Classification Method for Sparse Characterization of Tool Face State in Horizontal Well Drill String Torsional System

Abstract

This invention discloses a sparse representation classification method for tool face status in a horizontal well drill string torsion system, relating to the field of intelligent prediction technology. The method includes the following steps: establishing a basis function library for sparse representation; acquiring tool face angle time-series signals under different operating conditions; performing sparse decomposition based on the basis function library and the tool face angle time-series signals to obtain preliminary sparse coefficients; selecting basis functions with variance greater than a threshold as sensitive basis functions, and all sensitive basis functions for the current operating condition constitute the discriminant subspace for the current operating condition; taking the mean of the sparse coefficients of the tool face angle time-series signal for the current operating condition in the discriminant subspace as the centroid of the operating condition; repeating the aforementioned operations to obtain the centroids of multiple operating conditions and constructing a typical dictionary; this typical dictionary can be used for operating condition classification. This invention enables real-time identification and status indication of sliding steerable drilling conditions, providing reliable data support for wellbore trajectory control, slip efficiency optimization, and abnormal vibration early warning.

Description

Technical Field

[0001] This invention relates to the field of intelligent classification technology, specifically to a sparse characterization and classification method for the tool face state of a horizontal well drill string torsion swing system during drilling. Background Technology

[0002] In the development of unconventional energy sources such as shale gas and tight gas, long horizontal well sections are widely used, and the drill string exhibits significant torsional vibration and swaying phenomena within these horizontal sections. The tool face angle is a key control variable in sliding steerable drilling, and its dynamic changes reflect the coupled behavior between the drill string's torsional response, friction state, and steerable operation. However, under the influence of swaying and vibration, the actual downhole tool face will deviate significantly from the surface setting. Therefore, it is necessary to characterize the dynamic behavior of the tool face to provide a basis for operating condition identification, vibration suppression, and trajectory control.

[0003] In steerable drilling while drilling, the tool face angle is a key parameter controlling the wellbore trajectory direction, and its measurement accuracy, filtering, and control strategies have always been the focus of related research. However, existing studies have mostly focused on the numerical accuracy and dynamic control performance of the tool face angle, while paying insufficient attention to the structural operating condition information inherent in the time-domain waveform of the tool face angle itself.

[0004] In engineering practice, various near-bit vibration and toolface monitoring systems have been developed, capable of real-time acquisition of torsional vibration, toolface angle, and related dynamic data, which can be used for surface alarms and parameter optimization. Meanwhile, a series of deep learning-based analysis methods, such as Convolutional Neural Networks (CNN) and Long Short-Term Memory Networks (LSTM), have been proposed in the field of drilling condition identification both domestically and internationally, enabling some fault identification and operational condition classification. However, these methods generally rely on large amounts of labeled data, have complex model structures, and weak interpretability. Therefore, their applicability remains limited in environments with restricted downhole transmission and limited small-sample experimental data.

[0005] While existing research has made some progress in statistical feature extraction, frequency domain analysis, and deep learning-based automatic classification of toolface signals, the modeling and interpretation mechanisms for the waveform structural features of toolface signals themselves remain significantly insufficient. Current methods largely rely on macroscopic indicators such as energy and spectrum, or use difficult-to-interpret black-box models for category differentiation, failing to establish a clear correspondence between toolface waveforms and actual downhole conditions. Furthermore, current technology has not yet formed a reusable and scalable "typical waveform library" system, cannot summarize the inherent waveform patterns of different working conditions at the structural level, and struggles to maintain stable performance under conditions of strong noise, small samples, or even real-time recognition.

[0006] In real-world drilling environments characterized by complex well conditions, fluctuating operating conditions, and high noise levels, relying solely on traditional statistical or black-box methods is insufficient to meet the combined demands for real-time performance, interpretability, and cross-scenario adaptability. Therefore, establishing a structured, typical waveform modeling and classification method for toolface signals, mapping signal structural features to operating condition semantics, can significantly improve the efficiency and accuracy of operating condition identification while effectively reducing reliance on large amounts of labeled samples. By constructing a "typical waveform library" and covering a wider range of operating conditions with a small amount of sample data, a potentially valuable and efficient solution can be provided for downhole fault identification, operating condition classification, and traditional toolface monitoring systems. Summary of the Invention

[0007] To address at least one of the aforementioned problems, this invention provides a method for classifying the sparsity of the tool face state during drilling in a horizontal well drill string torsion system.

[0008] The technical solution of this invention to solve the above problems is as follows: A method for classifying the sparsity of the tool face state while drilling in a horizontal well drill string torsional yaw system, comprising the following steps:

[0009] S1. Establish a basis function library for sparse representation, wherein the basis function library contains basis functions for handling trend terms, periodic oscillation terms, decay terms, abrupt change terms, and non-smooth edge structure terms;

[0010] S2. Collect multiple tool face angle test data under different working conditions and preprocess them to obtain multiple tool face angle time series signals under different working conditions;

[0011] S3. Based on the basis function library and combined with the characteristics of the tool face angle time series signal, the corresponding basis function is used to perform sparse decomposition on the tool face angle time series signal to obtain the preliminary sparse coefficients.

[0012] S4. Based on the changes of the basis functions under the current working conditions, the basis functions with variance greater than the threshold are selected as sensitive basis functions. All sensitive basis functions under the current working conditions constitute the discrimination subspace of the current working conditions.

[0013] S5. Take the average of the sparse coefficients of the multiple tool face angle time series signals of the current working condition in the discrimination subspace, and use this average as the working condition centroid of the current working condition.

[0014] S6. Repeat S3-S5 to obtain the centroids of multiple working conditions and construct a typical dictionary;

[0015] S7. Take the tool surface signal to be identified, and obtain the target sparse coefficient of the tool surface signal according to the operations of S1~S3. Calculate the Euclidean distance between the target sparse coefficient in the discrimination subspace and the centroid of the working condition in the typical dictionary. Based on the minimum distance principle, obtain the classification of the tool surface signal to be identified.

[0016] The beneficial effects of this invention are as follows: By employing the tool face signal state classification method proposed in this invention, the accuracy and stability of downhole working condition identification can be significantly improved. The constructed typical waveform dictionary exhibits good structural discrimination ability under different working conditions, enabling the Euclidean distance discrimination of sparse coefficients to truly reflect the inherent differences between various working conditions. In field environments with small samples, strong noise interference, and unstable signal acquisition conditions, this invention can still maintain reliable identification performance, overcoming the limitations of traditional methods that are susceptible to noise and local disturbances leading to misjudgment.

[0017] The discriminant subspace verification mechanism introduced in this invention further improves the reliability of the final identification result. This mechanism effectively suppresses random deviations caused by factors such as noise, sampling differences, and transient disturbances by projecting and verifying the sparse coefficients, making the working condition identification process robust and reliable.

[0018] In summary, this invention enables real-time identification and status alerts for sliding steerable drilling conditions in practical applications, providing reliable data support for wellbore trajectory control, slip efficiency optimization, and abnormal vibration early warning, and has significant engineering application value and promising prospects for promotion. Attached Figure Description

[0019] Figure 1 This is a flowchart of the method according to an embodiment of the present invention;

[0020] Figure 2 This is a schematic diagram of the tool face angle testing equipment.

[0021] Figure 3 To stabilize the tool surface, the first set of original signals and the resampled signals are used;

[0022] Figure 4 To stabilize the second set of original signals and resampled signals on the tool surface;

[0023] Figure 5 The third set of original signals and resampled signals for stabilizing the tool surface;

[0024] Figure 6 Sensitive basis function selection plot for stable tool surfaces;

[0025] Figure 7 These are the three sets of resampled signals from the impact tool surface;

[0026] Figure 8 A comparison of typical and original waveforms for stable tool face conditions and impact tool face conditions. Detailed Implementation

[0027] The specific embodiments of the present invention will be clearly and completely described below with reference to examples. Obviously, the described examples are only some embodiments of the present invention, and not all embodiments.

[0028] like Figure 1 As shown, the sparse characterization and classification method for tool face state of a horizontal well drill string torsional yaw system includes the following steps:

[0029] S1. Establish a basis function library for sparse representation, wherein the basis function library contains basis functions for handling trend terms, periodic oscillation terms, decay terms, abrupt change terms, and non-smooth edge structure terms;

[0030] Based on extensive practical experience, the inventors discovered that existing tool face angle time series signals simultaneously contain multiple structural features such as trend changes, steady-state rotations, damped vibrations, local abrupt changes, and non-smooth edges. Using only a single type of basis function makes it difficult to effectively represent these signals. Therefore, based on the dynamic mechanism of tool faces, the inventors constructed a basis function library containing multiple types of basis functions for the multi-structured and interpretable representation of tool face angle time series signals.

[0031] For the five structural features mentioned above, the inventors have prepared multiple basis functions that meet the requirements in the basis function library.

[0032] The trend change term in the tool face angle time series signal can be processed by polynomial basis functions. Polynomial basis functions are mainly used to describe the overall trend and slowly changing smooth components of the signal, including constant terms, first-order and higher-order polynomials, etc., which can simulate the slowly changing trend caused by uneven drill string stress, friction accumulation and system bias.

[0033] The steady-state rotation term in the tool face angle time series signal can be processed using Fourier sine / cosine basis functions. Fourier sine / cosine basis functions are used to characterize steady-state rotation, torsion, and their harmonic components. They are the main source of periodic oscillation structures in the tool face signal and can reflect dynamic behaviors such as downhole rotation speed changes and periodic torsional vibrations.

[0034] The decaying vibration term in the tool face angle time series signal can be processed by the exponential decay basis function. The exponential decay basis function is used to capture the energy decay effect generated by the drill string after being subjected to impact, friction or collision. Its decay rate is obtained by discretization of multiple sets of parameters, which can simulate different physical processes from rapid decay to slow decay.

[0035] Local mutation terms in the tool face angle time series signal can be processed by ReLU-type piecewise activation basis functions. ReLU-type piecewise activation basis functions are used to represent local mutations, jumps, and breakpoint structures, and can reflect the transient jump characteristics of the tool face angle when it encounters jamming, instability, or short-term disturbances.

[0036] For non-smooth edges in tool face angle time series signals, one of the following can be used: sawtooth wave basis function, square wave basis function, and triangular wave basis function. These three basis functions are used to describe asymmetric abrupt changes, non-smooth edges, and non-sinusoidal periodic structures, which can supplement the nonlinear periodic behavior that is difficult to characterize by traditional Fourier basis functions.

[0037] Unlike traditional sparse representation methods that focus on using mathematical waveform atoms, the enhanced basis function library constructed in this invention is derived from the real physical components of the tool face signal. It is built upon five structural categories—trend, periodicity, decay, abrupt change, and non-smooth edge—as its basic construction principle, ensuring a one-to-one correspondence between basis functions and downhole dynamic events. This physical mechanism-driven dictionary design significantly improves the interpretability of the representation, allowing the sparse coefficients obtained after sparse representation to not only reflect the mathematical decomposition of the signal but also directly characterize the strength of structural components under different operating conditions. This facilitates the establishment of a stable and highly generalizable typical waveform library and operating condition feature space. The construction of the basis function library enables accurate, stable, and physically meaningful sparse representations of tool face signals with a finite number of structural basis functions, providing a solid foundation for subsequent operating condition classification, waveform reconstruction, and downhole condition identification.

[0038] S2. Collect multiple tool face angle test data under different working conditions and preprocess them to obtain multiple tool face angle time series signals under different working conditions;

[0039] During the testing of the tool face angle, an attitude sensor is typically installed on the drill string. This attitude sensor can acquire the acceleration of the drill string along the Y and Z axes. Subsequently, the tool face angle is calculated using the following formula:

[0040] In the formula, The tool face angle is represented by Accy, the y-axis acceleration is represented by Accz, and the z-axis acceleration is represented by Accz.

[0041] However, during the data collection process for the tool face angle, anomalies were observed: in horizontal wells, the drill string is affected by factors such as friction, torsional vibration, and periodic oscillation, causing the actual tool face angle to change continuously. However, when the angle crosses the sensor's measurement range boundary, it reverses, resulting in an unrealistic jump. Therefore, considering the actual situation, corrections are usually necessary.

[0042] In this step, to recover the true angle change trajectory, an unwinding algorithm (e.g., the unwrap function in signal processing) is used to unwind the original angle sequence. By detecting the angle difference between adjacent sampling points and setting a threshold, the turning point is automatically identified. An integer period is added to or subtracted from the position where the turning point occurs, thus obtaining a continuous, non-turning true angle sequence. The unwound signal accurately reflects the drill string dynamics and exhibits good retention of structures such as oscillation, stability holding, and torsional impact, providing reliable input for subsequent typical waveform construction.

[0043] Meanwhile, under different experimental conditions and data acquisition conditions, the sequence lengths of the original data obtained are inconsistent, making it impossible to directly perform sparse decomposition and feature extraction within a unified domain. To address this issue, this step employs linear interpolation to resample the angle sequences, uniformly mapping all signals to a fixed number of standard sampling points. This processing step ensures the consistency of the temporal structure, adjusting only the sampling density without altering the inherent signal structure, thus ensuring that differences between different experiments primarily stem from the operating conditions themselves, rather than sampling variations.

[0044] After unwinding and resampling, the tool face angle sequence suitable for subsequent sparse decomposition is finally obtained. Combined with the time series, the tool face angle time series signal is constructed.

[0045] To facilitate the establishment of the center of gravity for subsequent working conditions, the inventors discovered through extensive experimentation that at least three samples of the tool face angle time series signal should be collected for each working condition. Under these conditions, the practical application requirements can be met, demonstrating that the method of this embodiment can operate with a small number of samples. While more samples would make the final center of gravity for the working condition more representative, it would require higher costs.

[0046] The structure of the equipment for collecting tool face angle test data under different working conditions is as follows: Figure 2 As shown, the device consists of 12 parts: 1. Central control console; 2. Electrical control cabinet; 3. Motor bracket; 4. Servo motor; 5. Coupling; 6. Simulated drill string; 7. Simulated drill tool; 8. Attitude and displacement sensor; 9. Universal joint; 10. Magnetic powder brake; 11. Simulated wellbore; 12. Experimental stand.

[0047] Servo motor 4, serving as the main power source, is mounted on a motor bracket 3 with counterweights to reduce vibration during operation. Servo motor 4 is connected to the simulated drill string 6 via coupling 5 to achieve stable power transmission, driving the simulated drill tool 7 to rotate. Electrical control cabinet 2 can adjust parameters such as the speed and torque of servo motor 4 to simulate different operating conditions.

[0048] The simulated wellbore 11 is supported and fixed by a steel bracket, and its inner diameter, material, and other parameters are matched to those of the actual drilling wellbore to provide a realistic downhole environment. The simulated drill string 6 is placed inside the simulated wellbore 11, applying static and dynamic friction through line contact. A magnetic powder brake 10 connected to the end of the simulated drill string 6 precisely controls the resistance torque to reproduce complex working conditions such as drill string-wellwall friction. A universal joint 9 connects the simulated drill string 6 and the magnetic powder brake 10 to compensate for angular deviations during movement, ensuring continuous and stable power transmission.

[0049] The attitude and displacement sensor 8 is installed on the simulated drill 7, which can measure dynamic parameters such as acceleration, rotational speed and angular displacement in real time, and transmit the data to the central control console 1 wirelessly to realize real-time monitoring of the experimental process and subsequent data analysis.

[0050] S3. Based on the basis function library and combined with the characteristics of the tool face angle time series signal, the corresponding basis function is used to perform sparse decomposition on the tool face angle time series signal to obtain the preliminary sparse coefficients.

[0051] In this step, a least-squares sparse solution strategy is adopted to minimize the signal reconstruction error while suppressing the participation of redundant basis functions, so that the solved coefficients exhibit obvious sparse characteristics.

[0052] Specifically, for multiple different components in the tool face angle time series signal, corresponding basis functions are selected for sparse decomposition, thereby enabling the trend component, periodic oscillation component, decay component and local mutation component of the tool face angle time series signal to be distinguished in the coefficient space, realizing the structured expression of sparse coefficients and obtaining a parameterized expression result with physical correspondence.

[0053] In practical implementation, based on the sparse solution of the reconstruction error, the sparse coefficients satisfy the following optimization strategy: In the formula, This represents the preprocessed tool face time-domain signal, belonging to a length of The column vectors, whose physical meaning is the original dynamic behavior of the tool face angle changing with time during the drilling process; For the enhanced basis function dictionary matrix, by The group structure basis functions are composed of columns, with dimension 1. ; Let be the sparse coefficient vector to be determined, with dimension . It is used to reflect which structural components the signal is composed of, and the weight of each component; These are sparse regularization weighting coefficients used to balance "reconstruction accuracy" and "sparseness".

[0054] S4. Based on the changes of the basis functions under the current working conditions, the basis functions with variance greater than the threshold are selected as sensitive basis functions. All sensitive basis functions under the current working conditions constitute the discrimination subspace of the current working conditions.

[0055] In this step, the variance of the initial sparse coefficients obtained from multiple basis functions is first calculated row by row, and the calculation results are then statistically analyzed. In the formula, Represents the variance statistics set. Let A represent the set of vectors containing the elements of the k-th row of matrix A; matrix A represents the matrix composed of all the initial sparse coefficients under the current operating condition; subsequently, basis functions with variance greater than a threshold are selected as sensitive basis functions, and these sensitive basis functions are used as the discriminant subspace for the current operating condition. In the formula, S represents the discriminant subspace. The threshold is indicated. Based on the inventor's experience, this threshold is usually set to 0.04 to 0.06 times the maximum variance mentioned above, such as 0.05 times. Of course, those skilled in the art can set an appropriate threshold according to the actual situation.

[0056] Meanwhile, considering that there may be noise or local disturbances in the actual signal, in order to avoid excessive shrinkage of the screening results, a minimum number of base functions is set in this step. The minimum number of base functions is usually set to 10-20% of the total number of base functions, such as 5, 6, 7 or 8. When the number of base functions that meet the variance condition is too small, a small number of base functions with high variance ranking will be automatically added to maintain the expressive power of the discriminant subspace.

[0057] The sensitive basis function subset obtained through the above screening has the characteristics of "large inter-class differences and small intra-class variations". Its corresponding sparse coefficients can form more obvious differences in working condition distribution in the dimensionality-reduced feature space, thereby significantly improving the accuracy and interpretability of working condition identification.

[0058] This strategy strikes a balance between computational efficiency and sparsity. For typical operating condition identification or scenarios with relatively simple signal structures, the least squares solution method can obtain stable and structurally distinctive sparse coefficients. This solution framework can automatically enhance periodic features, suppress noise disturbances, and strengthen abrupt change edges, enabling the tool surface signal to form significant clusters with operating condition differences in the coefficient space.

[0059] S5. Take the average of the sparse coefficients of the multiple tool face angle time series signals of the current working condition in the discrimination subspace, and use this average as the working condition centroid of the current working condition.

[0060] After obtaining the discrimination subspace of the current working condition, the time series signals of multiple tool face angles of the current working condition are substituted back into the discrimination subspace for sparse decomposition, and finally multiple sparse coefficient sequences are obtained.

[0061] Specifically, for tool facet time series signals, as mentioned above, they are typically divided into five different terms: trend term, periodic oscillation term, decay term, abrupt change term, and non-smooth edge structure term. Different basis functions are required for decomposition of each term. However, in actual production, the inventors discovered that the discriminant subspace may not simultaneously contain sensitive basis functions capable of handling all five terms: trend term, periodic oscillation term, decay term, abrupt change term, and non-smooth edge structure term.

[0062] Therefore, considering the actual situation, the inventors made the following adjustments: if the discriminant subspace does not contain the basis function corresponding to a certain type of term, then that term will not participate in the coefficient update during the discrimination process; if the discriminant subspace contains the basis function corresponding to that term, then the sparse coefficients are projected or reconstructed only on the subset of basis functions to obtain consistent sparse coefficient components within the discriminant subspace; the sparse coefficient components obtained within the discriminant subspace are combined with the remaining coefficient components that did not participate in the discrimination to form a sparse coefficient vector consistent with the structure of the discriminant subspace.

[0063] Subsequently, the mean of multiple sparse coefficient sequences is calculated, which yields the centroid of the operating condition: In the formula, c represents the center of gravity of the working condition, and n represents the sample size. Let represent the sparse coefficient sequence of the nth sample.

[0064] The above-mentioned working condition centroid integrates the structural components common to all experiments, including trend changes, steady-state oscillations, decay processes and local disturbances. At the same time, it suppresses noise, disturbances or local anomalies in individual trials through multi-sample averaging, enabling it to more accurately characterize the inherent dynamic characteristics of the working condition.

[0065] Since the center of gravity of the working condition is relatively abstract, in order to make a quick preliminary judgment on the current working condition more intuitively, typical waveforms are also set in some implementation methods.

[0066] The typical waveform is obtained based on the centroid of the operating condition. After obtaining the centroid, it is projected onto the basis function library and combined with time-domain inversion to obtain a typical waveform with complete structural expression capabilities that can reflect the characteristics of the current operating condition. This operation is a routine operation in this field, so its specific construction steps will not be described in detail here.

[0067] Typical waveforms directly present the trend terms, main period structure, oscillation amplitude, local disturbances, and transient anomalies of the operating conditions in an intuitive and visual way, making the structural differences between operating conditions "visible" and "understandable." This visible representation not only facilitates rapid identification of operating conditions by on-site engineers but also provides important references for manual verification, anomaly diagnosis, trajectory control, and parameter tuning suggestions. Furthermore, typical waveforms, as a structured template, actually constitute the basic unit of the operating condition knowledge base and can serve as key inputs for future operating condition expansion, algorithm migration, model verification, and downhole anomaly identification. Therefore, even though typical waveforms are not directly used for the calculation of the discriminant formula, they remain an indispensable component of this invention in engineering applications and a crucial bridge linking sparse structural representation with actual drilling behavior.

[0068] S6. Repeat S3-S5 to obtain the centroids of multiple working conditions and construct a typical dictionary;

[0069] The typical waveforms mentioned above can also be included in the typical dictionary for easy reference by staff. In particular, for the sparse coefficients used to retrieve typical waveforms, considering the potential amplitude scale differences introduced by equipment gain, installation location, and measurement links, their amplitudes need to be normalized. Furthermore, the sampling time length, sampling frequency, or signal start point of different experiments are not entirely consistent. To eliminate these time-domain scale differences, this step maps all typical waveforms to a unified time axis and performs phase consistency correction on highly periodic operating conditions, ensuring that the typical structures of different operating conditions have comparability and a consistent time reference under the reference coordinates. Through this series of standardization steps, all operating condition templates in the typical dictionary are within a unified scale, a unified time reference, and a unified structural expression system, thus providing a reliable foundation for subsequent similarity measurement and projection analysis.

[0070] S7. Take the tool surface signal to be identified. Following the operations of S1 to S3, obtain the target sparse coefficient of the tool surface signal to be identified. Calculate the Euclidean distance between the target sparse coefficient and the centroid of the typical working condition in the dictionary. Based on the principle of minimum distance, obtain the classification of the tool surface signal to be identified.

[0071] To further illustrate the superiority of the method in the embodiments of the present invention, specific test examples are given below.

[0072] Using the device disclosed in this embodiment, relevant tests on the tool face working condition are performed. The attitude sensor collects the biaxial acceleration and calculates the initial tool face angle of the tool face working condition.

[0073] After unwinding and resampling the tool face angle, a tool face angle time series signal with 8 periods and 512 sampling points is obtained, as follows: Figures 3-5 As shown, where, Figure 3To stabilize the tool surface, the first set of original signals and the resampled signals are used; Figure 4 To stabilize the second set of original signals and resampled signals on the tool surface; Figure 5 The third set of original signals and resampled signals for stabilizing the tool surface;

[0074] The tool facet angle time series signal is sparsely decomposed using a basis function library. The variance of each basis function is obtained based on the initial sparse coefficients. Sensitive basis functions are then selected based on a threshold, such as... Figure 6 As shown.

[0075] This case study will utilize two types of toolface sample signals to invert typical waveforms. After preprocessing and basis function decomposition, the experimental samples are mean-valued along their dimensions using their complete sparse coefficients to obtain the centroid vector of the stable toolface condition. This centroid vector is then reprojected onto the complete basis function dictionary. Simultaneously, using the same method, the parameters of three sets of impact toolface conditions (see...) are... Figure 7 The tests were conducted, and the typical time-domain waveforms under the two toolface working conditions were finally obtained as follows: Figure 8 As shown, Figure 8 (a) in the figure represents a comparison between the typical waveform and the original waveform under the stable tool face condition. Figure 8 (b) in the figure represents a comparison between the typical waveform and the original waveform of the impact tool face condition.

[0076] The present invention has been disclosed above with preferred embodiments. However, those skilled in the art should understand that these embodiments are for illustrative purposes only and should not be construed as limiting the scope of the invention. Further improvements can be made without departing from the principles of the invention, and these improvements should also be considered as protections of the present invention.

Claims

1. A method for classifying the sparsity of the tool face state while drilling in a horizontal well drill string torsional yaw system, characterized in that, Includes the following steps: S1. Establish a basis function library for sparse representation, wherein the basis function library contains basis functions for handling trend terms, periodic oscillation terms, decay terms, abrupt change terms, and non-smooth edge structure terms; S2. Collect multiple tool face angle test data under different working conditions and preprocess them to obtain multiple tool face angle time series signals under different working conditions; S3. Based on the basis function library and combined with the characteristics of the tool face angle time series signal, the corresponding basis function is used to perform sparse decomposition on the tool face angle time series signal to obtain the preliminary sparse coefficients. S4. Based on the changes of the basis functions under the current working conditions, the basis functions with variance greater than the threshold are selected as sensitive basis functions. All sensitive basis functions under the current working conditions constitute the discrimination subspace of the current working conditions. S5. Take the average of the sparse coefficients of the multiple tool face angle time series signals of the current working condition in the discrimination subspace, and use this average as the working condition centroid of the current working condition. S6. Repeat S3-S5 to obtain the centroids of multiple working conditions and construct a typical dictionary; S7. Take the tool surface signal to be identified, and obtain the target sparse coefficient of the tool surface signal according to the operations of S1~S3. Calculate the Euclidean distance between the target sparse coefficient in the discrimination subspace and the centroid of the working condition in the typical dictionary. Based on the minimum distance principle, obtain the classification of the tool surface signal to be identified.

2. The method for classifying and characterizing the sparsity of the tool face state while drilling in a horizontal well drill string torsion system according to claim 1, characterized in that, In S1, the basis functions for handling the trend term are polynomial basis functions, the basis functions for handling the periodic oscillation term are Fourier sine / cosine basis functions, the basis functions for handling the decay term are exponential decay basis functions, the basis functions for handling the abrupt change term are ReLU-type piecewise activation basis functions, and the basis functions for handling the non-smooth edge structure term are sawtooth wave basis functions, square wave basis functions, and triangular wave basis functions.

3. The method for classifying and characterizing the sparsity of the tool face state while drilling in a horizontal well drill string torsion system according to claim 1, characterized in that, In S2, the structure of the device for collecting tool face angle test data under different working conditions is as follows: The device consists of a servo motor, a simulated drill string, a universal joint, and a displacement attitude sensor. The servo motor is used to control the drilling of the simulated drill string, the universal joint is used to cause the drill string to turn, and the displacement attitude sensor is used to collect the Y-axis acceleration, Z-axis acceleration, and angular velocity.

4. The method for classifying the sparsity of the tool face state while drilling in a horizontal well drill string torsion system according to claim 3, characterized in that, In S2, after the raw data is collected, the following steps are also included: S21. Calculate the angle of the tool face: In the formula, The tool face angle is represented by Accy, the y-axis acceleration is represented by Accz, and the z-axis acceleration is represented by Accz. S22. The initial tool face angle data is processed using an unwinding algorithm to obtain the actual tool face angle timing data; S23. The angle sequence is resampled using linear interpolation, and all signals are uniformly mapped to a fixed number of standard sampling points. Then, a tool face angle time series signal is established to show how the tool face angle changes over time.

5. The method for classifying and characterizing the sparsity of the tool face state in a horizontal well drill string torsion system according to claim 1, characterized in that, In S3, the trend term, periodic oscillation term, decay term, abrupt change term, and non-smooth edge structure term of the tool face angle time series signal are sparsely decomposed using the corresponding basis functions.

6. The method for classifying and characterizing the sparsity of the tool face state while drilling in a horizontal well drill string torsion system according to claim 1, characterized in that, In S4, the threshold is 0.04 to 0.06 times the maximum variance of the current operating condition.

7. The method for classifying and characterizing the sparsity of the tool face state in a horizontal well drill string torsion system according to claim 1, characterized in that, S5 also includes the following steps: after obtaining the centroid of the current working condition, project it onto the basis function library for inversion to obtain the typical waveform of the current working condition.

8. The method for classifying and characterizing the sparsity of the tool face state while drilling in a horizontal well drill string torsion system according to claim 1, characterized in that, In S5, the method for solving the sparse coefficients of the tool face angle time series signal in the discrimination subspace is as follows: The tool face angle time series signal includes a trend term, a periodic oscillation term, a decay term, abrupt change term, and a non-smooth edge structure term. If the discrimination subspace does not contain a basis function corresponding to a certain type of term, then that term does not participate in the coefficient update during the discrimination process; if the discrimination subspace contains a basis function corresponding to that term, then the sparse coefficients are projected or reconstructed only on the subset of basis functions, thereby obtaining consistent sparse coefficient components in the discrimination subspace. The sparse coefficient components obtained in the discriminant subspace are combined with the remaining coefficient components that did not participate in the discrimination to form a sparse coefficient vector consistent with the structure of the discriminant subspace.

Images