Data acquisition method and system of driver for oil exploitation

Abstract

The invention relates to the field of drilling ship manufacturing, in particular to a data acquisition method and system of a driver for oil exploitation. Six-degree-of-freedom motion data of a ship motion reference unit, pressure data of a hydraulic cylinder of a marine riser tensioner, hook load data of a drill string hung by a top driving system, torque data and radial acceleration data are synchronously collected, and collected multiple paths of signals are aggregated to form synchronous multi-source vibration data flow. According to the method, multiple types of vibration, load and pressure signals are synchronously collected and aggregated and segmented under the unified time reference in the operation process of the drilling ship, so that the consistency of multi-source data on the time scale is ensured, and the real dynamic coupling relation can be reflected by subsequent calculation; in the signal processing stage, key acceleration and pressure fluctuation information are extracted, and the actual working state of the marine riser system is represented through dynamic response characteristics constructed by the peak position, the peak amplitude and the power spectral density.

Description

Technical Field

[0001] This invention relates to the field of drilling vessel manufacturing, and more particularly to a data acquisition method and system for oil exploration actuators. Background Technology

[0002] The field of drilling vessel manufacturing technology covers the design, construction and integration of specialized vessels for offshore oil and gas resource exploration and exploitation, including hull structure design, power and propulsion system configuration, dynamic positioning system integration, drilling operation system layout, vibration and load monitoring system installation, marine environmental adaptability design, and the development of shipborne data acquisition and control systems.

[0003] While existing drilling vessel manufacturing technology encompasses structural design, propulsion systems, positioning systems, operational systems, and monitoring systems, in actual operation, under complex marine environments, environmental and operational loads change rapidly and instantaneously, easily distorting analysis results. Furthermore, if characteristic parameters are not calculated in conjunction with actual vibration modes, critical early anomaly signals may be masked. Therefore, improvements are needed. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the existing technology and to propose a data acquisition method and system for oil exploration using a drive.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a data acquisition method for an oil extraction drive, comprising the following steps: During drilling vessel operations, six-degree-of-freedom motion data of the hull motion reference unit, pressure data of the riser tensioner cylinder, drill string hook load data, torque data, and radial acceleration data of the top drive system are collected simultaneously. The collected multi-channel signals are aggregated to form a synchronous multi-source vibration data stream. The synchronous multi-source vibration data stream is then divided according to a preset time length to obtain a time-aligned coupled vibration data window. Based on the time-aligned coupled vibration data window, the vertical acceleration signal of the hull and the pressure fluctuation signal of the riser tensioner are extracted, the peak position and peak amplitude are extracted, and the power spectral density of the pressure fluctuation signal is calculated to establish a dynamic response feature set of the riser system. Based on the time-aligned coupled vibration data window, Fourier transforms are performed to generate a fractal vibration energy spectrum matrix. The power spectrum of each time window in the fractal vibration energy spectrum matrix is ​​normalized to obtain the energy probability distribution. Then, the energy probability distribution is used to calculate the top-driven multidimensional vibration energy result. Based on the dynamic response feature set of the water-proof pipe system, the joint rotational stiffness parameter in the preset vibration system characterization is adjusted to obtain the real-time system stiffness parameter. At the same time, based on the top drive multidimensional vibration energy result, the stiffness adaptive working condition anomaly index is calculated. By combining the real-time system stiffness parameter and the stiffness adaptive working condition anomaly index, the comprehensive state parameters of the drive system health and working condition are established.

[0006] Preferably, the step of acquiring the synchronous multi-source vibration data stream is as follows: Based on a unified time reference during drilling vessel operations, the six-degree-of-freedom motion data of the hull motion reference unit, the pressure data of the riser tensioner cylinder, the drill string hook load data, torque data, and radial acceleration data of the top drive system are read. The timestamps are parsed and the unit format and channel identifier are unified. The record boundaries and duplicate entries are corrected in ascending order of timestamps to obtain the acquired multi-channel signals. Based on the acquired multi-channel signals, records with the same sampling time are aligned according to the timestamp. The six-degree-of-freedom motion data of the hull motion reference unit, the pressure data of the riser tensioner cylinder, the drill string hook load data, torque data and radial acceleration data of the top drive system are spliced ​​in a fixed channel order and the source identifier and sampling timestamp are added. Missing positions are checked and kept aligned with null value marks to form a synchronous multi-source vibration data stream.

[0007] Preferably, the step of obtaining the time-aligned coupled vibration data window is as follows: Based on the synchronous multi-source vibration data stream, the start and end timestamps of continuous segments are calculated according to the preset time length. The six-degree-of-freedom motion data of the hull motion reference unit, the pressure data of the riser tensioner cylinder, the drill string hook load data, torque data and radial acceleration data of the top drive system are cut according to the start and end timestamps. If the length of the last segment is less than the preset time length, no segment is generated, forming a time-aligned coupled vibration data window.

[0008] Preferably, the steps for obtaining the dynamic response feature set of the riser system are as follows: Based on the time-aligned coupled vibration data window, the numerical sequence of the vertical acceleration channel in the hull motion reference unit and the numerical sequence of the pressure fluctuation channel in the riser tensioner cylinder are called. Continuous samples with the same start and end times are synchronously extracted according to the timestamp index to obtain the hull vertical acceleration signal and the riser tensioner pressure fluctuation signal. Based on the vertical acceleration signal of the hull and the pressure fluctuation signal of the riser tensioner, the cross-correlation numerical sequence of the two sets of signals under different time delays is calculated point by point, the time delay position and amplitude value of the corresponding response peak in the numerical sequence are located, and recorded as the peak position and peak amplitude. Based on the peak position and peak amplitude, the pressure fluctuation signal of the riser tensioner is called for frequency domain conversion and the energy distribution density of each frequency component is statistically analyzed. The energy distribution density is taken as the power spectral density of the pressure fluctuation signal. The peak position, peak amplitude and power spectral density are spliced ​​together in a fixed order to form a complete record, thus forming a dynamic response feature set of the riser system.

[0009] Preferably, the steps for obtaining the fractal vibrational energy spectrum matrix are as follows: Based on the time-aligned coupled vibration data window, the numerical sequences of the torque channel, drill string axial load channel, and radial acceleration channel in the top drive system are called. Continuous samples of the same length are segmented in a unified sampling frequency and timestamp order. Short-time Fourier transform is performed on each sample segment to extract the frequency components and energy distribution of the corresponding time period. The frequency component energy results of the three channels in the same time period are combined into matrix units according to the channel order. All matrix units are spliced ​​together according to the time period order to generate a fractal vibration energy spectrum matrix.

[0010] Preferably, the step of obtaining the top-driven multidimensional vibration energy result is as follows: Based on the fractal vibration energy spectrum matrix, the power spectrum values ​​of the torque channel, the power spectrum values ​​of the drill string axial load channel, and the power spectrum values ​​of the radial acceleration channel are extracted in each time window. The power spectrum values ​​of each channel at each frequency component are summed according to the time window to obtain the power spectrum sum. The amplitude is normalized by dividing the power spectrum value of the frequency component by the corresponding power spectrum sum to form the energy probability distribution. Based on the energy probability distribution, calculate the frequency moment weighted spectral entropy; Based on the aforementioned frequency moment weighted spectral entropy, the frequency moment weighted spectral entropy of the time window in the torque channel, the frequency moment weighted spectral entropy of the time window in the drill string axial load channel, and the frequency moment weighted spectral entropy of the time window in the radial acceleration channel are combined into a column vector. The column vectors are then aggregated sequentially according to the time window order to form the top-driven multidimensional vibration energy result.

[0011] Preferably, the steps for obtaining the real-time system stiffness parameters are as follows: Based on the dynamic response feature set of the riser system, the initial value of the joint rotational stiffness parameter characterized by the preset vibration system is read. The differences between the predicted output and the dynamic response feature set of the riser system in terms of peak position and peak amplitude are compared item by item and the sign is determined. The joint rotational stiffness parameter is adjusted according to the sign of the difference until the absolute value of the difference no longer decreases, and the real-time system stiffness parameter is obtained.

[0012] Preferably, the steps for obtaining the comprehensive health and operating condition parameters of the drive system are as follows: Calculate the stiffness adaptive working condition anomaly index based on the real-time system stiffness parameters; Based on the stiffness adaptive working condition anomaly index, the real-time system stiffness parameters and the stiffness adaptive working condition anomaly index are stored in the same time window and combined in the order of timestamps to form a sequence of joint rotation stiffness values ​​and working condition stability scores, thereby generating comprehensive health and working condition status parameters of the drive system.

[0013] This invention provides a data acquisition system, comprising: During drilling vessel operations, the data acquisition module simultaneously acquires six-degree-of-freedom motion data of the hull motion reference unit, pressure data of the riser tensioner cylinder, drill string hook load data, torque data, and radial acceleration data of the top drive system. The acquired multi-channel signals are aggregated to form a synchronous multi-source vibration data stream, which is then divided according to a preset time length to obtain a time-aligned coupled vibration data window. The dynamic response feature extraction module extracts the hull vertical acceleration signal and the riser tensioner pressure fluctuation signal based on the time-aligned coupled vibration data window, extracts the peak position and peak amplitude, calculates the power spectral density of the pressure fluctuation signal, and establishes a dynamic response feature set of the riser system. The vibration energy analysis module performs Fourier transforms on the time-aligned coupled vibration data windows to generate a fractal vibration energy spectrum matrix. It then normalizes the power spectrum of each time window within the fractal vibration energy spectrum matrix to obtain an energy probability distribution. Finally, it calculates the top-driven multidimensional vibration energy result based on the energy probability distribution. The health and operating condition assessment module, based on the dynamic response feature set of the riser system, adjusts the joint rotational stiffness parameter in the preset vibration system characterization to obtain the real-time system stiffness parameter. At the same time, based on the multi-dimensional vibration energy result of the top drive, it calculates the stiffness adaptive operating condition anomaly index. Combining the real-time system stiffness parameter and the stiffness adaptive operating condition anomaly index, it establishes the comprehensive health and operating condition status parameters of the drive system.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: This invention ensures the consistency of multi-source data across time scales by synchronously acquiring and aggregating multiple vibration, load, and pressure signals under a unified time reference during drilling vessel operations. This allows subsequent calculations to reflect the true dynamic coupling relationships. In the signal processing stage, key acceleration and pressure fluctuation information is extracted, and dynamic response features constructed from peak position, peak amplitude, and power spectral density characterize the actual working state of the riser system. In frequency domain analysis, the power spectra of different channels are normalized into energy probability distributions, and multidimensional vibration energy results are generated, reflecting both the vibration energy distribution characteristics of each channel and preserving the integrity of time-frequency information. In the parameter adjustment stage, the dynamic response features guide the correction of joint rotational stiffness parameters, ensuring a high degree of alignment between the predicted output and the actual response. Furthermore, the calculation of a stiffness-adaptive working condition anomaly index is introduced, integrating structural stiffness changes with the degree of vibration anomalies to achieve dynamic quantification of working condition stability. This not only improves the accuracy and real-time performance of parameter acquisition but also enhances sensitivity to abnormal states, enabling earlier identification of potential structural risks, ensuring the continuity and safety of drilling operations, and providing a reliable basis for optimizing operational strategies. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the steps of the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0017] Please see Figure 1 This invention provides a technical solution: a data acquisition method for an oil extraction drive, comprising the following steps: During drilling vessel operations, six-degree-of-freedom motion data of the hull motion reference unit, pressure data of the riser tensioner cylinder, drill string hook load data, torque data, and radial acceleration data of the top drive system are collected simultaneously. The collected multi-channel signals are aggregated to form a synchronous multi-source vibration data stream. The synchronous multi-source vibration data stream is then divided according to a preset time length to obtain a time-aligned coupled vibration data window. Based on a time-aligned coupled vibration data window, the vertical acceleration signal of the hull and the pressure fluctuation signal of the riser tensioner are extracted. The peak position and peak amplitude are extracted, and the power spectral density of the pressure fluctuation signal is calculated to establish a dynamic response feature set of the riser system. Based on the time-aligned coupled vibration data window, Fourier transforms are performed to generate a fractal vibration energy spectrum matrix. The power spectrum of each time window in the fractal vibration energy spectrum matrix is ​​normalized to obtain the energy probability distribution. Then, the energy probability distribution is used to calculate the top-driven multidimensional vibration energy result. Based on the dynamic response feature set of the riser system, the joint rotational stiffness parameter in the preset vibration system characterization is adjusted to obtain the real-time system stiffness parameter. At the same time, based on the top drive multidimensional vibration energy result, the stiffness adaptive working condition anomaly index is calculated. By combining the real-time system stiffness parameter and the stiffness adaptive working condition anomaly index, the comprehensive state parameters of the drive system health and working condition are established.

[0018] The steps for acquiring synchronous multi-source vibration data streams are as follows: Based on a unified time reference during drilling vessel operations, the six-degree-of-freedom motion data of the hull motion reference unit, the pressure data of the riser tensioner cylinder, the drill string hook load data, torque data, and radial acceleration data of the top drive system are read. The timestamps are parsed and the unit format and channel identifier are unified. The record boundaries and duplicate entries are corrected in ascending order of timestamps to obtain the acquired multi-channel signals. Based on the acquired multi-channel signals, records with the same sampling time are aligned according to the timestamp. The six-degree-of-freedom motion data of the hull motion reference unit, the pressure data of the riser tensioner cylinder, the drill string hook load data, torque data and radial acceleration data of the top drive system are spliced ​​in a fixed channel order and the source identifier and sampling timestamp are added. Missing positions are checked and kept aligned with null value marks to form a synchronous multi-source vibration data stream.

[0019] Specifically, based on a unified time reference during drilling vessel operations, the system reads six-degree-of-freedom motion data from the hull motion reference unit, pressure data from the riser tensioner cylinder, drill string hook load data, torque data, and radial acceleration data from the top drive system. First, it parses the timestamps encoded in ISO 8601 format (e.g., YYYY-MM-DDTHH:mm:ss.sssZ) in each data record, converting them into unified UNIX timestamp values ​​with millisecond-level precision. Then, it establishes a predefined mapping between units and channel identifiers. The standard standardizes all physical quantities to the International System of Units (SI). Specifically, all pressure data is converted to megapascals (MPA), all mechanical load data to kilonewtons (kN), all torque data to kilonewton-meters (kN / m), all acceleration data to meters per second squared (m / s²), and all angular velocity data to radians per second (radians / s). Each data channel is assigned a unique, fixed text identifier. For example, ship heave, roll, pitch, heave acceleration, roll angular velocity, and pitch angular velocity are identified as "MRU_Heave", "MRU_Roll", "MRU_Pitch", "MRU_Heave_Acc", "MRU_Roll_Vel", and "MRU_Pitch_Vel", respectively; riser tensioner pressure is identified as "Riser_Pressure"; and drill string hook load, torque, and radial acceleration are identified as "TDS_Hookload", "TDS_Torque", and "TDS_Radial_Acc", respectively. This identifier, along with the numerical value and timestamp, is stored. Then, after collecting the data records from all channels, a global ascending sort is performed strictly based on the parsed timestamp values. Next, the data sequence is traversed to correct record boundaries and duplicate entries. For duplicate entries, the timestamps and channel identifiers of adjacent records are compared. If they are completely identical, the first record in the sequence is retained, and all subsequent duplicate records are deleted. For record boundary correction, a statistical outlier detection method is used. Taking the pressure data of the riser tensioner as an example, a sample set containing at least 5000 data points is randomly selected from the historical stable operating condition data. The first quartile (Q1) and the third quartile (Q3) of this sample set are calculated, and then the interquartile range (IQR) is calculated. = Q3 - Q1), and set the upper and lower boundaries of the valid data range based on this value. For example, if Q1 is calculated to be 15.2 MPa and Q3 to be 16.8 MPa, then the IQR is 1.6 MPa. Set the lower boundary of the valid data range to Q1 minus 1.5 times the IQR (i.e., 15.2 - 1.5 * 1.6 = 12.8 MPa), and the upper boundary to Q3 plus 1.5 times the IQR (i.e., 16.8 + 1.5 * 1.6 = 19).(2 MPa) Any pressure readings exceeding this range were flagged as abnormal and removed from the dataset. Similar statistical boundary checks were applied to other channel data. After completing all cleaning and normalization steps, the acquired multi-channel signals were obtained.

[0020] Based on the acquired multi-channel signals, a unified target sampling frequency is first determined. This frequency setting should reference the highest effective signal frequency among all signal sources. For example, if the radial acceleration signal contains critical vibration information up to 20 Hz, the target sampling frequency should be set to more than twice the Nyquist frequency, such as 50 Hz. This means generating a master timestamp sequence with a constant time interval of 20 milliseconds. Then, using this master timestamp sequence as a reference, the data from all channels are aligned. Each target time point in the master timestamp sequence is traversed. For each channel, the two closest data points in the acquired multi-channel signals are found (one before and one after the target time point). Linear interpolation is used to calculate the signal value at the target time point. The calculation process involves subtracting the value of the previous data point from the value of the latter data point, multiplying this difference by the time difference between the target time point and the previous data point, dividing by the total time difference between the latter and the former data points, and finally adding the result to the value of the previous data point to obtain the target time value. If, within one sampling period (i.e., 20 milliseconds) before or after a target time point, a channel lacks any valid data points, interpolation cannot be performed, and that position will be considered as missing data. After completing the numerical calculations for all channels at the same target time point, these values ​​are horizontally concatenated into a data row according to a preset fixed channel order, such as from "MRU_Heave" to "TDS_Radial_Acc". A source identifier field is appended to the end of this data row to indicate the physical source of each data column, such as "MRU", "Riser", and "TDS". At the same time, the main timestamp of this row is recorded as the sampling time stamp. During the concatenation process, positions previously identified as missing data are filled with a predefined null value marker (e.g., a specific negative number -9999.99 that is outside the normal data range) to maintain the structural integrity and strict alignment of the data matrix. By repeating this alignment, concatenation, and filling process for all main timestamps throughout the entire time range, a synchronous multi-source vibration data stream is formed.

[0021] The steps for obtaining time-aligned coupled vibration data windows are as follows: Based on the synchronous multi-source vibration data stream, the start and end timestamps of continuous segments are calculated according to the preset time length. The six-degree-of-freedom motion data of the hull motion reference unit, the pressure data of the riser tensioner cylinder, the drill string hook load data, torque data and radial acceleration data of the top drive system are cut according to the start and end timestamps. If the length of the last segment is less than the preset time length, no segment is generated, forming a time-aligned coupled vibration data window.

[0022] Specifically, based on the synchronous multi-source vibration data stream, a time length for segmentation is first set. This length is determined by the frequency resolution requirements of subsequent frequency domain analysis and the ability to fully capture the dynamic processes of typical operating conditions. For example, to distinguish wave frequency components as low as 0.2 Hz in the spectrum, a data length of at least 5 seconds is theoretically required. Considering the periodic characteristics of phenomena such as stick-slip vibration during drilling, the preset time length is ultimately set to 10 seconds. Furthermore, to increase the number of analysis samples and smooth window boundary effects, an overlapping segmentation strategy is adopted, setting a window sliding step size, typically 50% of the window length, i.e., 5 seconds. Next, the start and end timestamps of continuous segments are calculated based on this setting. Specifically, the timestamp of the first data point in the synchronous multi-source vibration data stream is used as the start timestamp of the first segment, and its end timestamp is the start timestamp plus 10 seconds. The start timestamp of the second segment is the start timestamp of the first segment plus the sliding step size of 5 seconds, and its end timestamp is... The timestamp is a new start timestamp plus 10 seconds. This calculation process is executed cyclically, continuously generating new start and end timestamp pairs, until the calculated end timestamp of the segment exceeds the timestamp of the last data point in the data stream. For each set of calculated start and end timestamps, all data lines within that time interval are precisely cut from the synchronous multi-source vibration data stream. Each line contains synchronous data from all channels, including the six-degree-of-freedom motion data of the hull motion reference unit, the pressure data of the riser tensioner cylinder, the drill string hook load data of the top drive system, torque data, and radial acceleration data. During the iterative cutting process, the last possible segment is length-verified, that is, the actual time span from the start timestamp of the last segment to the end of the data stream is calculated. If this time span is less than the preset 10-second time length, the last segment data is discarded because it does not meet the minimum signal length requirement for analysis, and no corresponding segment is generated. Through this series of rigorous calculation, cutting, and verification steps, a time-aligned coupled vibration data window is formed.

[0023] The steps for obtaining the dynamic response feature set of the riser system are as follows: Based on the time-aligned coupled vibration data window, the numerical sequence of the vertical acceleration channel in the hull motion reference unit and the numerical sequence of the pressure fluctuation channel in the riser tensioner cylinder are called. Continuous samples with the same start and end time are synchronously extracted according to the timestamp index to obtain the hull vertical acceleration signal and the riser tensioner pressure fluctuation signal. Based on the vertical acceleration signal of the hull and the pressure fluctuation signal of the riser tensioner, the cross-correlation numerical sequence of the two sets of signals under different time delays is calculated point by point, and the time delay position and amplitude value of the corresponding response peak in the numerical sequence are located and recorded as the peak position and peak amplitude. Based on the peak position and peak amplitude, the pressure fluctuation signal of the riser tensioner is called for frequency domain conversion and the energy distribution density of each frequency component is statistically analyzed. The energy distribution density is taken as the power spectral density of the pressure fluctuation signal. The peak position, peak amplitude and power spectral density are spliced ​​together in a fixed order to form a complete record, thus forming the dynamic response feature set of the riser system.

[0024] Specifically, based on time-aligned coupled vibration data windows, from each data window, according to the preset channel identifiers "MRU_Heave_Acc" and "Riser_Pressure", the complete numerical sequence of the vertical acceleration channel in the hull motion reference unit and the complete numerical sequence of the pressure channel in the riser tensioner cylinder are extracted respectively. Since the original pressure signal contains the superposition of static tension and dynamic fluctuations, it is necessary to separate the pressure fluctuation component. This process is achieved by applying a moving average filter to the original pressure numerical sequence. The window length of this filter is set empirically. For a signal with a sampling frequency of 50 Hz and a window length of 100 sample points, the low-frequency trend signal calculated by this moving average is subtracted point by point from the original pressure signal. The resulting residual sequence is the pressure fluctuation signal of the riser tensioner. Subsequently, for each coupled vibration data window, the extracted vertical acceleration numerical sequence and the calculated pressure fluctuation numerical sequence are taken as a data pair. Based strictly on their shared start and end timestamps and sampling sequences within the data window, continuous samples of the same length are extracted to obtain the hull vertical acceleration signal and the riser tensioner pressure fluctuation signal.

[0025] Based on the ship's vertical acceleration signal and the riser tensioner pressure fluctuation signal, a normalized cross-correlation function is used to quantify the time response relationship between the two. First, a time delay search range is defined, set to -2 seconds to +2 seconds considering the physical response characteristics of the riser system. This corresponds to the time delay of -100 to +100 sampling points at a 50 Hz sampling rate. Then, for each time delay point, the ship's vertical acceleration signal is kept stationary, while the riser tensioner pressure fluctuation signal is shifted by the corresponding sampling point. The sum of the products of all corresponding data points in the overlapping region of the two signals is calculated to eliminate the influence of the signal amplitude itself. The calculated cross-correlation values ​​are normalized by dividing the cross-correlation value by the product of the standard deviations of the two signals, resulting in a normalized cross-correlation coefficient ranging from -1 to +1. This calculation is repeated for each time delay point within the search range, generating a complete cross-correlation numerical sequence. The horizontal axis of this sequence represents the time delay, and the vertical axis represents the normalized cross-correlation coefficient. Then, the maximum absolute value is searched in this cross-correlation numerical sequence. The time delay corresponding to this maximum value is recorded as the peak position in seconds, while the maximum value itself (retaining its original sign) is recorded as the peak amplitude, which is a dimensionless value.

[0026] Based on the peak position and peak amplitude, the pressure fluctuation signal of the diaphragm tensioner obtained from the same data window is further analyzed in the frequency domain to extract energy distribution characteristics. The Welch average periodogram method is used to calculate its power spectral density, with the following specific parameter settings: First, the 10-second pressure fluctuation signal is divided into several segments, each with 512 sample points to obtain a frequency resolution of approximately 0.1 Hz. A 50% overlap (256 sample points) is set between segments to reduce the variance of the spectrum estimation. A Hann window function is applied to each segment to suppress spectral leakage. The windowed signal segments are then analyzed separately... A fast Fourier transform is performed to calculate the squared amplitude of the spectrum. Finally, the power spectrum results of all segments are averaged to obtain the power spectral density of the pressure fluctuation signal of the diaphragm tensioner within the time window. This power spectral density is represented by a set of frequency values ​​and their corresponding energy density values. Finally, the scalar form of peak position and peak amplitude obtained in the previous step is concatenated with the vector form of the power spectral density. The concatenation is performed in a fixed order: first, the peak position value; second, the peak amplitude value; and finally, all the energy density values ​​arranged from low frequency to high frequency in the power spectral density vector. This concatenated long vector is taken as a complete record to form the dynamic response feature set of the diaphragm system.

[0027] The steps to obtain the fractal vibrational energy spectrum matrix are as follows: Based on the time-aligned coupled vibration data window, the numerical sequences of the torque channel, drill string axial load channel, and radial acceleration channel in the top drive system are called. Continuous samples of the same length are segmented in a unified sampling frequency and timestamp order. Short-time Fourier transform is performed on each sample segment to extract the frequency components and energy distribution of the corresponding time period. The frequency component energy results of the three channels in the same time period are combined into matrix units according to the channel order. All matrix units are spliced ​​in the time period order to generate a fractal vibration energy spectrum matrix.

[0028] Specifically, based on time-aligned coupled vibration data windows, from each data window, according to the preset channel identifiers "TDS_Torque", "TDS_Hookload", and "TDS_Radial_Acc", the numerical sequences of the torque channel, drill string axial load channel, and radial acceleration channel in the top drive system are called respectively. Then, short-time Fourier transform processing is performed on the numerical sequences of these three channels. The parameters for this process are as follows: the Hanning window function is selected, and the window length is set to 256 sampling points. This length is chosen to balance ensuring sufficient frequency resolution (approximately 0.2 Hz for a 50 Hz sampling rate) with the time positioning accuracy for capturing transient vibration characteristics. The overlap length between windows is set to 128 sampling points, i.e., a 50% overlap rate. For each channel signal, the processing flow is as follows: starting from the beginning of the signal, the first sample segment of length 256 is extracted, and this sample segment is compared with the Hanning window... The function is multiplied point by point, and a fast Fourier transform is performed on the windowed sample segment to obtain a set of complex frequency coefficients. Then, the square of the modulus of each frequency coefficient is calculated to obtain the power spectrum of the sample segment. This power spectrum vector is recorded. Then, the analysis window is slid forward 128 sampling points to extract the next sample segment with a length of 256. The process of windowing, transformation and power calculation is repeated until the numerical sequence of the entire channel is processed. This generates a two-dimensional matrix, i.e., a spectrum diagram, for each channel, which is composed of the power spectrum vector evolving over time. After the independent processing of the three channels of torque, drill string axial load and radial acceleration is completed, the power spectrum vectors calculated by these three channels in the same time period (i.e., the same analysis window) are vertically stacked into a 3-row N-column matrix unit in a fixed order of torque, drill string axial load and radial acceleration, where N is the number of frequency components. Finally, the matrix units of all time periods are horizontally spliced ​​according to the order of timestamps to generate a fractal vibration energy spectrum matrix.

[0029] The steps for obtaining the top-driven multidimensional vibration energy results are as follows: Based on the fractal vibration energy spectrum matrix, the power spectrum values ​​of the torque channel, the axial load channel of the drill string, and the radial acceleration channel are extracted in each time window. The power spectrum values ​​of each channel at each frequency component are summed according to the time window to obtain the power spectrum sum. The amplitude is normalized by dividing the power spectrum value of the frequency component by the corresponding power spectrum sum to form the energy probability distribution. Based on the energy probability distribution, the frequency-moment weighted spectral entropy is calculated using the following formula: ; in, The weighted spectral entropy is the frequency moment of time window k in channel d. Let k be the energy probability of the frequency component r in channel d within the time window k. The number of frequency components in channel d within the time window. The physical frequency value corresponding to the frequency component r. The Nyquist frequency is half the sampling rate. For time window indexing, For channel indexing, For frequency component index; Based on the frequency moment weighted spectral entropy, the frequency moment weighted spectral entropy of the time window in the torque channel, the frequency moment weighted spectral entropy of the time window in the drill string axial load channel, and the frequency moment weighted spectral entropy of the time window in the radial acceleration channel are combined into a column vector. The column vectors are then aggregated in the order of the time window to form the top-driven multidimensional vibration energy result.

[0030] Specifically, based on the fractal vibration energy spectrum matrix, each time window in the matrix is ​​processed. First, for a specific time window, the power spectrum numerical row vectors representing the torque channel, the drill string axial load channel, and the radial acceleration channel are extracted. Taking the torque channel as an example, its power spectrum numerical row vector contains the energy values ​​of all frequency components within that time window. Next, the sum of all values ​​in this row vector is calculated to obtain the sum of the power spectrum of the torque channel within that time window. This sum represents the total energy of the torque vibration at that moment. Subsequently, the power of each frequency component in this row vector is calculated... The power spectrum values ​​are all divided by the sum of the power spectra just calculated. Through this division operation, the original power spectrum values ​​are converted into a relative energy ratio value. This ratio value represents the proportion of energy of a specific frequency component in the total energy. The ratio values ​​of all frequency components form a new row vector. The sum of all elements in this new vector is always equal to 1, and the value of each element is between 0 and 1. The same summation and normalization operation is also performed on the power spectrum value row vectors of the drill string axial load channel and radial acceleration channel to obtain their respective relative energy ratio vectors. The relative energy ratio vectors of these three channels together constitute the energy probability distribution of this time window.

[0031] formula: The advantage of this formula lies in its application to traditional Shannon spectral entropy (Shannon spectral entropy). Based on this, a first-order frequency moment ( Weighted terms constituted by ) Traditional spectral entropy only measures the degree of disorder or uncertainty in the frequency domain distribution of signal energy. This weighting term, however, introduces an additional consideration of the average frequency position of the energy distribution, comparing the average frequency value of the energy with the Nyquist frequency. This means that when vibration energy concentrates in the high-frequency region, the value of this weighting term increases significantly, thus amplifying the spectral entropy value. Since drilling equipment often exhibits a significant increase in high-frequency components in its vibration signals when experiencing abnormal conditions such as wear, loosening, or stick-slip, this frequency moment weighting enhances the calculated index... It can not only reflect the complexity of the vibration state, but also more sensitively capture high-frequency vibration characteristics that indicate potential faults or deterioration of operating conditions, thus enhancing the ability to detect early faults. The steps for obtaining this parameter are as follows: this parameter is a discrete index of the time window, used to identify the time period for analysis. Its value is determined based on the number of window slides in the aforementioned short-time Fourier transform. For example, the first analysis window... The first value is 1, the second is 2, and so on. In a data window with a length of 10 seconds, a sampling rate of 50 Hz, a window length of 256, and an overlap of 128 points, the total number of samples is 500. The number of sliding windows is... After rounding, there are 3 complete windows, therefore The value range is 1, 2, 3; The acquisition steps are as follows: this parameter is the channel index of the vibration signal, used to distinguish different physical measurement quantities. In this application, according to the fixed order of data acquisition and processing, the torque channel is defined as... The drill string axial load channel is defined as The radial acceleration channel is defined as ; The steps to obtain this parameter are as follows: the parameter is a time window. In the passage The The energy probability of each frequency component is directly derived from the energy probability distribution calculated in the previous step. Its value is between 0 and 1, representing the proportion of the energy of that frequency component to the total energy. For a given time window... and channels All frequency components The sum is 1; The steps to obtain this parameter are as follows: the parameter is the total number of frequency components in the one-sided spectrum obtained after the short-time Fourier transform, and its value is determined by the number of transform points. The decision is made, and the calculation formula is as follows: In the embodiments of this application, the window length of the short-time Fourier transform is 256 sampling points, that is... Therefore, the number of frequency components ; The steps to obtain this parameter are as follows: The physical frequency value corresponding to each frequency component, in Hertz, is calculated based on the sampling frequency. The number of points in the Fourier transform The calculation formula is: ,in The index starts from 1. In this application, the sampling frequency... 50 Hz The value is 256, therefore the first... The frequency of each frequency component is ; The steps to obtain this parameter are as follows: the parameter is the Nyquist frequency, which is the highest analyzable frequency determined according to the sampling theorem, and its value is the sampling frequency. Half of the sampling frequency in this application It is 50 Hz, therefore the Nyquist frequency hertz.

[0032] Calculation process: To calculate the first time window ( Lower torque channel Frequency moment weighted spectral entropy For example, Define a simplified energy probability distribution In order to facilitate calculation, for example in Of the frequency components, only four have non-zero energy, while the rest have zero. Let the 11th frequency component be ( Energy probability , The 21st frequency component ( Energy probability , The 31st frequency component ( Energy probability , 41st frequency component ( Energy probability , The sum of these probabilities is , Calculate the corresponding physical frequency value : hertz, hertz, hertz, hertz, First, calculate the Shannon spectral entropy part. : , , , Next, calculate the numerator in the frequency moment weighting term. : , , Calculate the complete weighted terms: , Finally, multiply the two parts together to get... : .

[0033] This result indicates that, Within the given time window, the frequency-moment weighted spectral entropy of the torque channel is 1.2973. This value is a comprehensive state indicator. Comparing it with a baseline value obtained under normal operating conditions (e.g., the average entropy value calculated from a large amount of normal operating data, such as 1.1) can determine the degree of deviation from the current operating condition. If this value increases significantly, for example, exceeding 1.5 times the baseline value (i.e., greater than 1.65), it indicates that the vibration energy distribution is becoming more complex or shifting towards higher frequencies, potentially indicating an anomaly. This calculation result... This will be used as a component of the subsequent top-driven multidimensional vibration energy result vector.

[0034] Based on frequency-moment weighted spectral entropy, for each time window Extract the frequency moment weighted spectral entropy values ​​of the three channels calculated within this window, specifically the frequency moment weighted spectral entropy of the torque channel. Frequency moment weighted spectral entropy of the drill string axial load channel Weighted spectral entropy of the frequency moment of the radial acceleration channel These three independent scalar values ​​are organized into a three-dimensional column vector according to a fixed channel order (torque, drill string axial load, radial acceleration). Its form is This vector represents the overall vibration state of the top-driven system within the given time window in the three-dimensional feature space. Subsequently, for all consecutive time windows obtained through short-time Fourier transform (from... To the last window Repeat the vector construction process described above to generate a series of column vectors. Finally, these column vectors arranged in chronological order are aggregated into a time series matrix. Each column of this matrix is ​​a feature vector of a time window, and the rows correspond to different vibration channel features. This final matrix or vector sequence is the top-driven multidimensional vibration energy result.

[0035] The steps for obtaining real-time system stiffness parameters are as follows: Based on the dynamic response feature set of the riser system, the initial value of the joint rotational stiffness parameter characterized by the preset vibration system is read. The differences between the predicted output and the dynamic response feature set of the riser system in terms of peak position and peak amplitude are compared item by item and the sign is determined. The joint rotational stiffness parameter is adjusted according to the sign of the difference until the absolute value of the difference no longer decreases, and the real-time system stiffness parameter is obtained.

[0036] Specifically, based on the dynamic response feature set of the riser system, a pre-established simplified finite element model capable of characterizing the dynamic behavior of the riser system is first invoked. This model simplifies the riser into a series of structures connected by flexible beam elements and rotational spring elements, where the stiffness of the rotational spring elements is the joint rotational stiffness parameter to be identified. Simultaneously, the theoretical design rotational stiffness value of the riser joint, such as 100 MN·m / radian, is retrieved from the drilling vessel's design manual or equipment specifications and used as the initial value for the joint rotational stiffness parameter iteration. Next, an iterative optimization loop is initiated. In each loop, the hull vertical acceleration signal within the current coupled vibration data window is applied as an excitation input to the top of the finite element model, and dynamic simulation is run to obtain the model's predicted riser tensioner pressure response sequence. The predicted peak position and peak amplitude are then calculated. Finally, the model's predicted peak position is compared with the actual peak position extracted from the riser system's dynamic response feature set to calculate the positional difference. Simultaneously, the predicted peak amplitude is compared with the actual peak amplitude. The amplitude values ​​are compared to calculate the amplitude difference. Based on the sign of these two differences, an adjustment strategy is executed: if the predicted response peak appears later than the actual peak (positive position difference) or the response amplitude is larger than the actual peak (positive amplitude difference), it indicates that the current model stiffness is too low. Therefore, the joint rotational stiffness parameter is increased by a preset step size, and vice versa. The initial value of this step size is set to 1% of the initial value, i.e., 1 meganewton-meter per radian. After each adjustment, the simulation is rerun and the difference is calculated. A comprehensive error is defined as the weighted sum of the absolute values ​​of the position difference and the absolute values ​​of the amplitude difference. If the comprehensive error of this cycle is less than that of the previous cycle, the same adjustment direction is maintained and the iteration continues. If the comprehensive error begins to increase, the adjustment direction is reversed and the adjustment step size is halved. This iterative process continues until the change in comprehensive error in three consecutive iterations is less than a preset convergence threshold, which is set to 0.01. At this point, the joint rotational stiffness parameter in the model is considered to have converged to a value that best matches the actual dynamic response. This value is the obtained real-time system stiffness parameter.

[0037] The steps for obtaining the comprehensive health and operating condition parameters of the drive system are as follows: Based on the real-time system stiffness parameters, the stiffness adaptive working condition anomaly index is calculated using the following formula: ; in, The stiffness adaptive working condition anomaly index is given by time window k. The top driving multidimensional vibrational energy result vector of time window k. The frequency moment weighted spectral entropy of time window k in the torque channel. The frequency moment weighted spectral entropy of time window k in the drill string axial load channel. The frequency moment weighted spectral entropy of time window k in the radial acceleration channel. For reference stiffness value, As a reference mean vector, For reference covariance matrix, This is the stiffness sensitivity coefficient. For transpose, The inverse of the matrix. These are the real-time system stiffness parameters; Based on the stiffness adaptive working condition anomaly index, the real-time system stiffness parameters and the stiffness adaptive working condition anomaly index are stored in the same time window and combined in the order of timestamps to form a sequence of joint rotation stiffness values ​​and working condition stability scores, thereby generating comprehensive state parameters of drive system health and working condition.

[0038] Specifically, the formula: The advantage of the formula lies in the right-hand side of the exponent, namely the Mahalanobis distance. This is used to measure the deviation of the current working point, represented by the top-driven multidimensional vibration energy result, from the center of the normal working condition cluster. It can effectively capture instantaneous working condition fluctuations caused by factors such as formation changes and drill bit wear during drilling. The weighted terms on the left side of the formula... This introduces consideration of the system's structural health status, and incorporates the real-time identified system stiffness. Compared with the benchmark healthy stiffness In comparison, when the system experiences structural degradation such as loose joints, the stiffness will slowly but continuously deviate, and this weighting term will increase accordingly, thus amplifying the overall anomaly index. This design makes... It can not only react to real-time operating condition anomalies like traditional process monitoring methods, but also provide early warning of potential structural health problems due to stiffness degradation, thus realizing a comprehensive assessment of operating conditions and structural health. The steps to obtain this parameter are as follows: this parameter is a discrete index of the time window, used to identify the specific time period for analysis; its value is determined by the data stream segmentation method. If the data stream is segmented into... If there is a time window, then The value range is from 1 to An integer, for example, if the current calculation is for the data of the 10th time window, then... ; The steps to obtain this parameter are as follows: the parameter is the time window. The top drive multidimensional vibration energy result vector is a direct product of the previous step. It is a three-dimensional column vector composed of frequency moment weighted spectral entropy from three channels: torque, drill string axial load, and radial acceleration. This vector comprehensively characterizes the vibration energy distribution characteristics of the top drive system within this time window. For example, within the time window... At that time, it was obtained through calculation. ; The steps to obtain this parameter are as follows: the parameter is the time window. The real-time system stiffness parameters, obtained through iterative optimization using the finite element model in the previous step, reflect the equivalent joint rotational stiffness of the riser system at the current moment, measured in meganewton-meters per radian. For example, within a time window... At that time, the identified stiffness value is Mega Newton-meters per radian; The steps for obtaining this parameter are as follows: This parameter is a reference stiffness value, representing the joint rotational stiffness of the system under absolutely healthy conditions. The method of obtaining it is as follows: During the initial operation phase after the drilling system has been debugged or a new riser joint has been replaced, 24 consecutive hours of stable operating data are collected. For each data window within this period, the aforementioned real-time system stiffness parameter identification process is executed to obtain a stiffness value sequence. After removing outliers from this sequence that exceed the mean plus or minus three standard deviations, the average of all remaining stiffness values ​​is calculated. This average is used as the reference stiffness value. For example, after calculation, the following is obtained: Mega Newton-meters per radian; The steps to obtain this parameter are as follows: This parameter is a reference mean vector, representing the expected center of the top-driven multidimensional vibration energy result vector under healthy conditions. The method of obtaining it is: using and acquiring... Using the same 24-hour health status dataset, calculate the top-driving multidimensional vibrational energy result vector for all time windows in the dataset. Then, the average value is calculated for each corresponding dimension of all vectors to obtain a three-dimensional mean vector. For example, the calculated mean vector is... ; The steps for obtaining this parameter are as follows: The parameter is a reference covariance matrix, which describes the linear correlation and dispersion of the three frequency-moment weighted spectral entropy features under healthy conditions. It is obtained by: based on... and The same health status dataset and its corresponding Given a set of vectors, calculate the covariance matrix of this set of three-dimensional vector samples. For example, calculate the... ; The steps to obtain this parameter are as follows: This parameter is the stiffness sensitivity coefficient, a dimensionless adjustment parameter used to control the amplification effect of stiffness deviation on the total anomaly index. Its setting is based on risk assessment and expert experience, with the goal of ensuring that a stiffness decrease with clear physical significance (e.g., a 10% decrease, which may indicate significant loosening of joint bolts) produces an anomaly index increase sufficient to attract the operator's attention (e.g., a 50% increase in the total index while the Mahalanobis distance remains constant). Based on this goal, calculations can be performed: Solving for ,therefore .

[0039] Calculation process: With time window For example, substitute the obtained parameter values ​​into the calculation. Known parameters: , , , , , , First, calculate the Mahalanobis distance part. : , Calculate the inverse of the reference covariance matrix : , Calculate the square of the Mahalanobis distance : , , , Calculate Mahalanobis distance: , Then, calculate the stiffness weighting term: ; Finally, the anomaly index of the stiffness adaptive working condition is calculated. : .

[0040] The results show that, within time window 10, the stiffness adaptive operating condition anomaly index of the drive system is 1.970. This value is a comprehensive measure of health and operating condition deviation. Typically, a threshold based on the statistical distribution of health data is preset. For example, the threshold for normal operating conditions is set to 3 (based on the chi-square distribution characteristics of Mahalanobis distance, covering 99% of normal samples). The current calculated value of 1.970 is less than this threshold, indicating that although the real-time stiffness has decreased and the vibration characteristics have deviated to some extent, the overall state of the system is still within the normal and acceptable range. This value will be recorded for continuous state trend monitoring.

[0041] Based on the stiffness adaptive anomaly index, the system generates a composite data record containing key state information for each processed time window. Specifically, for each time window, the system pairs the real-time system stiffness parameter obtained in the previous step (a physical quantity representing the structural health status) with the stiffness adaptive anomaly index calculated in this step (a score representing comprehensive operational stability). This pair of values ​​is then associated with the start timestamp of the time window, creating a structured data entry. This entry contains at least three fields: timestamp, joint rotational stiffness value, and anomaly score. For example, for time window 10, with a start timestamp of "2023-11-01T14:30:50.000Z", the generated record would be {timestamp: "2023-11-01T14:30:50.000Z", stiffness:92.5, anomaly_score:} 1.970} By repeating this operation over all consecutive time windows, the system will generate a complete data sequence arranged in ascending order of timestamps. This sequence is essentially two synchronous time series: one is the joint rotation stiffness numerical sequence, which reflects the long-term evolution trend of the equipment's structural health status, and the other is the operating condition stability score sequence, which reveals the real-time dynamic changes in operating conditions. The combination of these two sequences constitutes the comprehensive state parameters of the drive system's health and operating conditions.

[0042] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A data acquisition method for an oil extraction actuator, characterized in that, Includes the following steps: During drilling vessel operations, six-degree-of-freedom motion data of the hull motion reference unit, pressure data of the riser tensioner cylinder, drill string hook load data, torque data, and radial acceleration data of the top drive system are collected simultaneously. The collected multi-channel signals are aggregated to form a synchronous multi-source vibration data stream. The synchronous multi-source vibration data stream is then divided according to a preset time length to obtain a time-aligned coupled vibration data window. Based on the time-aligned coupled vibration data window, the vertical acceleration signal of the hull and the pressure fluctuation signal of the riser tensioner are extracted, the peak position and peak amplitude are extracted, and the power spectral density of the pressure fluctuation signal is calculated to establish a dynamic response feature set of the riser system. Based on the time-aligned coupled vibration data window, Fourier transforms are performed to generate a fractal vibration energy spectrum matrix. The power spectrum of each time window in the fractal vibration energy spectrum matrix is ​​normalized to obtain the energy probability distribution. Then, the energy probability distribution is used to calculate the top-driven multidimensional vibration energy result. Based on the dynamic response feature set of the water-proof pipe system, the joint rotational stiffness parameter in the preset vibration system characterization is adjusted to obtain the real-time system stiffness parameter. At the same time, based on the top drive multidimensional vibration energy result, the stiffness adaptive working condition anomaly index is calculated. By combining the real-time system stiffness parameter and the stiffness adaptive working condition anomaly index, the comprehensive state parameters of the drive system health and working condition are established.

2. The data acquisition method for the oil extraction drive according to claim 1, characterized in that, The steps for acquiring the synchronous multi-source vibration data stream are as follows: Based on a unified time reference during drilling vessel operations, the six-degree-of-freedom motion data of the hull motion reference unit, the pressure data of the riser tensioner cylinder, the drill string hook load data, torque data, and radial acceleration data of the top drive system are read. The timestamps are parsed and the unit format and channel identifier are unified. The record boundaries and duplicate entries are corrected in ascending order of timestamps to obtain the acquired multi-channel signals. Based on the acquired multi-channel signals, records with the same sampling time are aligned according to the timestamp. The six-degree-of-freedom motion data of the hull motion reference unit, the pressure data of the riser tensioner cylinder, the drill string hook load data, torque data and radial acceleration data of the top drive system are spliced ​​in a fixed channel order and the source identifier and sampling timestamp are added. Missing positions are checked and kept aligned with null value marks to form a synchronous multi-source vibration data stream.

3. The data acquisition method for the oil extraction drive according to claim 1, characterized in that, The steps for obtaining the time-aligned coupled vibration data window are as follows: Based on the synchronous multi-source vibration data stream, the start and end timestamps of continuous segments are calculated according to the preset time length. The six-degree-of-freedom motion data of the hull motion reference unit, the pressure data of the riser tensioner cylinder, the drill string hook load data, torque data and radial acceleration data of the top drive system are cut according to the start and end timestamps. If the length of the last segment is less than the preset time length, no segment is generated, forming a time-aligned coupled vibration data window.

4. The data acquisition method for the oil extraction drive according to claim 1, characterized in that, The steps for obtaining the dynamic response feature set of the riser system are as follows: Based on the time-aligned coupled vibration data window, the numerical sequence of the vertical acceleration channel in the hull motion reference unit and the numerical sequence of the pressure fluctuation channel in the riser tensioner cylinder are called. Continuous samples with the same start and end times are synchronously extracted according to the timestamp index to obtain the hull vertical acceleration signal and the riser tensioner pressure fluctuation signal. Based on the vertical acceleration signal of the hull and the pressure fluctuation signal of the riser tensioner, the cross-correlation numerical sequence of the two sets of signals under different time delays is calculated point by point, the time delay position and amplitude value of the corresponding response peak in the numerical sequence are located, and recorded as the peak position and peak amplitude. Based on the peak position and peak amplitude, the pressure fluctuation signal of the riser tensioner is called for frequency domain conversion and the energy distribution density of each frequency component is statistically analyzed. The energy distribution density is taken as the power spectral density of the pressure fluctuation signal. The peak position, peak amplitude and power spectral density are spliced ​​together in a fixed order to form a complete record, thus forming a dynamic response feature set of the riser system.

5. The data acquisition method for an oil extraction drive according to claim 1, characterized in that, The steps for obtaining the fractal vibration energy spectrum matrix are as follows: Based on the time-aligned coupled vibration data window, the numerical sequences of the torque channel, drill string axial load channel, and radial acceleration channel in the top drive system are called. Continuous samples of the same length are segmented in a unified sampling frequency and timestamp order. Short-time Fourier transform is performed on each sample segment to extract the frequency components and energy distribution of the corresponding time period. The frequency component energy results of the three channels in the same time period are combined into matrix units according to the channel order. All matrix units are spliced ​​together according to the time period order to generate a fractal vibration energy spectrum matrix.

6. The data acquisition method for the oil extraction drive according to claim 1, characterized in that, The steps for obtaining the top-driven multidimensional vibration energy results are as follows: Based on the fractal vibration energy spectrum matrix, the power spectrum values ​​of the torque channel, the power spectrum values ​​of the drill string axial load channel, and the power spectrum values ​​of the radial acceleration channel are extracted in each time window. The power spectrum values ​​of each channel at each frequency component are summed according to the time window to obtain the power spectrum sum. The amplitude is normalized by dividing the power spectrum value of the frequency component by the corresponding power spectrum sum to form the energy probability distribution. Based on the energy probability distribution, calculate the frequency moment weighted spectral entropy; Based on the aforementioned frequency moment weighted spectral entropy, the frequency moment weighted spectral entropy of the time window in the torque channel, the frequency moment weighted spectral entropy of the time window in the drill string axial load channel, and the frequency moment weighted spectral entropy of the time window in the radial acceleration channel are combined into a column vector. The column vectors are then aggregated sequentially according to the time window order to form the top-driven multidimensional vibration energy result.

7. The data acquisition method for an oil extraction drive according to claim 1, characterized in that, The steps for obtaining the real-time system stiffness parameters are as follows: Based on the dynamic response feature set of the riser system, the initial value of the joint rotational stiffness parameter characterized by the preset vibration system is read. The differences between the predicted output and the dynamic response feature set of the riser system in terms of peak position and peak amplitude are compared item by item and the sign is determined. The joint rotational stiffness parameter is adjusted according to the sign of the difference until the absolute value of the difference no longer decreases, and the real-time system stiffness parameter is obtained.

8. The data acquisition method for the oil extraction drive according to claim 1, characterized in that, The steps for obtaining the comprehensive health and operating condition parameters of the drive system are as follows: Calculate the stiffness adaptive working condition anomaly index based on the real-time system stiffness parameters; Based on the stiffness adaptive working condition anomaly index, the real-time system stiffness parameters and the stiffness adaptive working condition anomaly index are stored in the same time window and combined in the order of timestamps to form a sequence of joint rotation stiffness values ​​and working condition stability scores, thereby generating comprehensive health and working condition status parameters of the drive system.

9. The data acquisition system of the oil exploration driver data acquisition method according to any one of claims 1-8, characterized in that, include: During drilling vessel operations, the data acquisition module simultaneously acquires six-degree-of-freedom motion data of the hull motion reference unit, pressure data of the riser tensioner cylinder, drill string hook load data, torque data, and radial acceleration data of the top drive system. The acquired multi-channel signals are aggregated to form a synchronous multi-source vibration data stream, which is then divided according to a preset time length to obtain a time-aligned coupled vibration data window. The dynamic response feature extraction module extracts the hull vertical acceleration signal and the riser tensioner pressure fluctuation signal based on the time-aligned coupled vibration data window, extracts the peak position and peak amplitude, calculates the power spectral density of the pressure fluctuation signal, and establishes a dynamic response feature set of the riser system. The vibration energy analysis module performs Fourier transforms on the time-aligned coupled vibration data windows to generate a fractal vibration energy spectrum matrix. It then normalizes the power spectrum of each time window within the fractal vibration energy spectrum matrix to obtain an energy probability distribution. Finally, it calculates the top-driven multidimensional vibration energy result based on the energy probability distribution. The health and operating condition assessment module, based on the dynamic response feature set of the riser system, adjusts the joint rotational stiffness parameter in the preset vibration system characterization to obtain the real-time system stiffness parameter. At the same time, based on the multi-dimensional vibration energy result of the top drive, it calculates the stiffness adaptive operating condition anomaly index. Combining the real-time system stiffness parameter and the stiffness adaptive operating condition anomaly index, it establishes the comprehensive health and operating condition status parameters of the drive system.

Images