any one of claims 1 to 7.

Images