Method and device for predicting stick-slip vibration of drill string
By combining traditional physical models with data-driven machine learning methods, the state space model is constructed and the XGBoost algorithm is used to solve the problem of difficult to predict the stick-slip vibration of the drill string in the existing technology with high accuracy, achieving more efficient and safe drilling operations.
Patent Information
- Application Number
- CN202311412300.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-10-27
- Publication Date
- 2025-04-29
AI Technical Summary
The prior art is difficult to predict and suppress the stick-slip vibration of the drill string with high accuracy, resulting in damage to the drill bit cutting tool, fatigue damage to the underground drilling tool, and may even cause the drilling tool to fall off, seriously affecting the drilling cycle and safety.
By obtaining the preprocessed well-measuring data, the training set and test set are divided by SPXY algorithm, the state space model is constructed and trained, and the optimal prediction model is finally constructed using the XGBoost algorithm to predict the prediction space of the stick-slip vibration of the drill string.
The generalization and interpretability of the prediction of the viscoscopic vibration of the drill string are improved, and the drilling pressure and speed combination values of the critical viscoscopic vibration are determined, which enhances the safety and efficiency of drilling operations.
Smart Images
Figure CN119903424A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of oil drilling, and in particular to a method and a device for predicting stick-slip vibration of a drill string. Background Art
[0002] During the drilling process, when the energy generated by the rotary system of the drilling rig cannot reach the drill bit, the drill string is inevitably affected by the friction of the well wall, causing the actual speed of the downhole drill string to slow down or even stop. This continuous stick-slip vibration will cause damage to the drill bit cutting tool and fatigue damage to the downhole drill tool, and may even cause the weak part of the drill tool to break and fall off, seriously affecting the drilling cycle and drilling safety. Therefore, in order to ensure the safety and efficiency of drilling operations, it is particularly important to predict and effectively suppress the stick-slip vibration of the downhole drill string.
[0003] The existing methods for predicting drill string stick-slip vibration mainly include vibration information measurement method and physical model analysis method. The vibration information measurement method mainly measures the vibration information of the drill string to understand the real characteristics of the drill string vibration, determine the stick-slip vibration of the drill string, and further study the influencing factors of the stick-slip vibration of the drill string. This method requires the use of highly sensitive sensors to obtain the vibration state of the drill string, and the investment cost is relatively high. The physical model analysis method can preliminarily simulate the stick-slip vibration phenomenon of the drill string using a simple physical model, and analyze the influence of related parameters on the stick-slip vibration. This method cannot accurately determine the critical values of drilling pressure and rotation speed that suppress the stick-slip vibration of the drill string. Summary of the invention
[0004] The present invention provides a method and device for predicting stick-slip vibration of a drill string, which are used to overcome the limitations of physical models that require calibration parameters and data-driven models such as limited training data or lack of physical knowledge, and effectively improve drilling efficiency.
[0005] In a first aspect, the present invention provides a method for predicting stick-slip vibration of a drill string, comprising:
[0006] Acquire the preprocessed logging data and determine the stick-slip level of the logging data;
[0007] The SPXY algorithm is used to divide the logging data into a training set and a test set according to a preset ratio; the training set includes the logging training data and its corresponding stick-slip level labels; the test set includes the logging test data and its corresponding stick-slip level labels;
[0008] The constructed state space model is trained by the well logging training data and the corresponding stick-slip level labels, and a target state space model is obtained after being tested by the well logging test data and the corresponding stick-slip level labels; the state space model is composed of a state transition model and an observation model, and both the state transition model and the observation model have several combination types;
[0009] By using the XGBoost algorithm, the stick-slip degree prediction result output by the target state space physical model is used as the input data of the constructed optimal prediction network to obtain the target optimal prediction model;
[0010] The acquired mud logging data to be analyzed is used, and based on the target optimal prediction model, a stick-slip vibration prediction space corresponding to the mud logging data to be analyzed is constructed.
[0011] Optionally, obtaining the preprocessed well logging data and determining the stick-slip level of the well logging data includes:
[0012] Dividing the logging data into stick-slip cycles of preset multiples;
[0013] Determine the ratio between the difference between the maximum drilling speed and the minimum drilling speed of the divided logging data in the preset multiple stick-slip cycles and the average drilling speed;
[0014] The ratio is used as a measure of the stick-slip level SSI to determine the stick-slip level of the divided logging data.
[0015] Optionally, obtaining the preprocessed well logging data and determining the stick-slip level of the well logging data further includes:
[0016] The acquired logging data are subjected to wavelet decomposition by using the preset wavelet basis function and decomposition scale to obtain wavelet decomposition coefficients at different scales;
[0017] Performing threshold processing on the wavelet decomposition coefficients to obtain processed wavelet parameters;
[0018] The processed wavelet parameters are subjected to inverse wavelet transform to reconstruct the de-noised well logging data.
[0019] Optionally, by using the XGBoost algorithm, the stick-slip degree prediction result output by the target state space physical model is used as input data of the constructed optimal prediction network to obtain the target optimal prediction model, including:
[0020] Determine the stick-slip degree prediction results under different combination types obtained by inputting the logging data into the target state space physical model;
[0021] The stick-slip degree prediction results under different combination types are trained using the XGBoost algorithm to obtain the target optimal prediction model.
[0022] Optionally, the acquired mud logging data to be analyzed and based on the target optimal prediction model, a stick-slip vibration prediction space corresponding to the mud logging data to be analyzed is constructed, including:
[0023] Determine the combination value of rotation speed and drilling pressure in the logging data to be analyzed, and form a corresponding grid;
[0024] Inputting the optimal prediction model with the combined value to obtain a corresponding prediction result;
[0025] By marking colors of all the grids according to corresponding prediction results, a stick-slip vibration prediction space corresponding to the mud logging data to be analyzed is constructed.
[0026] In a second aspect, the present invention provides a device for predicting stick-slip vibration of a drill string, comprising:
[0027] An acquisition module, used for acquiring the pre-processed well logging data and determining the stick-slip level of the well logging data;
[0028] A data set division module is used to divide the well logging data into a training set and a test set according to a preset ratio by using an SPXY algorithm; the training set includes the well logging training data and its corresponding stick-slip level labels; the test set includes the well logging test data and its corresponding stick-slip level labels;
[0029] a target state space model determination module, used to train the constructed state space model through the well logging training data and its corresponding stick-slip level labels, and obtain the target state space model after testing the well logging test data and its corresponding stick-slip level labels; the state space model is composed of a state transition model and an observation model, and both the state transition model and the observation model have several combination types;
[0030] An optimal prediction module determination module is used to obtain a target optimal prediction model by using the stick-slip degree prediction result output by the target state space physical model as input data of the constructed optimal prediction network through the XGBoost algorithm;
[0031] The prediction module is used to acquire the logging data to be analyzed and construct a stick-slip vibration prediction space corresponding to the logging data to be analyzed based on the target optimal prediction model.
[0032] Optionally, the acquisition module includes:
[0033] A division submodule, used for dividing the logging data into stick-slip cycles of preset multiples;
[0034] A ratio determination submodule, used to determine the ratio between the difference between the maximum drilling speed and the minimum drilling speed of the divided logging data in the preset multiple stick-slip cycle and the average drilling speed;
[0035] The level determination submodule is used to determine the stick-slip level of the divided logging data by using the ratio as a measure of the stick-slip level SSI.
[0036] Optionally, the acquisition module further includes:
[0037] The decomposition submodule is used to perform wavelet decomposition on the acquired logging data through a preset wavelet basis function and decomposition scale to obtain wavelet decomposition coefficients at different scales;
[0038] A threshold processing submodule, used for performing threshold processing on the wavelet decomposition coefficients to obtain processed wavelet parameters;
[0039] The inverse transformation submodule is used to perform inverse wavelet transformation on the processed wavelet parameters to reconstruct the denoised well logging data.
[0040] In a third aspect, the present application provides an electronic device, comprising a processor and a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the steps in the method provided in the first aspect are performed.
[0041] In a fourth aspect, the present application provides a storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps in the method provided in the first aspect are performed.
[0042] It can be seen from the above technical solutions that the present invention has the following advantages:
[0043] The present invention provides a method and device for predicting stick-slip vibration of a drill string. The method comprises: obtaining preprocessed logging data and determining the stick-slip degree of the logging data; using an SPXY algorithm to divide the logging data into a training set and a test set according to a preset ratio; the training set comprises the logging training data and its corresponding stick-slip degree label; the test set comprises the logging test data and its corresponding stick-slip degree label; training a constructed state space model through the logging training data and its corresponding stick-slip degree label, and obtaining a target state space model after testing the logging test data and its corresponding stick-slip degree label; the state space model consists of a state transition model and an observation model, and both the state transition model and the observation model have several combination types; using an XGBoost algorithm, using the stick-slip degree prediction result output by the target state space physical model as input data of the constructed optimal prediction network to obtain a target optimal prediction model; acquiring the logging data to be analyzed, and constructing a stick-slip vibration prediction space corresponding to the logging data to be analyzed based on the target optimal prediction model. Combining traditional physical models with data-driven machine learning methods overcomes the limitations of physical models that require parameter calibration and data-driven models such as limited training data or lack of physical knowledge, thereby improving the generalization and interpretability of prediction results, further determining the critical drilling pressure and speed combination values for stick-slip vibration, and accelerating the process of oil and gas exploration and development towards efficient and safe development. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0045] Figure 1 This is a flow chart of a method for predicting stick-slip vibration of a drill string according to a first embodiment of the present invention;
[0046] Figure 2 This is a flow chart of a second embodiment of a method for predicting stick-slip vibration of a drill string according to the present invention;
[0047] Figure 3 It is a schematic diagram of a three-dimensional prediction space of stick-slip vibration under a combination of rotation speed and bit pressure of a drill string stick-slip vibration prediction method according to a second embodiment of the present invention;
[0048] Figure 4It is a schematic diagram of a two-dimensional prediction space of stick-slip vibration under a combination of rotation speed and bit pressure of a drill string stick-slip vibration prediction method according to a second embodiment of the present invention;
[0049] Figure 5 This is a stick-slip degree distribution diagram of Example 2 of a method for predicting stick-slip vibration of a drill string according to the present invention under different drilling pressure and rotation speed combination values;
[0050] Figure 6 The present invention is a structural block diagram of an embodiment of a device for predicting stick-slip vibration of a drill string according to the present invention. DETAILED DESCRIPTION
[0051] The embodiments of the present invention provide a method and device for predicting stick-slip vibration of a drill string, which are used to overcome the limitations of physical models that require calibration of parameters and data-driven models such as limited training data or lack of physical knowledge, and effectively improve drilling efficiency.
[0052] In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0053] For example, see Figure 1 , Figure 1 The flowchart of the first embodiment of a method for predicting stick-slip vibration of a drill string of the present invention includes:
[0054] S101, obtaining pre-processed logging data and determining the stick-slip level of the logging data;
[0055] S102, using the SPXY algorithm to divide the well logging data into a training set and a test set according to a preset ratio; the training set includes the well logging training data and its corresponding stick-slip level labels; the test set includes the well logging test data and its corresponding stick-slip level labels;
[0056] S103, training the constructed state space model through the well logging training data and its corresponding stick-slip level labels, and obtaining a target state space model after testing through the well logging test data and its corresponding stick-slip level labels; the state space model is composed of a state transition model and an observation model, and both the state transition model and the observation model have several combination types;
[0057] S104, using the stick-slip degree prediction result output by the target state space physical model as input data of the constructed optimal prediction network through the XGBoost algorithm, to obtain a target optimal prediction model;
[0058] S105, acquiring the mud logging data to be analyzed, and constructing a stick-slip vibration prediction space corresponding to the mud logging data to be analyzed based on the target optimal prediction model.
[0059] A method for predicting stick-slip vibration of a drill string provided by an embodiment of the present invention comprises the following steps: obtaining preprocessed logging data and determining the stick-slip level of the logging data; dividing the logging data into a training set and a test set according to a preset ratio by using an SPXY algorithm; the training set comprises the logging training data and its corresponding stick-slip level labels; the test set comprises the logging test data and its corresponding stick-slip level labels; training a constructed state space model by using the logging training data and its corresponding stick-slip level labels, and obtaining a target state space model after testing the logging test data and its corresponding stick-slip level labels; the state space model comprises a state transition model and an observation model, and both the state transition model and the observation model have several combination types; obtaining a target optimal prediction model by using an XGBoost algorithm and using a stick-slip level prediction result output by the target state space physical model as input data of the constructed optimal prediction network; obtaining the logging data to be analyzed, and constructing a stick-slip vibration prediction space corresponding to the logging data to be analyzed based on the target optimal prediction model. Combining traditional physical models with data-driven machine learning methods overcomes the limitations of physical models that require parameter calibration and data-driven models such as limited training data or lack of physical knowledge, thereby improving the generalization and interpretability of prediction results, further determining the critical drilling pressure and speed combination values for stick-slip vibration, and accelerating the process of oil and gas exploration and development towards efficient and safe development.
[0060] For example 2, please refer to Figure 2 , Figure 2 This is a flow chart of a second embodiment of a method for predicting stick-slip vibration of a drill string according to the present invention, the method comprising:
[0061] S201, obtaining pre-processed logging data;
[0062] In the embodiment of the present invention, the logging data are selected from gamma rays at different depths, wellbores and times, and the logging data are drilling pressure, rotation speed, torque and drilling speed.
[0063] In an optional embodiment, obtaining the pre-processed well logging data includes:
[0064] The acquired logging data are subjected to wavelet decomposition by using the preset wavelet basis function and decomposition scale to obtain wavelet decomposition coefficients at different scales;
[0065] Performing threshold processing on the wavelet decomposition coefficients to obtain processed wavelet parameters;
[0066] The processed wavelet parameters are subjected to inverse wavelet transform to reconstruct the denoised well logging data, wherein the denoised well logging data is the processed well logging data.
[0067] Due to the possibility of logging instrument failure or wellbore collapse during oil and gas exploration and development, the logging data may be noisy or missing. In order to ensure the accuracy of subsequent modeling, the original logging data needs to be preprocessed to improve the authenticity and effectiveness of the data.
[0068] The preprocessing process of the embodiment of the present invention is: first identify and delete outliers and missing values, and then use the nonlinear wavelet transform threshold method to denoise the logging data. The specific calculation process is:
[0069] (1) Through the pre-set wavelet basis function And the decomposition scale, the original signal f(t) is decomposed by wavelet to obtain a at the jth scale j The wavelet decomposition coefficient w(a j ,τ), that is:
[0070]
[0071] Among them, a j is the wavelet transform scale parameter, and τ is the wavelet transform translation parameter. j The wavelet decomposition coefficients calculated under different τ are represented by the vector w j express.
[0072] (2) Choose a suitable threshold λ j For the wavelet coefficient w j Threshold all elements in to get w' j ,Right now:
[0073] λ j =f×max(w j ),f∈(0,1);
[0074]
[0075] Among them, max(w j ) is the wavelet coefficient w j The maximum value under jn is the wavelet coefficient w jThe value of the nth element in after thresholding.
[0076] (3) The processed wavelet coefficient w' j Perform inverse wavelet transform to complete signal reconstruction and obtain the denoised signal f(t)'.
[0077] S202, dividing the logging data according to preset multiples of stick-slip cycles;
[0078] S203, determining the ratio between the difference between the maximum drilling speed and the minimum drilling speed of the divided logging data in the preset multiple stick-slip cycles and the average drilling speed;
[0079] S204, using the ratio as a measure of stick-slip level SSI, to determine the stick-slip level of the divided logging data;
[0080] In the embodiment of the present invention, the ratio of the maximum and minimum drilling speed differences generated by a preset multiple of stick-slip cycles (at least one complete stick-slip cycle) to the average drilling speed is used as a measure of the stick-slip degree SSI: wherein the determination expression of SSI is:
[0081]
[0082] Among them, max (bitRPM) is the maximum drilling speed difference, min (bitRPM) is the minimum drilling speed difference, is the average drilling speed.
[0083] S205, using the SPXY algorithm to divide the well logging data into a training set and a test set according to a preset ratio; the training set includes the well logging training data and its corresponding stick-slip level labels; the test set includes the well logging test data and its corresponding stick-slip level labels;
[0084] In the embodiment of the present invention, after the calculated SSI value is divided into four stick-slip degree levels of low, medium, high and severe, the original data set is divided into a training set and a test set in a ratio of 3:1 or 4:1 using the SPXY algorithm.
[0085] S206, training the constructed state space model through the well logging training data and its corresponding stick-slip level labels, and obtaining a target state space model after testing through the well logging test data and its corresponding stick-slip level labels; the state space model is composed of a state transition model and an observation model, and both the state transition model and the observation model have several combination types;
[0086] In an embodiment of the present invention, the state transition model selects physical equations, fully connected neural network (Fully Connected Neural Network, FCNN), long short-term memory network (Long Short-Term Memory, LSTM), and Markov Recurrent Neural Network (Markov Recurrent Neural Network, MRNN).
[0087] The physical equations are a set of equations that combine the drill string dynamics model and the drill bit-rock interaction model, as shown below:
[0088] dt(t)=dt(t)·10 -5 ;
[0089]
[0090] α(t+1)=(ω(t)-RPM(t))·dt(t)+α(t);
[0091]
[0092]
[0093] depth(t+1)=a2·RPM(t)·WOB(t)·dt(t)+depth(t);
[0094] gamma(t+1)=a3·RPM(t)·WOB(t)·depth(t)·dt(t)+gamma(t);
[0095] y=|(a4·RPM(t)-a5·ω(t))·a6|;
[0096] Among them, μ is the friction coefficient, ω is the downhole rotation speed, α=ω-RPM, and y is the degree of stick-slip vibration.
[0097] The output of the state transition model is used as the input of the observation model.
[0098] The observation models selected are physical equation, XGBoost, and FCNN respectively.
[0099] According to the above state transition model and observation model, five state space physical models are constructed as follows:
[0100] (1) State space model 1: Both the state transition and observation models use physical equations.
[0101] (2) State space model 2: The state transition model uses physical equations, and the observation model uses the XGBoost algorithm.
[0102] (3) State space model 3: FCNN is used for both state transition and observation models.
[0103] (4) State space model 4: The state transition model uses LSTM and the observation model uses FCNN.
[0104] (5) State space model 5: The state transition model uses MRNN and the observation model uses FCNN.
[0105] S207, determining the stick-slip degree prediction results under different combination types obtained by inputting the logging data into the target state space physical model;
[0106] S208, using the XGBoost algorithm to train the stick-slip degree prediction results under different combination types to obtain the target optimal prediction model;
[0107] In an embodiment of the present invention, the prediction results of the above five model combinations are used as inputs to a data-driven machine learning model, ultimately achieving high-precision prediction of the stick-slip vibration degree of the drill string.
[0108] In the specific implementation, the machine learning algorithm uses XGBoost. The "integration" idea of this algorithm can integrate multiple weakly supervised models into a strong supervised model to provide performance that is better than any weakly supervised model, which helps to reduce prediction variance and improve the generalization and robustness of the model.
[0109] The prediction performance of all constructed models for low and severe stick-slip vibration levels was compared and the results are shown below:
[0110]
[0111] The comparison results show that the performance of the machine learning method guided by the physical model is better than that of the algorithm with a single physical model. The XGBoost model has the highest accuracy in predicting the degree of low stick-slip vibration, and the model established by the invented method has the highest accuracy in predicting the degree of severe stick-slip vibration.
[0112] S209, determining the combination value of the rotation speed and the drilling pressure in the logging data to be analyzed, and forming a corresponding grid;
[0113] S210, inputting the optimal prediction model with the combined value to obtain a corresponding prediction result;
[0114] S211, constructing a stick-slip vibration prediction space corresponding to the mud logging data to be analyzed by marking colors of all the grids according to corresponding prediction results.
[0115] In the embodiment of the present invention, based on the trained model, the rotation speed and drilling pressure combination value are used as the initial condition to construct a three-dimensional or two-dimensional space, such as Figure 3 The second embodiment of the method for predicting stick-slip vibration of a drill string of the present invention is shown in the schematic diagram of the three-dimensional prediction space of stick-slip vibration under the combination of rotation speed and drilling pressure. In the three-dimensional space, X and Y are the rotation speed and drilling pressure respectively, and the clock is the time step. Figure 4 Embodiment 2 of a method for predicting stick-slip vibration of a drill string of the present invention is shown in a schematic diagram of a two-dimensional prediction space of stick-slip vibration under a combination of rotation speed and drilling pressure. In the two-dimensional space, X and Y are rotation speed and drilling pressure, respectively. Each cell is a combination of rotation speed and drilling pressure, and the color indicates the stick-slip degree. The darker the color, the more serious the stick-slip degree. Based on this grid, exploration and development personnel can be guided to identify areas with a high possibility of stick-slip.
[0116] A stick-slip vibration prediction space was created for all test wells, and a safe operable area was obtained based on the rotation speed and bit pressure, such as Figure 5 Embodiment 2 of a method for predicting stick-slip vibration of a drill string of the present invention shows that in the upper left corner of the stick-slip degree distribution diagram under different drilling pressure and rotation speed combination values, high drilling pressure and low rotation speed can effectively reduce the stick-slip vibration degree.
[0117] A method for predicting stick-slip vibration of a drill string provided by an embodiment of the present invention comprises the following steps: obtaining preprocessed logging data and determining the stick-slip level of the logging data; dividing the logging data into a training set and a test set according to a preset ratio by using an SPXY algorithm; the training set comprises the logging training data and its corresponding stick-slip level labels; the test set comprises the logging test data and its corresponding stick-slip level labels; training a constructed state space model by using the logging training data and its corresponding stick-slip level labels, and obtaining a target state space model after testing the logging test data and its corresponding stick-slip level labels; the state space model comprises a state transition model and an observation model, and both the state transition model and the observation model have several combination types; obtaining a target optimal prediction model by using an XGBoost algorithm and using a stick-slip level prediction result output by the target state space physical model as input data of the constructed optimal prediction network; obtaining the logging data to be analyzed, and constructing a stick-slip vibration prediction space corresponding to the logging data to be analyzed based on the target optimal prediction model. The traditional physical model is combined with the data-driven machine learning method. Through the XGBoost algorithm, the optimal combination corresponding to the logging data to be analyzed is selected from different combination types for prediction, overcoming the limitations of physical models that require parameter calibration and data-driven models such as limited training data or lack of physical knowledge, thereby improving the generalization and interpretability of the prediction results, further determining the critical drilling pressure and speed combination values for stick-slip vibration, and accelerating the process of oil and gas exploration and development towards efficient and safe development.
[0118] For example 3, please refer to Figure 6 , Figure 6 The structural block diagram of an embodiment of a prediction device for stick-slip vibration of a drill string according to the present invention comprises:
[0119] An acquisition module 501 is used to acquire the pre-processed logging data and determine the stick-slip level of the logging data;
[0120] The data set division module 502 is used to divide the well logging data into a training set and a test set according to a preset ratio by using an SPXY algorithm; the training set includes the well logging training data and its corresponding stick-slip level labels; the test set includes the well logging test data and its corresponding stick-slip level labels;
[0121] The target state space model determination module 503 is used to train the constructed state space model through the well logging training data and its corresponding stick-slip level label, and obtain the target state space model after testing the well logging test data and its corresponding stick-slip level label; the state space model is composed of a state transition model and an observation model, and both the state transition model and the observation model have several combination types;
[0122] The optimal prediction module determination module 504 is used to obtain the target optimal prediction model by using the stick-slip degree prediction result output by the target state space physical model as the input data of the constructed optimal prediction network through the XGBoost algorithm;
[0123] The prediction module 505 is used to acquire the mud logging data to be analyzed and construct a stick-slip vibration prediction space corresponding to the mud logging data to be analyzed based on the target optimal prediction model.
[0124] In an optional embodiment, the acquisition module 501 includes:
[0125] A division submodule, used for dividing the logging data into stick-slip cycles of preset multiples;
[0126] A ratio determination submodule, used to determine the ratio between the difference between the maximum drilling speed and the minimum drilling speed of the divided logging data in the preset multiple stick-slip cycle and the average drilling speed;
[0127] The level determination submodule is used to determine the stick-slip level of the divided logging data by using the ratio as a measure of the stick-slip level SSI.
[0128] In an optional embodiment, the acquisition module 501 further includes:
[0129] The decomposition submodule is used to perform wavelet decomposition on the acquired logging data through a preset wavelet basis function and decomposition scale to obtain wavelet decomposition coefficients at different scales;
[0130] A threshold processing submodule, used for performing threshold processing on the wavelet decomposition coefficients to obtain processed wavelet parameters;
[0131] The inverse transformation submodule is used to perform inverse wavelet transformation on the processed wavelet parameters to reconstruct the denoised well logging data.
[0132] In an optional embodiment, the optimal prediction module determination module 504 includes:
[0133] A prediction result determination submodule, used for determining the stick-slip degree level prediction results under different combination types obtained by inputting the logging data into the target state space physical model;
[0134] The optimal model determination submodule is used to use the XGBoost algorithm to train the stick-slip degree prediction results under different combination types to obtain the target optimal prediction model.
[0135] In an optional embodiment, the prediction module 505 includes:
[0136] A grid forming submodule, used to determine the combined value of rotation speed and drilling pressure in the logging data to be analyzed, and form a corresponding grid;
[0137] A prediction submodule, used for inputting the optimal prediction model with the combined value to obtain a corresponding prediction result;
[0138] The construction submodule is used to construct the stick-slip vibration prediction space corresponding to the mud logging data to be analyzed by marking colors of all the grids according to corresponding prediction results.
[0139] An embodiment of the present invention further provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of a method for predicting stick-slip vibration of a drill string as described in any of the above embodiments.
[0140] An embodiment of the present invention further provides a computer storage medium having a computer program stored thereon, wherein when the computer program is executed by the processor, the steps of a method for predicting stick-slip vibration of a drill string as described in any of the above embodiments are implemented.
[0141] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.
[0142] In the several embodiments provided in the present application, it should be understood that the methods, devices, electronic devices and storage media disclosed in the present invention can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.
[0143] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0144] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0145] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a readable storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned readable storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.
[0146] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for predicting stick-slip vibration of a drill string, characterized in that: include: Acquire the preprocessed logging data and determine the stick-slip level of the logging data; The SPXY algorithm is used to divide the logging data into a training set and a test set according to a preset ratio; the training set includes the logging training data and its corresponding stick-slip level labels; the test set includes the logging test data and its corresponding stick-slip level labels; The constructed state space model is trained by the well logging training data and the corresponding stick-slip level labels, and a target state space model is obtained after being tested by the well logging test data and the corresponding stick-slip level labels; the state space model is composed of a state transition model and an observation model, and both the state transition model and the observation model have several combination types; By using the XGBoost algorithm, the stick-slip degree prediction result output by the target state space physical model is used as the input data of the constructed optimal prediction network to obtain the target optimal prediction model; The acquired mud logging data to be analyzed is used, and based on the target optimal prediction model, a stick-slip vibration prediction space corresponding to the mud logging data to be analyzed is constructed.
2. The method for predicting stick-slip vibration of a drill string according to claim 1, characterized in that: Obtaining the preprocessed logging data and determining the stick-slip level of the logging data includes: Dividing the logging data into stick-slip cycles of preset multiples; Determine the ratio between the difference between the maximum drilling speed and the minimum drilling speed of the divided logging data in the preset multiple stick-slip cycles and the average drilling speed; The ratio is used as a measure of the stick-slip level SSI to determine the stick-slip level of the divided logging data.
3. The method for predicting stick-slip vibration of a drill string according to claim 2, characterized in that: Acquiring the preprocessed logging data and determining the stick-slip level of the logging data also includes: The acquired logging data are subjected to wavelet decomposition by using the preset wavelet basis function and decomposition scale to obtain wavelet decomposition coefficients at different scales; Performing threshold processing on the wavelet decomposition coefficients to obtain processed wavelet parameters; The processed wavelet parameters are subjected to inverse wavelet transform to reconstruct the de-noised well logging data.
4. The method for predicting stick-slip vibration of a drill string according to claim 1, characterized in that: By using the XGBoost algorithm, the stick-slip degree prediction result output by the target state space physical model is used as the input data of the constructed optimal prediction network to obtain the target optimal prediction model, including: Determine the stick-slip degree prediction results under different combination types obtained by inputting the logging data into the target state space physical model; The stick-slip degree prediction results under different combination types are trained using the XGBoost algorithm to obtain the target optimal prediction model.
5. The method for predicting stick-slip vibration of a drill string according to claim 1, characterized in that: The acquired logging data to be analyzed is used to construct a stick-slip vibration prediction space corresponding to the logging data to be analyzed based on the target optimal prediction model, including: Determine the combination value of rotation speed and drilling pressure in the logging data to be analyzed, and form a corresponding grid; Inputting the optimal prediction model with the combined value to obtain a corresponding prediction result; By marking colors of all the grids according to corresponding prediction results, a stick-slip vibration prediction space corresponding to the mud logging data to be analyzed is constructed.
6. A prediction device for stick-slip vibration of a drill string, characterized in that: include: An acquisition module, used for acquiring the pre-processed well logging data and determining the stick-slip level of the well logging data; A data set division module is used to divide the well logging data into a training set and a test set according to a preset ratio by using an SPXY algorithm; the training set includes the well logging training data and its corresponding stick-slip level labels; the test set includes the well logging test data and its corresponding stick-slip level labels; a target state space model determination module, used to train the constructed state space model through the well logging training data and its corresponding stick-slip level labels, and obtain the target state space model after testing the well logging test data and its corresponding stick-slip level labels; the state space model is composed of a state transition model and an observation model, and both the state transition model and the observation model have several combination types; An optimal prediction module determination module is used to obtain a target optimal prediction model by using the stick-slip degree prediction result output by the target state space physical model as input data of the constructed optimal prediction network through the XGBoost algorithm; The prediction module is used to acquire the logging data to be analyzed and construct a stick-slip vibration prediction space corresponding to the logging data to be analyzed based on the target optimal prediction model.
7. The drill string stick-slip vibration prediction device according to claim 6, characterized in that: The acquisition module comprises: A division submodule, used for dividing the logging data into stick-slip cycles of preset multiples; A ratio determination submodule, used to determine the ratio between the difference between the maximum drilling speed and the minimum drilling speed of the divided logging data in the preset multiple stick-slip cycle and the average drilling speed; The level determination submodule is used to determine the stick-slip level of the divided logging data by using the ratio as a measure of the stick-slip level SSI.
8. The drill string stick-slip vibration prediction device according to claim 7, characterized in that: The acquisition module also includes: The decomposition submodule is used to perform wavelet decomposition on the acquired logging data through a preset wavelet basis function and decomposition scale to obtain wavelet decomposition coefficients at different scales; A threshold processing submodule, used for performing threshold processing on the wavelet decomposition coefficients to obtain processed wavelet parameters; The inverse transformation submodule is used to perform inverse wavelet transformation on the processed wavelet parameters to reconstruct the denoised well logging data.
9. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the method according to any one of claims 1 to 5 is executed.
10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 5 is performed.