A downhole information interpretation method and system based on drill bit vibration data

Abstract

The application discloses a downhole information interpretation method and system based on drill bit vibration data, and the method comprises the following steps: integrating an intelligent sensing module in the water eye of an existing drill bit, collecting downhole vibration data and environmental parameters in the drilling process at a high sampling frequency; synchronously aligning the downhole vibration data and the ground drilling parameters; calibrating and compensating the synchronized vibration data by using the environmental parameters and extracting features to obtain a vibration feature parameter vector reflecting the interaction between the drill bit and the stratum; and identifying the stratum lithology or the drilling working condition based on the vector and in combination with the environmental parameters. The application solves the problems of information distortion and low resolution caused by the fact that the sensor is far away from the drill bit and the data bandwidth limitation of the traditional measurement-while-drilling system, provides an unprecedented high-precision data basis for post-drilling fine analysis, geological model calibration and future drilling decision optimization, and the good compatibility with the existing drill bit does not require redesign and manufacture of the drill bit, thereby significantly reducing the application cost.

Description

Technical Field

[0001] This invention relates to the field of drilling information processing, and specifically to a method and system for interpreting downhole information based on drill bit vibration data. Background Technology

[0002] In the field of oil and gas drilling, a precise understanding of the lithology of the downhole formation and drilling conditions is of great significance for optimizing drilling parameters and improving drilling efficiency.

[0003] Traditionally, measurement while drilling (MWD) systems have been the primary means of acquiring downhole information. However, existing technologies suffer from the following drawbacks: The sensors in traditional MWD systems are typically located far from the drill bit, causing vibration signals to attenuate and distort significantly after traveling long distances within the drill string, resulting in inaccurate information received from the surface or upper drilling tools. Furthermore, limited by the bandwidth of traditional data transmission methods such as mud pulses, MWD systems have extremely low data sampling rates, making it impossible to capture high-frequency vibration characteristics that accurately reflect the drill bit-rock interaction mechanism and complex operating conditions (such as high-frequency stick-slip and drill bit bouncing).

[0004] To address the aforementioned signal distortion issues, the industry has attempted to mount sensors as close to the drill bit as possible. However, these solutions typically require specific mounting cavities or channels in the drill bit body, necessitating the design and manufacture of specialized drill bits integrating the sensors. These specialized drill bits are not only extremely expensive and complex to manufacture, but also lack versatility, being incompatible with the numerous standard drill bits used in the field, significantly limiting their application and promotion. Furthermore, modifications to the drill bit's main structure may introduce potential stress concentration risks, affecting the drill bit's structural strength and service life.

[0005] Therefore, based on this type of distorted and low-resolution data, it is difficult to perform fine-grained classification of formation lithology and to conduct in-depth post-drilling dynamics analysis. This leads to a vague understanding of the formation and operating conditions, restricting the optimization of subsequent drilling parameters, the improvement of drill bit selection, and strategies for dealing with complex formations, thus affecting the overall efficiency and economy of drilling operations.

[0006] It is evident that how to achieve near-bit high-fidelity data acquisition without changing the existing standard drill bit design or significantly increasing costs is a technical challenge that urgently needs to be solved in this field. Summary of the Invention

[0007] The purpose of this invention is to overcome the above-mentioned technical deficiencies and propose a downhole information interpretation method and system based on drill bit vibration data to solve the problems of information distortion and low resolution in traditional measurement methods. At the same time, this invention is committed to achieving high compatibility with existing drilling equipment through an ingenious modular integration scheme, which can be applied without changing or redesigning the drill bit, thereby significantly reducing the implementation cost and promotion threshold of the technology.

[0008] To achieve the above-mentioned technical objectives, the present invention adopts the following technical solution: This invention provides a downhole information interpretation system based on drill bit vibration data, comprising: The drill bit (1) has an integrated smart sensor module (2) installed in the water inlet (11) of the drill bit (1). The smart sensor module (2) includes a mounting housing (21) for fixed connection with the water inlet (11), a six-axis inertial sensing unit (22) for collecting vibration data, a temperature sensor (23) for collecting downhole temperature parameters, and a pressure sensor (24) for collecting downhole pressure parameters. The six-axis inertial sensing unit (22), the temperature sensor (23), and the pressure sensor (24) are all integrated in the mounting housing (21).

[0009] Furthermore, the mounting housing (21) is an annular housing, which includes an outer shell (212) and an inner shell (211). The outer shell (212) is fixedly connected to the water eye (11), and the outer shell (212) is fixedly connected to the inner shell (211), forming a mounting cavity for mounting the six-axis inertial sensing unit (22), temperature sensor (23), and pressure sensor (24). The shape and size of the outer shell (212) match the water eye (11).

[0010] A method for interpreting downhole information based on drill bit vibration data includes: S1. The drilling parameters are acquired through the ground recording system, and the downhole vibration data and environmental parameters are acquired through the intelligent sensing module integrated into the drill bit water hole. The environmental parameters include temperature parameters acquired through the temperature sensor and pressure parameters acquired through the pressure sensor. S2. Using preset characteristic events in the drilling process as time anchors, synchronize and align the time series of downhole vibration data and the time series of drilling parameters to obtain downhole multidimensional vibration data; S3. Perform drift calibration or noise compensation on downhole multidimensional vibration data using temperature parameters, and perform signal processing and feature extraction on the calibrated and compensated downhole multidimensional vibration data to obtain one or more vibration feature parameter vectors reflecting the interaction between the drill bit and the formation. S4. Match the vibration feature parameter vector with the formation-vibration spectrum feature library to determine the lithology of the current drill bit; at the same time, extract pressure-derived features from the pressure parameters, and combine the vibration feature parameter vector with the pressure-derived features to form a fused feature vector, and use a machine learning model to identify the drilling conditions experienced by the current drill bit.

[0011] Furthermore, the preset feature event in step S2 is the drilling stop period when a single drill pipe is connected. Synchronization alignment is achieved by calculating the time offset to maximize the sliding cross-correlation function between the downhole and surface reference signals. The sliding cross-correlation function is defined as follows: Where x(t) is the time series of downhole vibration data, y(t) is the time series of surface drilling parameters, τ is the time offset, and t is the time variable.

[0012] Furthermore, methods for drift calibration or noise compensation of downhole multidimensional vibration data using temperature parameters include: based on temperature parameters, adjusting the original measured values ​​A of the downhole multidimensional vibration data using a pre-calibrated temperature-drift polynomial model. 原始 Compensation is performed to obtain the corrected measurement value A. 校正 The compensation formula is: A 校正 =A 原始 −(k2(T−T ref ) 2 +k1(T−T ref )) Where T is the temperature parameter, T ref To calibrate the reference temperature, k1 and k2 are the first-order and second-order temperature drift coefficients obtained before calibration.

[0013] Furthermore, the method for signal processing of the calibrated and compensated downhole multidimensional vibration data includes: using Fast Fourier Transform to process the discrete calibrated and compensated downhole multidimensional vibration data x n Complex spectrum value X converted to the frequency domain k The transformation relationship is defined by the following formula: Where, x n For discrete, calibrated, compensated downhole multidimensional vibration data, X k denoted as the complex spectrum value in the frequency domain, where N is the number of sampling points, n is the sampling point index of the time series, k is the frequency component index in the frequency domain, and j is the imaginary unit.

[0014] Furthermore, the method for feature extraction of the calibrated and compensated downhole multidimensional vibration data includes: using wavelet transform to perform time-frequency analysis on the non-stationary downhole multidimensional vibration data to identify the local time-frequency characteristics of transient shocks and periodic patterns, specifically including: Multi-level decomposition of downhole multi-dimensional vibration data yields multiple sub-band signals with different frequency ranges; Calculate the energy E of each of the multiple sub-band signals. i The calculation formula is as follows: Among them, E i Let i be the energy of the i-th sub-band signal. Let M be the k-th discrete sampling point of the i-th sub-band signal, and M be the total number of sampling points of the sub-band signal. The calculated energy of each sub-band is used to construct a time-frequency feature vector, which is used to identify transient impacts and periodic modes. The process of identifying transient impacts and periodic modes includes: By comparing the high-frequency sub-band energy in the time-frequency feature vector with a preset impact threshold, when the high-frequency sub-band energy exceeds the impact threshold in a short period of time, it is identified as a transient impact. By analyzing the energy distribution in the time-frequency feature vector, when the proportion of energy in a certain sub-band in the total energy continuously exceeds the preset concentration threshold, it is identified as a periodic pattern.

[0015] Furthermore, methods for establishing a stratigraphic-vibration spectral feature library include: By segmenting and correlating downhole vibration data from historical drilling processes with known geological logging data according to drill bit depth, the stratigraphic segments in each region can be determined. Extract and store typical vibration characteristic parameter vectors that correspond one-to-one with each type of stratum lithology, as identification templates in the library.

[0016] Furthermore, pressure-derived characteristics include pressure fluctuation variance. The calculation formula is as follows: Among them, P i For pressure sampling points within the time window, μ p The average pressure within the window is given by N, where N is the number of sampling points within the window, and i is the index. Methods for forming a fused feature vector by combining vibration characteristic parameter vectors with pressure-derived features include: The m vibration features extracted from the downhole multidimensional vibration data are used to form a vibration feature parameter vector with dimension m. The n pressure-derived features extracted from the pressure parameters are used to form a pressure-derived feature vector with dimension n. Before splicing, the vibration characteristic parameter vector and the pressure derived characteristic vector are first normalized to eliminate the dimensional influence between different features. By concatenating the normalized pressure-derived feature vector into the normalized vibration feature parameter vector in terms of dimension, a fused feature vector with dimension m+n is formed, and this fused feature vector is used as the input of the machine learning model.

[0017] Furthermore, the machine learning model is a pre-trained support vector machine model, which classifies the input fused feature vector x using the following decision function f(x): Where f(x) is the decision function output of the SVM model, sgn is the sign function, and x is the input fusion feature vector. i For support vectors, K is the kernel function, and α i is the Lagrange multiplier, b is the bias term, l is the total number of support vectors, and i is the index.

[0018] Compared with existing technologies, the beneficial effects of the downhole information interpretation method and system based on drill bit vibration data provided by this invention include: 1. By directly collecting and storing data at the vibration source (inside or near the drill bit), the attenuation and distortion of vibration signals during propagation in the drill string are fundamentally avoided, and high-fidelity, high-bandwidth original vibration data that most accurately reflects the interaction between the drill bit and the formation is obtained. 2. Based on the acquired high-fidelity data, high-resolution formation lithology identification and retrospective analysis of complex drilling conditions (such as micro-fragmentation, stick-slip, and bouncing) can be performed after drilling is completed. Its accuracy and detail far exceed those of traditional measurement-while-drilling technology. 3. This refined analysis result can be used to establish or calibrate a more accurate formation-vibration characteristic model, providing an unprecedented reliable basis for optimizing drilling parameters, selecting drill bits, and mitigating risks in adjacent wells or the next well section. This macroscopically improves the efficiency and safety of drilling operations and promotes the refinement and intelligent development of drilling engineering. 4. This invention designs the intelligent sensing module as an independent, reusable unit that can be installed inside the drill bit's water hole. This design requires no structural modifications or special customization to the drill bit body and can be directly adapted and installed in various models of existing standard drill bits. This "plug-and-play" mode greatly reduces the implementation threshold and single-operation cost of the technology, avoids the expensive cost of custom drill bits for integrating sensors, facilitates rapid technology promotion and large-scale application, and has outstanding economic value and field applicability. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the downhole information interpretation method based on drill bit vibration data of the present invention. Figure 2 This is a schematic diagram of the downhole information interpretation system based on drill bit vibration data according to the present invention; Figure 3 yes Figure 2 A magnified view of a portion of region A in the middle; Explanation of reference numerals in the attached drawings: 1-Drill bit, 11-Water eye, 2-Intelligent sensing module, 21-Annular housing, 211-Inner housing, 212-Outer housing, 22-Six-axis inertial sensing unit, 23-Temperature sensor, 24-Pressure sensor, 25-Memory, 26-Rechargeable battery. Detailed Implementation

[0020] The technical solutions in the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. The described embodiments are only possible technical implementations of the present invention, but are not limited thereto. Other embodiments obtained by those skilled in the art in conjunction with the embodiments of the present invention without creative effort are also within the protection scope of the present invention.

[0021] Please refer to Figure 2 and Figure 3 A downhole information interpretation system based on drill bit vibration data includes an intelligent sensing module 2 integrated in the water hole of the drill bit 1. The intelligent sensing module 2 includes a mounting housing 21, a six-axis inertial sensing unit 22, a temperature sensor 23, and a pressure sensor 24.

[0022] Choosing the water inlet as the installation location is a crucial and cost-effective design choice. Water inlets are an inherent feature of all PDC drill bits, used to spray drilling fluid to clean and cool the bottom of the well. This invention perfectly utilizes this existing space by designing the sensing module as a ring-shaped mounting housing 21 adapted to the standard water inlet size, avoiding any destructive modifications or redesigns to the drill bit's main structure. This means users can directly utilize a large inventory of drill bits in their warehouse, simply installing this module to achieve an "intelligent" upgrade. This not only ensures that the original mechanical and hydraulic properties of the drill bit remain unaffected but also fundamentally solves the problem of high-cost customized drill bits, paving the way for the widespread, low-cost application of the technology.

[0023] The mounting housing 21 is made of high-strength metal and is ring-shaped. This structure ensures its own strength, preventing damage in the complex downhole environment, while also forming a flow channel for drilling fluid. The mounting housing 21 is fixed inside one or more water holes 11 of the drill bit 1. It includes an inner housing 211 and an outer housing 212. The shape and size of the outer housing 212 match the water hole 11. The outer housing 212 is fixed inside the water hole 11 by welding or threaded connection, while the inner housing 211 is fixed inside the outer housing 212. A sealed mounting cavity is formed between the inner housing 211 and the outer housing 212 for mounting the aforementioned six-axis inertial sensing unit, temperature sensor, and pressure sensor. The mounting housing 21 can also be made of ceramic or other materials with certain strength and corrosion resistance.

[0024] The six-axis inertial sensing unit 22 is used to collect vibration data such as three-axis acceleration and three-axis angular velocity. It features high precision and high stability, accurately capturing minute vibration changes of the drill bit during drilling. The six-axis inertial sensing unit 22 is fixed in the mounting cavity by bolts or glue and is connected to the data transmission line to transmit the collected data. Eight-axis or ten-axis multi-axis inertial sensing units can also be used to obtain more comprehensive motion information.

[0025] Temperature sensor 23 is used to collect temperature parameters downhole. It is typically made using principles such as thermistors or thermocouples, and has high temperature measurement accuracy. Temperature sensor 23 is also fixed inside the mounting cavity and maintained at a certain distance from other components such as the six-axis inertial sensing unit 22 to avoid mutual interference. Temperature sensor 23 can also employ newer sensors such as fiber optic grating temperature sensors to improve measurement accuracy and reliability.

[0026] The pressure sensor 24 is used to collect downhole pressure parameters. It can be made using principles such as piezoresistive or capacitive sensors and can monitor downhole pressure changes in real time. The pressure sensor 24 is installed in a suitable position within the mounting cavity and connected to the outside environment through a pressure transmission channel opened on the annular housing 21 to accurately measure downhole pressure. The pressure sensor 24 can also be other types of sensors such as a vibrating wire pressure sensor.

[0027] In addition, a memory 25 and a power supply are fixed inside the mounting cavity. Considering the difficulty of real-time high-speed transmission of downhole data, this invention preferably uses a high-temperature resistant, impact-resistant, and large-capacity memory 25 to store all the raw high-frequency data collected by various sensors during drilling. After drilling is completed, the intelligent sensing module 2 is pulled out with the drill string, and the data in the memory 25 is then retrieved for subsequent analysis via a wired or wireless interface. Simultaneously, an independent power supply module is also provided inside the mounting cavity. This power supply is preferably a high-temperature resistant rechargeable battery 26, which can provide a stable and continuous power supply for the entire intelligent sensing module 2 in the harsh downhole environment. Before each downhole operation, the rechargeable battery 26 can be charged on the surface, enabling the module to be reused and reducing operating costs. In other embodiments, the rechargeable battery 26 can also be self-powered by downhole turbine generators or other methods, generating electricity through the flow of drilling fluid.

[0028] These components of the intelligent sensing module 2 work together to continuously collect downhole vibration data and environmental parameters of the drill bit during the drilling process at a high sampling frequency, providing a reliable data source for subsequent analysis.

[0029] Please refer to Figure 1 This invention provides a downhole information interpretation method based on drill bit vibration data, including data acquisition, synchronization alignment, drift calibration and feature extraction, comparison analysis and identification. Specifically, by directly collecting data near the drill bit, using preset feature events for time synchronization, and calibrating and extracting features from the data, high-fidelity bottom-hole vibration signals are obtained. These signals are then compared with a pre-built formation-vibration spectrum feature library to achieve accurate and high-resolution identification of formation lithology. Furthermore, analysis using a pre-built machine learning model achieves accurate and high-resolution identification of drilling conditions. This is because directly collecting data at the drill bit source ensures the authenticity and integrity of the information; precise time synchronization and data calibration improve analysis accuracy; and the combination of feature comparison and model analysis allows for in-depth information interpretation from multiple dimensions.

[0030] Includes the following steps: S1. The drilling parameters are acquired through the ground recording system, and the downhole vibration data and environmental parameters are acquired through the intelligent sensing module integrated into the drill bit water hole. The environmental parameters include temperature parameters acquired through the temperature sensor and pressure parameters acquired through the pressure sensor. S2. Using preset characteristic events in the drilling process as time anchors, synchronize and align the time series of downhole vibration data and the time series of drilling parameters to obtain downhole multidimensional vibration data; Synchronization alignment involves synchronizing the time series x(t) of downhole vibration data with the time series y(t) of drilling parameters collected by the surface recording system, using preset characteristic events in the drilling process as time anchors.

[0031] Specifically, the preset characteristic event is the drilling stoppage period when a single drill pipe is connected. During this period, downhole vibration will significantly decrease, and parameters such as surface rotation speed and drilling pressure will also show obvious characteristics. The data processing unit uses a sliding cross-correlation algorithm to find the time offset τ that maximizes the similarity between the downhole vibration signal sequence x(t) and the surface parameter signal sequence y(t). The sliding cross-correlation function is defined as:

[0032] Here, x(t) represents the time series of downhole vibration data, y(t) represents the time series of surface drilling parameters, τ is the time offset, and t is the time variable. This function calculates precise time correction values, unifying the time axes of the downhole and surface datasets. Specifically, y(t+τ) is used to shift the surface signal y(t) by a time increment τ to align it with the downhole signal x(t). By systematically "sliding" (i.e., changing τ) and calculating the similarity at each location, we can ultimately find the time difference that best matches the two signals, thus achieving synchronization. It should be understood that other events with distinct characteristics can also be used as time anchors, such as drill bit changes.

[0033] S3. Perform drift calibration or noise compensation on downhole multidimensional vibration data using temperature parameters, and perform signal processing and feature extraction on the calibrated and compensated downhole multidimensional vibration data to obtain one or more vibration feature parameter vectors reflecting the interaction between the drill bit and the formation. Specifically, methods for drift calibration or noise compensation of downhole multidimensional vibration data using temperature parameters include: adjusting the original measured values ​​A of the downhole multidimensional vibration data using a pre-calibrated temperature-drift polynomial model based on the temperature parameters. 原始 Compensation is performed to obtain the corrected measurement value A. 校正 The compensation formula is: A 校正 =A 原始 −(k2(T−T ref ) 2 +k1(T−T ref )) Where T is the temperature parameter, T ref To calibrate the reference temperature, k1 and k2 are the pre-calibrated first-order and second-order temperature drift coefficients. This eliminates the influence of the high-temperature downhole environment on the sensor accuracy. Other more complex calibration models can also be used to improve calibration accuracy.

[0034] Methods for signal processing of calibrated and compensated downhole multidimensional vibration data include: using Fast Fourier Transform (FFT) to process the discrete calibrated and compensated downhole multidimensional vibration data x. n Complex spectrum value X converted to the frequency domain k The transformation relationship is defined by the following formula: Where, x n For discrete, calibrated, compensated downhole multidimensional vibration data, X k Here, represents the complex spectrum value in the frequency domain, N is the number of sampling points, n is the sampling point index of the time series, k is the frequency component index in the frequency domain, and j is the imaginary unit. The FFT can be used to extract features such as the dominant frequency, harmonic distribution, and energy proportion of each frequency band. Other frequency domain transformation methods, such as the discrete cosine transform, can also be used.

[0035] The method for feature extraction from calibrated and compensated downhole multidimensional vibration data includes: using wavelet transform to perform time-frequency analysis on non-stationary downhole multidimensional vibration data to identify the local time-frequency characteristics of transient shocks and periodic modes. Specifically, this involves multi-level decomposition of the vibration data to obtain multiple sub-band signals with different frequency ranges; and calculating the energy E of each sub-band signal. i The calculation formula is as follows:

[0036] Among them, E i Let i be the energy of the i-th sub-band signal. Let M be the k-th discrete sampling point of the i-th sub-band signal, and M be the number of sampling points for that sub-band signal. The calculated energy of each sub-band forms a time-frequency feature vector, used to identify transient impacts and periodic patterns. By comparing the high-frequency sub-band energy in the time-frequency feature vector with a preset impact threshold, a transient impact is identified when the high-frequency sub-band energy exceeds the threshold within a short period. By analyzing the energy distribution in the time-frequency feature vector, a periodic pattern is identified when the proportion of energy in a certain sub-band consistently exceeds a preset concentration threshold. Other time-frequency analysis methods, such as the Hilbert-Huang transform, can also be used.

[0037] S4. Match the vibration feature parameter vector with the formation-vibration spectrum feature library to determine the lithology of the current drill bit; at the same time, extract pressure-derived features from the pressure parameters, and combine the vibration feature parameter vector with the pressure-derived features to form a fused feature vector, and use a machine learning model to identify the drilling conditions experienced by the current drill bit.

[0038] Formation lithology identification involves matching real-time calculated vibration characteristic parameter vectors with a pre-established formation-vibration spectrum feature database to determine the formation lithology of the current drill bit location. The method for establishing the formation-vibration spectrum feature database includes: segmenting and correlating historical downhole vibration data from the drilling process with known geological logging data according to drill bit depth to determine each formation segment;

[0039] Extract and store typical vibration characteristic parameter vectors that correspond one-to-one with each type of stratum lithology, as identification templates in the library.

[0040] Specifically, by linking a large amount of historical data with known geological information, the database stores "vibration fingerprints" corresponding to different lithologies such as sandstone and mudstone. By finding the template with the highest similarity, the lithology of the currently drilled strata can be determined. Machine learning algorithms can also be used to optimize and update the feature database to improve the accuracy of identification.

[0041] Identifying drilling conditions involves combining a vibration characteristic parameter vector with pressure-derived features extracted from the pressure parameter P to form a fused feature vector. The pressure-derived features include the pressure fluctuation variance, calculated using the following formula:

[0042] Among them, P i For pressure sampling points within the time window, μ p The average pressure within the window is denoted by N, and N is the number of sampling points within the window.

[0043] The method for forming a fused feature vector by combining vibration feature parameter vectors and pressure-derived features includes: extracting m vibration features from downhole multidimensional vibration data to form a vibration feature parameter vector of dimension m; extracting n pressure-derived features from pressure parameters to form a pressure-derived feature vector of dimension n; before splicing, normalizing the vibration feature parameter vector and the pressure-derived feature vector to eliminate the dimensional influence between different features; and splicing the normalized pressure-derived feature vector into the normalized vibration feature parameter vector in terms of dimension to form a fused feature vector of dimension m + n.

[0044] The fused feature vector is input into a pre-trained machine learning model, which outputs a classification result for the current drilling condition. The machine learning model is a pre-trained support vector machine (SVM) model, which classifies the input fused feature vector x using the following decision function f(x):

[0045] Where f(x) is the decision function output of the SVM model, sgn is the sign function, and x is the input fusion feature vector. i For support vectors, K is the kernel function, and α i is the Lagrange multiplier, b is the bias term, l is the total number of support vectors, and i is the index.

[0046] The implementation principle of this invention is as follows: By integrating an intelligent sensing module near the drill bit, high-fidelity downhole vibration data and environmental parameters are collected and stored during drilling. After the drill string is retrieved, precise time synchronization alignment, data drift calibration, and depth feature extraction are performed, and the data is compared and analyzed with a pre-built feature model. This enables high-precision retrospective analysis and identification of formation lithology and drilling conditions throughout the entire drilling process. Compared with existing technologies, this invention can obtain more accurate and comprehensive bottom-hole source information, providing a more reliable basis for subsequent engineering summaries, geological model corrections, and future drilling decision optimization, thus helping to improve the overall efficiency and safety level of drilling projects.

[0047] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the invention by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the invention should be included within the scope of protection of the invention.

Claims

1. A downhole information interpretation system based on drill bit vibration data, characterized in that, include: The drill bit (1) has an integrated smart sensor module (2) installed in the water inlet (11) of the drill bit (1). The smart sensor module (2) includes a mounting housing (21) for fixed connection with the water inlet (11), a six-axis inertial sensing unit (22) for collecting vibration data, a temperature sensor (23) for collecting downhole temperature parameters, and a pressure sensor (24) for collecting downhole pressure parameters. The six-axis inertial sensing unit (22), the temperature sensor (23), and the pressure sensor (24) are all integrated in the mounting housing (21).

2. The downhole information interpretation system based on drill bit vibration data according to claim 1, characterized in that, The mounting housing (21) is an annular housing. The mounting housing (21) includes an outer shell (212) and an inner shell (211). The outer shell (212) is fixedly connected to the water eye (11), and the outer shell (212) is fixedly connected to the inner shell (211), forming a mounting cavity for mounting the six-axis inertial sensing unit (22), temperature sensor (23), and pressure sensor (24). The shape and size of the outer shell (212) match the water eye (11).

3. A method for interpreting downhole information based on drill bit vibration data, characterized in that, include: S1. The drilling parameters are acquired through the ground recording system, and the downhole vibration data and environmental parameters are acquired through the intelligent sensing module integrated into the drill bit water hole. The environmental parameters include temperature parameters acquired through the temperature sensor and pressure parameters acquired through the pressure sensor. S2. Using preset characteristic events in the drilling process as time anchors, synchronize and align the time series of downhole vibration data and the time series of drilling parameters to obtain downhole multidimensional vibration data; S3. Perform drift calibration or noise compensation on downhole multidimensional vibration data using temperature parameters, and perform signal processing and feature extraction on the calibrated and compensated downhole multidimensional vibration data to obtain one or more vibration feature parameter vectors reflecting the interaction between the drill bit and the formation. S4. Match the vibration feature parameter vector with the formation-vibration spectrum feature library to determine the lithology of the current drill bit; at the same time, extract pressure-derived features from the pressure parameters, and combine the vibration feature parameter vector with the pressure-derived features to form a fused feature vector, and use a machine learning model to identify the drilling conditions experienced by the current drill bit.

4. The downhole information interpretation method based on drill bit vibration data according to claim 3, characterized in that, The preset feature event in step S2 is the drilling stop period when a single drill pipe is connected. Synchronization alignment is achieved by calculating the time offset and maximizing the sliding cross-correlation function between the downhole and surface reference signals. The sliding cross-correlation function is defined as follows: Where x(t) is the time series of downhole vibration data, y(t) is the time series of surface drilling parameters, τ is the time offset, and t is the time variable.

5. The downhole information interpretation method based on drill bit vibration data according to claim 1, characterized in that, Methods for drift calibration or noise compensation of downhole multidimensional vibration data using temperature parameters include: calibrating the original measured value A of the downhole multidimensional vibration data using a pre-calibrated temperature-drift polynomial model based on the temperature parameters. 原始 Compensation is performed to obtain the corrected measurement value A. 校正 The compensation formula is: A 校正 =A 原始 −(k2(T−T ref ) 2 +k1(T−T ref )) Where T is the temperature parameter, T ref To calibrate the reference temperature, k1 and k2 are the first-order and second-order temperature drift coefficients obtained before calibration.

6. The downhole information interpretation method based on drill bit vibration data according to claim 1, characterized in that, Methods for signal processing of calibrated and compensated downhole multidimensional vibration data include: using Fast Fourier Transform to process the discrete calibrated and compensated downhole multidimensional vibration data x n Complex spectrum value X converted to the frequency domain k The transformation relationship is defined by the following formula: Where, x n For discrete, calibrated, compensated downhole multidimensional vibration data, X k denoted as the complex spectrum value in the frequency domain, where N is the number of sampling points, n is the sampling point index of the time series, k is the frequency component index in the frequency domain, and j is the imaginary unit.

7. The downhole information interpretation method based on drill bit vibration data according to claim 1, characterized in that, Methods for feature extraction from calibrated and compensated downhole multidimensional vibration data include: using wavelet transform to perform time-frequency analysis on non-stationary downhole multidimensional vibration data to identify the local time-frequency characteristics of transient shocks and periodic patterns, specifically including: Multi-level decomposition of downhole multi-dimensional vibration data yields multiple sub-band signals with different frequency ranges; Calculate the energy E of each of the multiple sub-band signals. i The calculation formula is as follows: Among them, E i Let i be the energy of the i-th sub-band signal. Let M be the k-th discrete sampling point of the i-th sub-band signal, and M be the total number of sampling points of the sub-band signal. The calculated energy of each sub-band is used to construct a time-frequency feature vector, which is used to identify transient impacts and periodic modes. The process of identifying transient impacts and periodic modes includes: By comparing the high-frequency sub-band energy in the time-frequency feature vector with a preset impact threshold, when the high-frequency sub-band energy exceeds the impact threshold in a short period of time, it is identified as a transient impact. By analyzing the energy distribution in the time-frequency feature vector, when the proportion of energy in a certain sub-band in the total energy continuously exceeds the preset concentration threshold, it is identified as a periodic pattern.

8. The downhole information interpretation method based on drill bit vibration data according to claim 1, characterized in that, Methods for establishing a stratigraphic-vibration spectrum feature library include: By segmenting and correlating downhole vibration data from historical drilling processes with known geological logging data according to drill bit depth, the stratigraphic segments in each region can be determined. Extract and store typical vibration characteristic parameter vectors that correspond one-to-one with each type of stratum lithology, as identification templates in the library.

9. The downhole information interpretation method based on drill bit vibration data according to claim 4, characterized in that, Pressure-derived characteristics include pressure fluctuation variance The calculation formula is as follows: Among them, P i For pressure sampling points within the time window, μ p The average pressure within the window is given by N, where N is the number of sampling points within the window, and i is the index. Methods for forming a fused feature vector by combining vibration characteristic parameter vectors with pressure-derived features include: The m vibration features extracted from the downhole multidimensional vibration data are used to form a vibration feature parameter vector with dimension m. The n pressure-derived features extracted from the pressure parameters are used to form a pressure-derived feature vector with dimension n. Before splicing, the vibration characteristic parameter vector and the pressure derived characteristic vector are first normalized to eliminate the dimensional influence between different features. By concatenating the normalized pressure-derived feature vector into the normalized vibration feature parameter vector in terms of dimension, a fused feature vector with dimension m+n is formed, and this fused feature vector is used as the input of the machine learning model.

10. The downhole information interpretation method based on drill bit vibration data according to claim 1, characterized in that, The machine learning model is a pre-trained support vector machine model, which classifies the input fused feature vector x using the following decision function f(x): Where f(x) is the decision function output of the SVM model, sgn is the sign function, and x is the input fusion feature vector. i For support vectors, K is the kernel function, and α i is the Lagrange multiplier, b is the bias term, l is the total number of support vectors, and i is the index.

Images