Intelligent control method and device for high-precision drilling rotary guiding track

Abstract

The embodiment of the invention provides an intelligent control method and device for a high-precision drilling rotary guiding track, and belongs to the technical field of drilling control. The method comprises the following steps: constructing an observation sequence for representing a drilling state based on multi-source observation data; executing state estimation based on the observation sequence, and generating a state estimation result containing borehole attitude parameters and dynamic indexes; constructing a constraint optimization model based on the state estimation result and a preset trajectory target, and executing model prediction control in the constraint optimization model to generate a control quantity sequence for driving a rotary steering tool; and acting the control quantity sequence on the rotary steering tool to execute track adjustment, and performing residual calculation and model parameter updating in real time based on execution feedback information and a corresponding state estimation result until drilling is completed. According to the scheme, high-precision closed-loop intelligent control over the well drilling track is achieved, track deviation is remarkably reduced, and drilling stability is improved.

Description

Technical Field

[0001] This invention relates to the field of drilling control technology, specifically to a high-precision intelligent control method and a high-precision intelligent control device for drilling rotary steering trajectory. Background Technology

[0002] Rotary steerable drilling technology is a key method in current directional and horizontal well construction. Its basic principle is to continuously adjust the drill bit attitude using a downhole rotary steerable tool during drilling to ensure the wellbore trajectory extends according to a preset target. Compared to traditional directional drilling tools, rotary steerable technology can continuously correct the trajectory during drilling, significantly improving construction efficiency. However, existing technologies still have significant challenges in high-precision trajectory control.

[0003] Existing rotary steering trajectory control methods typically rely on distributed measurement systems at the surface and downhole. Trajectory deviations are calculated through wellbore attitude measurements and feedback from a limited number of operating parameters, and then the control program generates steering commands. However, under complex formation conditions, the downhole environment exhibits significant nonlinear characteristics, such as friction between the drill string and the wellbore, stick-slip and impact vibrations, and local formation heterogeneity. These factors can cause fluctuations or even distortions in wellbore attitude data and operating parameters. Existing methods often use linear approximation models or empirical formulas, which fail to fully reflect the actual trajectory state.

[0004] Because of the discrepancy between the state estimation and the actual wellbore trajectory, control commands often fail to match the actual drilling results after being issued and executed. This causes the trajectory to gradually deviate from the preset target, requiring manual monitoring and frequent corrections to maintain the required drilling conditions. This accumulation of deviations not only reduces the accuracy of trajectory control but also introduces the risk of wellbore collisions and uneven well network deployment, increasing the construction period and costs.

[0005] Therefore, the core deficiency of existing rotary steerable trajectory control methods lies in the lack of accurate modeling and estimation of complex downhole nonlinear characteristics, making it impossible to achieve high-precision determination of wellbore trajectory status, thus directly limiting the improvement of trajectory control accuracy. This problem has become a bottleneck for the further development of rotary steerable drilling technology. Summary of the Invention

[0006] The purpose of this invention is to provide a high-precision intelligent control method and device for rotary steerable drilling trajectory, so as to at least solve the problem of insufficient trajectory state estimation accuracy and the resulting accumulation of trajectory control deviations in the existing rotary steerable drilling process.

[0007] To achieve the above objectives, the first aspect of the present invention provides a high-precision intelligent control method for rotary steering trajectories in drilling. The method includes: acquiring multi-source observation data, including wellbore attitude data and bottom hole combined operation data, during the target drilling process, and constructing an observation sequence to characterize the drilling state based on the multi-source observation data; performing state estimation based on the observation sequence to generate a state estimation result including wellbore attitude parameters and dynamic indices; constructing a constrained optimization model based on the state estimation result and a preset trajectory target, and performing model predictive control in the constrained optimization model to generate a control quantity sequence for driving the rotary steering tool; applying the control quantity sequence to the rotary steering tool to perform trajectory adjustment, and performing residual calculation and model parameter updates in real time based on execution feedback information and corresponding state estimation results until drilling is completed.

[0008] Optionally, the wellbore attitude data includes inclination angle, azimuth angle, and spatial curvature parameters; the bottom hole combined operation data includes drilling pressure, rotational speed, and torque; constructing an observation sequence to characterize the drilling state based on the multi-source observation data includes: uniformly aligning the timestamps of the multi-source observation data, and performing interpolation resampling based on the aligned data to obtain a synchronization sequence; performing outlier removal and filtering on the synchronization sequence, and obtaining corresponding attitude denoising sequences and operation denoising sequences based on the processed wellbore attitude data and bottom hole combined operation data, respectively; extracting the rate of change of inclination angle, azimuth angle, and spatial curvature based on the attitude denoising sequence as a first feature; extracting rotational speed, torque, and mechanical power surrogate based on the operation denoising sequence as a second feature; and connecting the first and second features in chronological order after normalization to form an observation sequence to characterize the drilling state.

[0009] Optionally, state estimation is performed based on the observation sequence to generate a state estimation result containing wellbore attitude parameters and dynamic indices, including: inputting the observation sequence into a nonlinear state estimation model, extracting sequence components related to wellbore attitude and solving for well inclination angle, azimuth angle and spatial curvature to obtain wellbore attitude parameters; simultaneously extracting sequence components related to bottom hole combined operation and performing frequency domain analysis to identify stick-slip frequency bands and shock vibration frequency bands and calculate corresponding intensity values ​​to form dynamic indices; and storing the wellbore attitude parameters and the dynamic indices as a combination of state estimation results.

[0010] Optionally, the nonlinear state estimation model is constructed by establishing nonlinear observation equations and state transition equations between the observation sequence and state variables, and incorporating an unscented Kalman filter algorithm or an extended Kalman filter algorithm into the equation solving framework. This model is used to output state estimation results containing wellbore attitude parameters and dynamic indices during the state estimation process.

[0011] Optionally, sequence components related to the combined operation at the bottom of the well are extracted and frequency domain analysis is performed to identify stick-slip frequency bands and shock vibration frequency bands and calculate corresponding intensity values ​​to form dynamic indices. This includes: extracting sequence components related to the combined operation at the bottom of the well from the observation sequence; performing a fast Fourier transform based on the extracted sequence components to obtain a spectral distribution; identifying stick-slip frequency bands and shock vibration frequency bands in the spectral distribution, and calculating amplitude and energy intensity within the identified frequency bands to form dynamic indices for state estimation results.

[0012] Optionally, a constrained optimization model is constructed based on the state estimation results and the preset trajectory target, including: comparing the wellbore attitude parameters and dynamic indices in the state estimation results with the preset trajectory target to determine the current trajectory deviation and stability constraint; establishing a set of nonlinear constraints including spatial curvature constraints, build-up rate constraints, and anti-collision constraints based on the trajectory deviation and stability constraints; transforming the preset trajectory target into an objective function, and combining the objective function with the set of nonlinear constraints to form an optimization solution structure; and calling a rolling time-domain model predictive control algorithm in the optimization solution structure to obtain the constrained optimization model.

[0013] Optionally, model predictive control is performed in the constrained optimization model to generate a sequence of control quantities for driving the rotary steerable tool, including: predicting the future wellbore attitude change trend based on the state estimation results in the rolling time domain; calculating the deviation between the future wellbore attitude change trend and the preset trajectory target to obtain a trajectory deviation amount for optimization; iteratively solving the objective function in the constrained optimization model under the trajectory deviation amount constraint to obtain a combination of control variables in multiple candidate solution spaces; selecting a solution from the candidate solution space that simultaneously satisfies the spatial curvature constraint, build-up rate constraint, and anti-collision constraint, and outputting a sequence of control quantities consisting of deflection plate angle, phase parameter, and drilling pressure and rotation speed setpoints.

[0014] Optionally, the control sequence is applied to the rotary steerable tool to perform trajectory adjustment, and residual calculation and model parameter updates are performed in real time based on execution feedback information and corresponding state estimation results. This includes: sending the deflection plate angle and phase parameters in the control sequence to the deflection drive component of the rotary steerable tool, and simultaneously inputting the drilling pressure and rotation speed setpoints to the surface control system to adjust the drilling pressure actuator and the rotary drive device; after the control sequence is applied, the wellbore attitude data and operating data returned by the rotary steerable tool are collected in real time and compared with the state estimation results to calculate the residual; when the residual is lower than a preset residual threshold, the original control cycle is maintained; when the residual exceeds the preset residual threshold, model parameter updates and control cycle adjustments are triggered, and a new control sequence is generated to be applied to the rotary steerable tool again.

[0015] Optionally, when the residual exceeds a preset residual threshold, model parameter updates and control cycle adjustments are triggered, and a new control quantity sequence is generated and applied to the rotary steering tool again. This includes: comparing the residual with the preset residual threshold; if the residual exceeds the preset residual threshold, calling the surrogate model used for constraint optimization and correcting the friction coefficient and formation stiffness parameters based on the residual change to form updated model parameters; adjusting the control cycle according to the magnitude of the residual exceeding the preset residual threshold, with a shorter control cycle for a larger excess; and re-executing model predictive control under the updated model parameters and new control cycle configuration to generate a new control quantity sequence and apply it to the rotary steering tool again.

[0016] A second aspect of the present invention provides a high-precision intelligent control device for rotary steering trajectories in drilling. The device includes: an acquisition unit, configured to acquire multi-source observation data, including wellbore attitude data and bottom hole combined operation data, during the target drilling process, and construct an observation sequence characterizing the drilling state based on the multi-source observation data; an estimation unit, configured to perform state estimation based on the observation sequence, generating a state estimation result including wellbore attitude parameters and dynamic indices; a scheme generation unit, configured to construct a constrained optimization model based on the state estimation result and a preset trajectory target, and execute model predictive control in the constrained optimization model to generate a control quantity sequence for driving the rotary steering tool; and an execution unit, configured to apply the control quantity sequence to the rotary steering tool to perform trajectory adjustment, and perform residual calculation and model parameter update in real time based on execution feedback information and corresponding state estimation results until drilling is completed.

[0017] Through the above technical solution, this invention acquires wellbore attitude data and bottomhole combined operation data during drilling, and fuses multi-source observation data to form a more comprehensive description of the downhole state. Based on this, nonlinear state estimation yields state estimation results including attitude parameters and dynamic indices, making the trajectory calculation closer to the actual wellbore state. A constrained optimization model is established in conjunction with a preset trajectory target, and the control quantity sequence is solved within the model predictive control framework. This allows for the generation of reasonable directional commands while considering spatial curvature, build-up rate, and stability conditions. Furthermore, by applying the control quantities to the rotary steerable tool and using feedback information in real time for residual determination and parameter updates, a closed-loop correction mechanism can be formed, continuously correcting trajectory deviations. Overall, this method effectively improves the control accuracy and stability of rotary steerable drilling trajectories, reduces the accumulation of trajectory deviations, and improves construction efficiency and wellbore quality.

[0018] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0019] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of the steps of a high-precision intelligent control method for drilling rotary steering trajectory provided by one embodiment of the present invention; Figure 2 This is a structural diagram of a high-precision intelligent control device for drilling rotary guide trajectory provided in one embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for illustration and explanation of the embodiments of this application and are not intended to limit the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0021] It should be noted that if the embodiments of this application involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0022] Furthermore, if the embodiments of this application involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed in this application.

[0023] like Figure 1 As shown, this invention provides a high-precision intelligent control method for drilling rotary steering trajectory, the method comprising: Step S1: Acquire multi-source observation data, including wellbore attitude data and bottom hole combined operation data, during the target drilling process, and construct an observation sequence to characterize the drilling status based on the multi-source observation data.

[0024] Specifically, the wellbore attitude data includes inclination angle, azimuth angle, and spatial curvature parameters; the bottom hole combined operation data includes drilling pressure, rotational speed, and torque; an observation sequence for characterizing the drilling state is constructed based on the multi-source observation data, including: uniformly aligning the timestamps of the multi-source observation data, and performing interpolation resampling based on the aligned data to obtain a synchronization sequence; performing outlier removal and filtering on the synchronization sequence, and obtaining corresponding attitude denoising sequences and operation denoising sequences based on the processed wellbore attitude data and bottom hole combined operation data, respectively; extracting the rate of change of inclination angle, azimuth angle, and spatial curvature based on the attitude denoising sequence as a first feature; extracting rotational speed, torque, and mechanical power surrogate based on the operation denoising sequence as a second feature; and connecting the first and second features in chronological order after normalization to form an observation sequence for characterizing the drilling state.

[0025] In the specific implementation process, it is first necessary to continuously collect multi-source observation data during the target drilling operation to fully characterize the downhole trajectory and operating status. The multi-source observation data mentioned here mainly includes two categories: wellbore attitude data and bottomhole assembly operation data. Wellbore attitude data is usually provided in real time by the measurement-while-drilling unit (MWD), and its key indicators include inclination angle, azimuth angle, and spatial curvature parameters calculated from the combination of inclination angle and azimuth angle. These parameters directly reflect the spatial position of the wellbore and the degree of trajectory curvature, and are the basic information for trajectory accuracy control. Bottomhole assembly operation data is mainly provided by measuring elements installed at various nodes of the bottomhole assembly, including three indicators: drilling pressure, rotational speed, and torque. These are used to characterize the stress state of the drill bit, rotational speed, and torsional condition of the drill string, respectively, and are important bases for reflecting mechanical loading characteristics.

[0026] After acquiring the two types of data mentioned above, it is necessary to process and unify the data from different sources. Since wellbore attitude data and bottom-hole combined operation data come from different acquisition channels, their sampling frequencies and timestamps often differ. Directly splicing them together would lead to timing misalignment and feature distortion. Therefore, before data fusion, the timestamps of the multi-source data must first be aligned and unified. Specifically, interpolation and resampling operations are performed on various types of data based on a unified time reference, mapping them to the same time step to generate a synchronization sequence. After this processing, the attitude data and operation data at each moment can correspond one-to-one, avoiding trajectory estimation errors caused by asynchronous sampling.

[0027] After obtaining the synchronization sequence, further data quality improvement is needed. The drilling environment is complex, and sensor signals are easily affected by factors such as downhole high temperature and pressure, drill string vibration, and electromagnetic interference, often resulting in noise or outliers in the raw data. Using this data directly without processing will significantly reduce the reliability of subsequent trajectory estimation and optimization models. Therefore, outlier removal and filtering must be performed on the synchronization sequence. Outlier removal can be based on set threshold rules or statistical distribution judgments to remove outliers that are far from the normal range; filtering can use methods such as Kalman filtering or low-pass filtering to reduce high-frequency noise. After this step, smoother sequence data is obtained, which can be further divided into attitude denoising sequences and operational denoising sequences. The former mainly preserves the variation patterns of well inclination angle, azimuth angle, and spatial curvature, while the latter corresponds to the stable trends of drilling pressure, rotational speed, and torque.

[0028] After obtaining the denoised attitude and operation sequences, further feature extraction is required to ensure the data reflects the dynamic changes in the wellbore state. For the attitude denoised sequence, the rates of change of the inclination angle, azimuth angle, and spatial curvature are first calculated. These three parameters quantify the speed and direction of the wellbore spatial attitude change over time, reflecting the deviation trend of the wellbore trajectory. Using these rates of change as the first feature input enhances the sensitivity of the state estimation model to trajectory changes. For the operation denoised sequence, the rates of change of rotational speed and torque, as well as the mechanical power surrogate formed by the combination of rotational speed and torque, are calculated. The mechanical power surrogate reflects the power consumption level of the drill bit at the bottom of the well and is directly related to trajectory stability and wellbore morphology. Through this step, the operation data is not only the raw signal but is transformed into a second feature with direct guiding value for the trajectory.

[0029] Finally, the first and second features are processed uniformly. To avoid the imbalance caused by different dimensions and numerical ranges, all features need to be normalized to ensure they are expressed at the same scale. The normalized features are stacked and encoded in chronological order to form a continuous sequence of observation vectors. This sequence is called the observation sequence used to characterize the drilling state. It contains feature information from both wellbore attitude and bottomhole combined operation, reflecting both the trend of geometric position changes and the loading status of dynamic operation.

[0030] Step S2: Perform state estimation based on the observation sequence to generate state estimation results including wellbore attitude parameters and dynamic indices.

[0031] Specifically, the observation sequence is input into a nonlinear state estimation model to extract sequence components related to the wellbore attitude and solve for the inclination angle, azimuth angle, and spatial curvature to obtain wellbore attitude parameters; at the same time, sequence components related to bottom hole combined operation are extracted and frequency domain analysis is performed to identify stick-slip frequency bands and shock vibration frequency bands and calculate the corresponding intensity values ​​to form dynamic indices; the wellbore attitude parameters and the dynamic indices are combined and stored as state estimation results.

[0032] Furthermore, the nonlinear state estimation model is constructed by establishing nonlinear observation equations and state transition equations between the observation sequence and state variables, and by introducing the unscented Kalman filter algorithm or the extended Kalman filter algorithm into the equation solving framework. This model is used to output state estimation results containing wellbore attitude parameters and dynamic indices during the state estimation process.

[0033] Furthermore, sequence components related to the combined operation at the bottom of the well are extracted and frequency domain analysis is performed to identify the stick-slip frequency band and the shock vibration frequency band and calculate the corresponding intensity values ​​to form dynamic indices. This includes: extracting sequence components related to the combined operation at the bottom of the well from the observation sequence; performing a fast Fourier transform based on the extracted sequence components to obtain the spectral distribution; identifying the stick-slip frequency band and the shock vibration frequency band in the spectral distribution, and calculating the amplitude and energy intensity within the identified frequency band range to form dynamic indices for the state estimation results.

[0034] In the specific implementation process, state estimation needs to be performed based on the observation sequence to obtain more accurate trajectory and operational status information, thereby providing reliable input for subsequent constraint optimization and control decisions. The core objective of this state estimation is to extract information related to the wellbore's geometric attitude and information related to the combined operational status at the bottom of the well, based on preprocessed multi-source data. This information is then fused using a reasonable mathematical model to obtain a state estimation result that simultaneously reflects the wellbore's spatial position and dynamic conditions. This result contains at least two types of information: wellbore attitude parameters, mainly including inclination angle, azimuth angle, and spatial curvature; and dynamic indices, mainly including intensity indices of stick-slip vibration and impact vibration.

[0035] Specifically, the first step is to input the observation sequence into the nonlinear state estimation model. This observation sequence is constructed from the wellbore attitude data and bottom hole combined operation data in the previous steps after alignment, filtering, and feature extraction. It already possesses a unified time reference and a high signal-to-noise ratio. After inputting it into the model, the first step is to extract the sequence components related to the wellbore attitude, such as the time-varying sequence of inclination angle, the time-varying sequence of azimuth angle, and the spatial curvature sequence calculated based on their combination. In the nonlinear state estimation model, these components are substituted into the observation equations, and through joint solution with the state transition equations, smoother and more accurate wellbore attitude parameters are iteratively derived. Unlike traditional linear models, this approach emphasizes the introduction of nonlinear features, which can better reflect the true attitude changes of the wellbore under complex formation conditions.

[0036] Simultaneously, it is necessary to extract sequence components related to the combined operation at the bottom of the well from the observation sequence. These components mainly include parameters such as drilling pressure, rotational speed, and torque. Relying solely on these raw quantities often fails to directly characterize the downhole dynamics; therefore, in this embodiment, frequency domain analysis is performed on these sequence components. Specifically, based on the Fast Fourier Transform (FFT) method, the time-domain signal is transformed into a frequency-domain spectral distribution. This allows the identification of the stick-slip band and the impact vibration band in the frequency domain. These two bands are crucial indicators of downhole dynamics; the stick-slip band corresponds to intermittent sticking and slippage between the drill string and the wellbore, while the impact vibration band corresponds to periodic impacts between the drill bit or drill string and the wellbore. By identifying these two frequency bands in the spectral distribution and calculating their amplitude and energy intensity, quantitative dynamic indicators can be formed. These indicators not only reflect the stability of the operating conditions but also indirectly infer whether trajectory control is affected by abnormal operating conditions.

[0037] Furthermore, to ensure the accuracy and robustness of the state estimation, the construction method of the nonlinear state estimation model also needs to be explained. This model is implemented by establishing nonlinear observation equations and state transition equations between the observation sequence and the state variables. The observation equations describe the nonlinear mapping relationship between observed quantities (e.g., wellbore inclination angle, torque, etc.) and internal state variables, while the state transition equations characterize the dynamic evolution of state variables over time. In practical calculations, traditional linear filtering methods cannot effectively handle nonlinear relationships. Therefore, this implementation introduces the Unscented Kalman Filter (UKF) algorithm or the Extended Kalman Filter (EKF) algorithm into the solution framework. The Unscented Kalman Filter avoids linearization errors through sampling point propagation and can provide high estimation accuracy in nonlinear scenarios; the Extended Kalman Filter approximates the nonlinear solution through a first-order Taylor expansion, with relatively low computational cost, making it suitable for operating conditions with high real-time requirements. Both methods can be embedded into the nonlinear state estimation model, and the appropriate algorithm can be selected according to actual needs. The nonlinear state estimation model constructed in this way can output state estimation results containing wellbore attitude parameters and dynamic indices after multiple rounds of iterative calculations following the input observation sequence.

[0038] Specifically, let the state variable vector be... ,in, The well inclination angle; It is the azimuth angle; For spatial curvature; This represents the rate of change in drilling pressure. Let be the rate of change of torque; T be the matrix transpose. The state transition equation is defined as: in, For control input (such as speed setting); Process noise.

[0039] in, It is a nonlinear function defined by empirical formulas in drill string mechanics; The sampling interval; This is process noise.

[0040] Furthermore, the observation vector is defined as: The superscript "obs" represents the attitude and operational data acquired by the sensors. The observation equation is defined as: in, ; To observe noise.

[0041] Taking the extended Kalman filter solution process as an example, the state estimation steps include: 1. Prediction, and the corresponding prediction rules are: in, Q is the process noise covariance matrix.

[0042] 2. Update, the corresponding update rules are as follows: in, R is the observation noise covariance matrix.

[0043] It should be noted that, in frequency domain analysis, in addition to Fast Fourier Transform (FFT), a sliding window mechanism can be introduced. This involves performing piecewise FFT analysis within a certain time window, which can better capture the dynamic changes of stick-slip or shock vibrations, rather than simply obtaining a global spectral distribution. By calculating the amplitude and energy intensity within the identified stick-slip and shock vibration frequency bands and continuously tracking them over time, more stable and dynamically sensitive dynamic indicators can be formed. These indicators, combined with wellbore attitude parameters, are stored to constitute the final state estimation result.

[0044] Step S3: Construct a constrained optimization model based on the state estimation results and the preset trajectory target, and execute model predictive control in the constrained optimization model to generate a sequence of control quantities for driving the rotary guide tool.

[0045] Specifically, constructing a constrained optimization model based on the state estimation results and the preset trajectory target includes: comparing the wellbore attitude parameters and dynamic indices in the state estimation results with the preset trajectory target to determine the current trajectory deviation and stability constraints; establishing a set of nonlinear constraints, including spatial curvature constraints, build-up rate constraints, and collision resistance constraints, based on the trajectory deviation and stability constraints; transforming the preset trajectory target into an objective function, and combining the objective function with the set of nonlinear constraints to form an optimization solution structure; and calling a rolling time-domain model predictive control algorithm in the optimization solution structure to obtain the constrained optimization model.

[0046] Furthermore, model predictive control is performed in the constrained optimization model to generate a sequence of control variables for driving the rotary steerable tool, including: predicting the future wellbore attitude change trend based on the state estimation results in the rolling time domain; calculating the deviation between the future wellbore attitude change trend and the preset trajectory target to obtain a trajectory deviation amount for optimization; iteratively solving the objective function in the constrained optimization model under the trajectory deviation amount constraint to obtain a combination of control variables in multiple candidate solution spaces; selecting a solution from the candidate solution space that simultaneously satisfies the spatial curvature constraint, build-up rate constraint, and anti-collision constraint, and outputting a sequence of control variables consisting of deflection plate angle, phase parameter, and drilling pressure and rotation speed setpoints.

[0047] In this embodiment of the invention, based on the state estimation results and the preset trajectory target, a constraint optimization model needs to be further constructed to achieve closed-loop control of the drilling trajectory. The core of this constraint optimization model is to transform the trajectory control problem into a mathematical optimization problem. In this optimization problem, on the one hand, the trajectory deviation must be minimized through the objective function, and on the other hand, various physical constraints must be satisfied. In this way, the optimized output can meet the accuracy requirements of trajectory control without exceeding the safety boundaries of downhole mechanical and geometric constraints.

[0048] Specifically, the first step is to compare the state estimation results with the preset trajectory target. The state estimation results include wellbore attitude parameters and dynamic indices. Wellbore attitude parameters include inclination angle, azimuth angle, and spatial curvature, which reflect the geometric shape of the wellbore in three-dimensional space. Dynamic indices are obtained through frequency domain analysis, including the intensity values ​​of stick-slip vibration and impact vibration, which characterize the dynamic stability of the drill string during drilling. The preset trajectory target is the ideal wellbore path designed before construction, which is usually defined with well depth as the independent variable, defining the target inclination angle, target azimuth angle, and preset trajectory curvature. By comparing the actual state with the target trajectory point by point, the trajectory deviation can be obtained. The trajectory deviation here is not only a geometric deviation but can also be considered in conjunction with the dynamic indices. For example, when the stick-slip index exceeds a certain threshold, the trajectory, even with a small geometric deviation, will be considered unstable. Therefore, the trajectory deviation and stability constraint are obtained simultaneously during the comparison process.

[0049] Based on this, a set of nonlinear constraints needs to be established. This set includes at least three aspects: first, spatial curvature constraints, meaning that the curvature of the wellbore at any depth must not exceed a preset limit, otherwise it may cause excessive bending of the drill string or even stuck pipe; second, build-up rate constraints, meaning that the rate of change of the wellbore trajectory slope cannot be too large, to ensure that the directional tool can adjust its trajectory within a reasonable range without causing drastic attitude jumps; and third, anti-collision constraints, meaning that in multi-well group development scenarios, a certain safe distance must be maintained between wellbores to avoid trajectory intersections or wellbore collisions. All of the above constraints are nonlinear constraints, and their mathematical form can be written as inequalities. For example, the spatial curvature constraint can be written as... The slope constraint can be written as Collision resistance constraints can be written as ,in The spatial curvature of the wellbore at a certain depth; This is the preset upper limit value for spatial curvature; The build-up rate is the rate of change of the wellbore trajectory dip angle θ relative to the well depth L. The preset upper limit of the slope rate In a multi-well group scenario, the spatial distance between any two wells i and j at position x is given. This is the preset safety distance threshold.

[0050] Furthermore, the preset trajectory target is transformed into an objective function. Generally, the objective function is defined as the sum of squares of the trajectory deviations, i.e. ,in These represent the deviations in inclination angle, azimuth angle, and curvature, respectively. Of course, in some embodiments, a weighting term for dynamic stability can be added to the objective function, allowing the optimization process to simultaneously consider trajectory accuracy and operational stability. Ultimately, the objective function, combined with the aforementioned set of nonlinear constraints, forms a complete optimization solution structure, the so-called constrained optimization model.

[0051] After the constrained optimization model is constructed, the rolling time model predictive control (MPC) algorithm needs to be invoked to perform the optimization solution. Rolling time refers to considering only the trajectory and state evolution within a finite prediction time domain in each control cycle, optimizing the control variables within that prediction time domain, and finally executing only the first control action. The prediction window is then updated and the optimization is repeated in the next control cycle. The advantage of this method is that it can correct the trajectory in real time in uncertain environments, exhibiting strong adaptability.

[0052] Specifically, within the rolling time domain, the future wellbore attitude change trend is first predicted based on the state estimation results. The prediction process incorporates the state transition equation, substituting the current inclination angle, azimuth angle, and curvature into the equation to obtain the wellbore trajectory evolution results for the next few steps. Subsequently, the deviation between the predicted future wellbore attitude change trend and the preset trajectory target is calculated point-by-point to obtain the trajectory deviation in the prediction time domain. This trajectory deviation is one of the inputs for subsequent optimization solutions.

[0053] After obtaining the trajectory deviation, the iterative solution process for constraint optimization begins. At this point, the objective function consists of the trajectory deviation within the prediction time domain, and the constraints comprise spatial curvature constraints, build-up rate constraints, and collision resistance constraints. Iterative optimization methods (such as sequential quadratic programming (SQP) or interior-point methods) are used to search for combinations of control variables that satisfy the constraints in the candidate solution space. The control variables mainly include deflection plate angles, tool phase, and setpoints related to drilling pressure and rotational speed. These variables are key parameters that directly affect the rotary steerable tool.

[0054] After iterative solving, multiple candidate solution combinations can be obtained. At this point, a selection process is needed, ultimately choosing the solution that minimizes the objective function while satisfying all nonlinear constraints as the optimized output. The final output control sequence includes deflector angle, phase parameter, drill pressure setpoint, and rotational speed setpoint. These control sequences are then sent to the rotary steering tool, which can then drive the tool to adjust its trajectory according to the optimized instructions.

[0055] Step S4: Apply the control sequence to the rotary steering tool to perform trajectory adjustment, and perform residual calculation and model parameter update in real time based on the execution feedback information and corresponding state estimation results until drilling is completed.

[0056] Specifically, the control sequence is applied to the rotary steerable tool to perform trajectory adjustment, and residual calculation and model parameter updates are performed in real time based on execution feedback information and corresponding state estimation results. This includes: sending the deflection plate angle and phase parameters in the control sequence to the deflection drive component of the rotary steerable tool, and simultaneously inputting the drilling pressure and rotation speed setpoints to the surface control system to adjust the drilling pressure actuator and the rotary drive device; after the control sequence is applied, the wellbore attitude data and operating data returned by the rotary steerable tool are collected in real time and compared with the state estimation results to calculate the residual; when the residual is lower than a preset residual threshold, the original control cycle is maintained; when the residual exceeds the preset residual threshold, model parameter updates and control cycle adjustments are triggered, and a new control sequence is generated to be applied to the rotary steerable tool again.

[0057] Furthermore, when the residual exceeds a preset residual threshold, model parameter updates and control cycle adjustments are triggered, and a new control quantity sequence is generated and applied to the rotary steering tool again. This includes: comparing the residual with the preset residual threshold; if the residual exceeds the preset residual threshold, calling the surrogate model used for constraint optimization and correcting the friction coefficient and formation stiffness parameters based on the residual change to form updated model parameters; adjusting the control cycle according to the magnitude of the residual exceeding the preset residual threshold, with a shorter control cycle for a larger excess; and re-executing model predictive control under the updated model parameters and new control cycle configuration to generate a new control quantity sequence and apply it to the rotary steering tool again.

[0058] In practical implementation, the control sequence output by the aforementioned constraint optimization model needs to be applied to the rotary steerable tool to drive the drill bit attitude and drilling parameters to adjust the trajectory according to the optimized instructions. This control sequence, obtained through model predictive control, contains several key variables, including deflection angle, phase parameters, drilling pressure setpoint, and rotational speed setpoint. These quantities correspond to the direct control methods of the rotary steerable tool and downhole dynamics, respectively, and form the basis for achieving dynamic correction of the wellbore trajectory.

[0059] Specifically, the deflection angle and phase parameters from the control sequence are first sent to the deflection drive assembly of the rotary steerable tool, enabling the downhole tool to adjust its deflection direction and amplitude according to the instructions, thereby correcting the drilling direction of the drill bit. Simultaneously, the drill pressure and rotational speed setpoints are input to the surface control terminal. By adjusting the drill pressure actuator and the rotary drive, the drill bit is maintained at the set loading state during drilling. Through this method, the downhole rotary steerable tool and the surface transmission device work together, allowing trajectory adjustments to be gradually implemented into the actual drilling process according to the calculation results of the optimization model.

[0060] After the control sequence is completed, it is necessary to promptly obtain real-time feedback information from the rotary steerable tool. This feedback information includes wellbore attitude data and bottomhole combined operation data, consistent with the parameters in the aforementioned state estimation results. For example, attitude data includes inclination angle, azimuth angle, and curvature, while operation data includes drilling pressure, torque, and rotational speed. This data is returned to the upper-level processing stage at the end of the control cycle for comparison with the current state estimation results. By calculating the difference between the feedback data and the state estimation results, the residual can be obtained. The residual here means "the difference between prediction and reality," reflecting the accuracy of the model's description of the downhole state and representing the real-time deviation level of trajectory control.

[0061] Next, the control strategy needs to be determined based on a comparison between the residuals and a preset threshold. If the residuals are lower than the preset residual threshold, it indicates that the model predictions and actual feedback are basically consistent. In this case, the original control cycle can be maintained, and the new control sequence can continue to be generated using the original logic without additional model updates. If the residuals exceed the preset residual threshold, it indicates a significant deviation between the trajectory prediction and the actual situation. This deviation may originate from abrupt changes in formation conditions, abnormal drill string dynamics (such as stick-slip or impact), or tool parameter mismatch. In this case, it is necessary to trigger model parameter updates and control cycle adjustments to prevent further accumulation of trajectory deviations.

[0062] In scenarios triggering updates, a surrogate model for constraint optimization is first invoked. The surrogate model can be understood as a fast approximation of the constraint optimization model; it structurally retains spatial curvature constraints, build-up rate constraints, and collision resistance constraints, but allows dynamic adjustment at the parameter level. Specifically, the residuals are compared to thresholds. If the residuals exceed the threshold, the friction coefficient and formation stiffness parameters in the surrogate model are adjusted based on the change in residuals. The friction coefficient is a crucial factor influencing drill string dynamics, directly affecting downhole mechanical stability; the formation stiffness parameter affects the relationship between wellbore attitude changes and drill bit stress. By adjusting these two parameters, the surrogate model can more closely approximate actual downhole conditions, preventing the continuous expansion of prediction bias.

[0063] After the model parameters are corrected, the control cycle needs to be adjusted based on the extent to which the residuals exceed the threshold. The logic here is: the larger the residuals, the more severe the deviation, and the more frequently the model needs to be updated; therefore, the control cycle must be shortened. For example, when the residuals slightly exceed the threshold, the prediction step size can be appropriately shortened; while when the residuals significantly exceed the threshold, both the prediction step size and the control step size need to be shortened simultaneously to generate new control input sequences more quickly. In this way, trajectory control can dynamically adapt to different levels of operating condition deviations, thereby maintaining stability.

[0064] Finally, with updated model parameters and a new control cycle configuration, model predictive control is re-executed. The new MPC optimization process predicts future trajectory evolution based on the revised parameter set and iteratively solves the objective function under nonlinear constraints. The resulting new control sequence is more consistent with actual operating conditions than the original sequence, incorporating revised deflector angles, phase parameters, and drill pressure and rotation speed setpoints. This new control sequence is then applied to the rotary steerable tool, driving trajectory adjustment. As this closed-loop process is repeated, trajectory control continuously corrects the deviation between the model and reality in each cycle, ensuring the wellbore trajectory closely follows the preset target.

[0065] like Figure 2As shown, this invention provides a high-precision intelligent control device for rotary steering trajectories in drilling. The device includes: an acquisition unit, used to acquire multi-source observation data including wellbore attitude data and bottom hole combined operation data during the target drilling process, and construct an observation sequence to characterize the drilling state based on the multi-source observation data; an estimation unit, used to perform state estimation based on the observation sequence, generating a state estimation result including wellbore attitude parameters and dynamic indices; a scheme generation unit, used to construct a constrained optimization model based on the state estimation result and a preset trajectory target, and execute model predictive control in the constrained optimization model to generate a control quantity sequence for driving the rotary steering tool; and an execution unit, used to apply the control quantity sequence to the rotary steering tool to perform trajectory adjustment, and perform residual calculation and model parameter update in real time based on execution feedback information and corresponding state estimation results until drilling is completed.

[0066] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a microcontroller, chip, or processor to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0067] The optional embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details described above. Within the scope of the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical solutions of the embodiments of the present invention, and these simple modifications all fall within the protection scope of the embodiments of the present invention. It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the embodiments of the present invention will not further describe the various possible combinations.

[0068] Furthermore, various different embodiments of the present invention can be combined in any way, as long as they do not violate the spirit of the embodiments of the present invention, they should also be regarded as the content disclosed by the embodiments of the present invention.

Claims

1. A high-precision intelligent control method for drilling rotary steering trajectory, characterized in that, The method includes: Acquire multi-source observation data, including wellbore attitude data and bottom hole combined operation data, during the target drilling process, and construct an observation sequence to characterize the drilling state based on the multi-source observation data; Based on the observation sequence, state estimation is performed to generate state estimation results that include wellbore attitude parameters and dynamic indices; Based on the state estimation results and the preset trajectory target, a constraint optimization model is constructed, and model predictive control is executed in the constraint optimization model to generate a sequence of control quantities for driving the rotary guide tool; The control sequence is applied to the rotary steering tool to perform trajectory adjustment, and residual calculation and model parameter updates are performed in real time based on the execution feedback information and corresponding state estimation results until drilling is completed.

2. The high-precision intelligent control method for drilling rotary steering trajectory according to claim 1, characterized in that, The wellbore attitude data includes inclination angle, azimuth angle, and spatial curvature parameters; The bottom hole combined operation data includes drilling pressure, rotation speed, and torque; Based on the multi-source observation data, an observation sequence for characterizing the drilling status is constructed, including: The timestamps of the multi-source observation data are uniformly aligned, and interpolation resampling is performed based on the aligned data to obtain a synchronization sequence; outlier removal and filtering are performed on the synchronization sequence, and corresponding attitude denoising sequences and operation denoising sequences are obtained based on the processed wellbore attitude data and bottom hole combined operation data, respectively. Based on the attitude denoising sequence, the rate of change of well inclination angle, azimuth angle and spatial curvature are extracted as the first feature; based on the operation denoising sequence, the rotational speed, torque and mechanical power are extracted as the second feature. The first and second features are normalized and then connected in chronological order to form an observation sequence for characterizing the drilling status.

3. The high-precision intelligent control method for drilling rotary steering trajectory according to claim 1, characterized in that, Based on the observation sequence, state estimation is performed to generate state estimation results containing wellbore attitude parameters and dynamic indices, including: The observation sequence is input into a nonlinear state estimation model to extract the sequence components related to the wellbore attitude and solve for the inclination angle, azimuth angle and spatial curvature to obtain the wellbore attitude parameters. Simultaneously, sequence components related to the combined operation at the bottom of the well are extracted and frequency domain analysis is performed to identify the stick-slip frequency band and the shock vibration frequency band and calculate the corresponding intensity values ​​to form dynamic indicators; The wellbore attitude parameters and the dynamic indices are combined and stored as state estimation results.

4. The high-precision intelligent control method for drilling rotary steering trajectory according to claim 3, characterized in that, The nonlinear state estimation model is constructed by establishing nonlinear observation equations and state transition equations between the observation sequence and state variables, and incorporating the unscented Kalman filter algorithm or the extended Kalman filter algorithm into the equation solving framework. It is used to output state estimation results containing wellbore attitude parameters and dynamic indices during the state estimation process.

5. The high-precision intelligent control method for drilling rotary steering trajectory according to claim 3, characterized in that, Extract sequence components related to bottom hole combined operation and perform frequency domain analysis to identify stick-slip frequency bands and shock vibration frequency bands, calculate corresponding intensity values, and form dynamic indices, including: Extract sequence components related to bottom hole combined operation from the observation sequence; The spectral distribution is obtained by performing a Fast Fourier Transform on the extracted sequence components; In the spectral distribution, stick-slip bands and shock vibration bands are identified, and amplitudes and energy intensities are calculated within the identified bands to form dynamic indices for the state estimation results.

6. The high-precision intelligent control method for drilling rotary steering trajectory according to claim 1, characterized in that, Based on the state estimation results and the preset trajectory target, a constrained optimization model is constructed, including: The wellbore attitude parameters and dynamic indices in the state estimation results are compared with the preset trajectory target to determine the current trajectory deviation and stability constraint. Based on the trajectory deviation and stability constraints, a set of nonlinear constraints is established, including spatial curvature constraints, slope constraint, and collision resistance constraints. The trajectory objective is transformed into an objective function, and the objective function is combined with the set of nonlinear constraints to form an optimization solution structure; The rolling time domain model predictive control algorithm is invoked in the optimization solution structure to obtain the constrained optimization model.

7. The high-precision intelligent control method for drilling rotary steering trajectory according to claim 1, characterized in that, Model predictive control is performed in the constrained optimization model to generate a sequence of control quantities for driving the rotary steering tool, including: Predict the future wellbore attitude change trend based on the state estimation results within the rolling time domain; The deviation between the future wellbore attitude change trend and the preset trajectory target is calculated to obtain the trajectory deviation amount used for optimization solution; The objective function in the constrained optimization model is iteratively solved under the constraint of the trajectory deviation to obtain the combination of control variables in multiple candidate solution spaces; Select a solution from the candidate solution space that simultaneously satisfies the spatial curvature constraint, the build-up rate constraint, and the anti-collision constraint, and output a sequence of control quantities consisting of the deflection angle, phase parameter, and the setpoints for drilling pressure and rotation speed.

8. The high-precision intelligent control method for drilling rotary steering trajectory according to claim 7, characterized in that, The control sequence is applied to the rotary guide tool to perform trajectory adjustment, and residual calculation and model parameter updates are performed in real time based on execution feedback information and corresponding state estimation results, including: The deflection angle and phase parameters in the control sequence are sent to the deflection drive assembly of the rotary guide tool, while the drilling pressure and rotation speed setpoints are input to the ground control system to adjust the drilling pressure actuator and the rotary drive device. After the control sequence is completed, the wellbore attitude data and operation data returned by the rotary steering tool are collected in real time and compared with the state estimation results to calculate the residual. When the residual is lower than the preset residual threshold, the original control cycle is maintained. When the residual exceeds the preset residual threshold, the model parameters are updated and the control cycle is adjusted, and a new control quantity sequence is generated to act on the rotary guide tool again.

9. The high-precision intelligent control method for drilling rotary steering trajectory according to claim 8, characterized in that, When the residual exceeds a preset residual threshold, model parameter updates and control cycle adjustments are triggered, and a new control quantity sequence is generated to act on the rotary guide tool again, including: The residual is compared with the preset residual threshold. If the residual exceeds the preset residual threshold, the surrogate model used for constraint optimization is invoked and the friction coefficient and formation stiffness parameters therein are corrected based on the residual change to form updated model parameters. The control cycle is adjusted according to the magnitude by which the residual exceeds the preset residual threshold; the greater the magnitude of the exceedance, the shorter the control cycle. Model predictive control is re-executed with updated model parameters and a new control cycle configuration to generate a new sequence of control variables and apply them to the rotary guide tool again.

10. A high-precision intelligent control device for drilling rotary steering trajectory, characterized in that, The device includes: The acquisition unit is used to acquire multi-source observation data, including wellbore attitude data and bottom hole combined operation data, during the target drilling process, and to construct an observation sequence to characterize the drilling state based on the multi-source observation data. The estimation unit is used to perform state estimation based on the observation sequence and generate state estimation results including wellbore attitude parameters and dynamic indices; The scheme generation unit is used to construct a constrained optimization model based on the state estimation results and the preset trajectory target, and to execute model predictive control in the constrained optimization model to generate a sequence of control quantities for driving the rotary guide tool; The execution unit is used to apply the control sequence to the rotary steering tool to perform trajectory adjustment, and to perform residual calculation and model parameter update in real time based on execution feedback information and corresponding state estimation results until drilling is completed.

Images