Shaft leakage pressure prediction method and system based on transient fluctuation theory and neural network algorithm

Abstract

The invention provides a shaft leakage pressure prediction method and system based on a transient fluctuation theory and a neural network algorithm. The method comprises the steps of performing structured preprocessing on well history data, logging data, logging data and seismic data to obtain model input historical data; solving the transient wave equation by adopting a characteristic line method, and calculating shaft leakage pressure to obtain a shaft leakage pressure theoretical value; taking the shaft leakage pressure theoretical value as a label, taking the model input historical data as model input, and training a plurality of neural networks to obtain an optimal neural network; training the optimal neural network again by using the shaft leakage pressure measured value as a label and the model input measured data as the model input to obtain the trained optimal neural network; and predicting the stratum leakage pressure at the new well position by using the optimal neural network and generating a pressure profile. According to the scheme provided by the invention, the reliability of a prediction result can be ensured. The probability of complex accidents is effectively reduced, and it is guaranteed that the ultra-deep well drilling process is safe and efficient.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of wellbore leakage pressure prediction, and particularly relates to a wellbore leakage pressure prediction method and system based on transient wave theory and neural network algorithm. BACKGROUND

[0002] In recent years, oil and gas resource exploration is advancing to deep and ultra-deep layers, and the deepest well in China has a vertical depth of more than 10,000 meters. The problem brought by ultra-deep layers is that the measurement and prediction accuracy of seismic and logging data is not enough, and it is difficult to accurately predict the formation pressure and fracture pressure profile of the ultra-deep layer, so it is more difficult to determine the leakage pressure of the ultra-deep layer. However, the leakage pressure is an important parameter related to the safety of the drilling process. If the bottom hole circulating equivalent density is greater than the leakage pressure, wellbore leakage will occur, and if it is not effectively controlled, it may cause well wall collapse, sticking and blowout, and other complex accidents, which seriously endanger the safety of drilling.

[0003] Disadvantages of the prior art:

[0004] The existing leakage pressure prediction technology is mostly based on the measured data of drilled wells for parallel prediction, but the change of formation trend, depth and lithology will all affect the calculation of the leakage pressure of the formation, and the existing model cannot consider comprehensively.

[0005] The prediction of the leakage pressure in the prior art cannot comprehensively consider the change of logging, recording and other numerical values, and cannot consider the influence of engineering parameters such as leakage amount and leakage rate, which will all affect the calculation of the leakage pressure. SUMMARY

[0006] To solve the above technical problems, the application provides a technical scheme of a wellbore leakage pressure prediction method based on transient wave theory and neural network algorithm to solve the above technical problems.

[0007] The first aspect of the application discloses a wellbore leakage pressure prediction method based on transient wave theory and neural network algorithm, which comprises the following steps:

[0008] Step S1, structure the well history data, logging data, recording data and seismic data to obtain structured data; and pre-process the structured data to obtain model input historical data;

[0009] Step S2, solve the transient wave equation by using the method of characteristic lines to calculate the wellbore leakage pressure and obtain the theoretical value of the wellbore leakage pressure;

[0010] Step S3, take the theoretical value of the wellbore leakage pressure as a label, and the model input historical data as a model input to train multiple neural networks; and obtain the optimal neural network by evaluating the average deviation of the prediction of the multiple neural networks after training;

[0011] Step S4: Using the measured wellbore leakage pressure as the label, and the measured data as the model input, the optimal neural network is trained again to obtain the trained optimal neural network.

[0012] Step S5: Apply the optimal neural network to predict the formation leakage pressure at the new well location and generate a pressure profile.

[0013] According to the method of the first aspect of the present invention, in step S1, the structured data includes:

[0014] Formation Poisson's ratio, formation elastic modulus, lithological density, effective porosity, sonic transit time, permeability, resistivity, water saturation, pore pressure, spontaneous potential, and spontaneous gamma.

[0015] According to the method of the first aspect of the present invention, in step S1, the structured data is generated by manual editing or machine recognition to form editable original structured data; outliers and missing values ​​in the original structured data are preprocessed, the preprocessing including data cleaning, data integration, data standardization and data reduction, and finally structured data is obtained.

[0016] According to the method of the first aspect of the present invention, in step S3, the plurality of neural networks include:

[0017] Convolutional neural networks, generative adversarial neural networks, attention mechanism neural networks, fully connected neural networks, recurrent neural networks, and transformer neural networks.

[0018] According to the method of the first aspect of the present invention, in step S3, training multiple neural networks using the theoretical value of the wellbore leakage pressure as a label and the historical data of the model input as the model input includes:

[0019] The corresponding loss functions are constructed by applying the mean square error of the theoretical value of wellbore leakage pressure and the predicted value of each neural network.

[0020] The loss function is then applied to train the corresponding neural network.

[0021] According to the method of the first aspect of the present invention, in step S4, the optimal neural network is applied to predict the wellbore leakage pressure of the drilled well, and the prediction is compared with the measured value of the wellbore leakage pressure to obtain the mean square error value.

[0022] According to the method of the first aspect of the present invention, in step S4, the mean square error value is applied to further verify and train the optimal neural network, gradually approximating the measured value, and obtaining the trained optimal neural network.

[0023] A second aspect of this invention discloses a wellbore leakage pressure prediction system based on transient fluctuation theory and neural network algorithms, the system comprising:

[0024] The first processing module is configured to: structure the well history data, well logging data, well logging data, and seismic data to obtain structured data; and preprocess the structured data to obtain historical data for model input.

[0025] The second processing module is configured to use the method of characteristics to solve the transient wave equation, calculate the wellbore leakage pressure, and obtain the theoretical value of the wellbore leakage pressure.

[0026] The third processing module is configured to use the theoretical value of the wellbore leakage pressure as a label and the historical data of the model input as the model input to train multiple neural networks; and to obtain the optimal neural network by evaluating the average deviation of the predictions of the multiple neural networks after training.

[0027] The fourth processing module is configured to use the measured wellbore leakage pressure as a label, the measured data as the model input, and retrain the optimal neural network to obtain the trained optimal neural network.

[0028] The fifth processing module is configured to apply the optimal neural network to predict the formation leakage pressure at the new well location and generate a pressure profile.

[0029] According to a system based on a second aspect of the present invention, the structured data includes:

[0030] Formation Poisson's ratio, formation elastic modulus, lithological density, effective porosity, sonic transit time, permeability, resistivity, water saturation, pore pressure, spontaneous potential, and spontaneous gamma.

[0031] According to the system of the second aspect of the present invention, the structured data is generated by manual editing or machine recognition to form editable original structured data; outliers and missing values ​​in the original structured data are preprocessed, the preprocessing including data cleaning, data integration, data standardization and data reduction, to finally obtain structured data.

[0032] According to a system based on a second aspect of the present invention, the plurality of neural networks include:

[0033] Convolutional neural networks, generative adversarial neural networks, attention mechanism neural networks, fully connected neural networks, recurrent neural networks, and transformer neural networks.

[0034] According to a system of a second aspect of the present invention, training multiple neural networks using the theoretical value of the wellbore leakage pressure as a label and historical data as model input includes:

[0035] The corresponding loss functions are constructed by applying the mean square error of the theoretical value of wellbore leakage pressure and the predicted value of each neural network.

[0036] The loss function is then applied to train the corresponding neural network.

[0037] According to the system of the second aspect of the present invention, the optimal neural network is used to predict the wellbore leakage pressure of the drilled well, and the prediction is compared with the measured value of the wellbore leakage pressure to obtain the mean square error value.

[0038] The mean squared error value is then used to further verify and train the optimal neural network, gradually approximating the measured value to obtain the trained optimal neural network.

[0039] A third aspect of this invention discloses an electronic device. The electronic device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the steps of the wellbore leakage pressure prediction method based on transient fluctuation theory and neural network algorithm according to any one of the first aspects of this disclosure.

[0040] A fourth aspect of this invention discloses a computer-readable storage medium. The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a wellbore leakage pressure prediction method based on transient fluctuation theory and neural network algorithm according to any one of the first aspects of this disclosure.

[0041] In summary, the proposed solution effectively combines mathematical models, field measured data, and predicted data variation patterns. It integrates multiple data structures from rock mechanics, seismic analysis, and well logging, ensuring reliable prediction of leakage pressure under the constraints of measured data. Furthermore, the predicted profile can provide a reference for on-site design and construction, effectively reducing the probability of complex accidents and ensuring the safe and efficient drilling process in ultra-deep wells. Attached Figure Description

[0042] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0043] Figure 1 A flowchart illustrating a wellbore leakage pressure prediction method based on transient fluctuation theory and neural network algorithm according to an embodiment of the present invention;

[0044] Figure 2This is a structural diagram of a wellbore leakage pressure prediction system based on transient fluctuation theory and neural network algorithm according to an embodiment of the present invention;

[0045] Figure 3 This is a structural diagram of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0046] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] The first aspect of this invention discloses a wellbore leakage pressure prediction method based on transient fluctuation theory and neural network algorithm.

[0048] Example 1:

[0049] Figure 1 The flowchart below shows a wellbore leakage pressure prediction method based on transient fluctuation theory and neural network algorithm according to Embodiment 1 of the present invention. Figure 1 As shown, the method includes:

[0050] Step S1: Structure the well history data, well logging data, well logging data, and seismic data to obtain structured data; preprocess the structured data to obtain the model input historical data;

[0051] Step S2: Solve the transient wave equation using the method of characteristics to calculate the wellbore leakage pressure and obtain the theoretical value of the wellbore leakage pressure;

[0052] Step S3: Using the theoretical value of wellbore leakage pressure as a label and the historical data of the model input as the model input, train multiple neural networks; by evaluating the average deviation of the predictions of the multiple neural networks after training, obtain the optimal neural network;

[0053] Step S4: Using the measured wellbore leakage pressure as the label, and the measured data as the model input, the optimal neural network is trained again to obtain the trained optimal neural network.

[0054] Step S5: Apply the optimal neural network to predict the formation leakage pressure at the new well location and generate a pressure profile.

[0055] Example 2:

[0056] Example 2 discloses a wellbore leakage pressure prediction method based on transient fluctuation theory and neural network algorithm, the method comprising:

[0057] Step S1: Structure the well history data, well logging data, well logging data, and seismic data to obtain structured data; preprocess the structured data to obtain the model input historical data;

[0058] In step S1, the well history data, well logging data, well logging data and seismic data are structured to obtain structured data; the structured data is preprocessed to obtain the model input historical data.

[0059] In some embodiments, in step S1, the structured data includes:

[0060] Formation Poisson's ratio, formation elastic modulus, lithological density, effective porosity, sonic transit time, permeability, resistivity, water saturation, pore pressure, spontaneous potential, and spontaneous gamma.

[0061] Specifically, some of the field data needs to be identified manually or by machine to form editable structured data; outliers and missing values ​​in the original data documents are preprocessed, including data cleaning, data integration, data standardization and data reduction, to finally obtain preprocessed data.

[0062] Step S2: Solve the transient wave equation using the method of characteristics to calculate the wellbore leakage pressure and obtain the theoretical value of the wellbore leakage pressure;

[0063] Step S3: Using the theoretical value of wellbore leakage pressure as a label and the historical data of the model input as the model input, train multiple neural networks; by evaluating the average deviation of the predictions of the multiple neural networks after training, obtain the optimal neural network;

[0064] Step S4: Using the measured wellbore leakage pressure as the label, and the measured data as the model input, the optimal neural network is trained again to obtain the trained optimal neural network.

[0065] Step S5: Apply the optimal neural network to predict the formation leakage pressure at the new well location and generate a pressure profile.

[0066] Example 3:

[0067] Example 3 discloses a wellbore leakage pressure prediction method based on transient fluctuation theory and neural network algorithm, the method comprising:

[0068] Step S1: Structure the well history data, well logging data, well logging data, and seismic data to obtain structured data; preprocess the structured data to obtain the model input historical data;

[0069] In step S1, the well history data, well logging data, well logging data and seismic data are structured to obtain structured data; the structured data is preprocessed to obtain the model input historical data.

[0070] In some embodiments, in step S1, the structured data includes:

[0071] Formation Poisson's ratio, formation elastic modulus, lithological density, effective porosity, sonic transit time, permeability, resistivity, water saturation, pore pressure, spontaneous potential, and spontaneous gamma.

[0072] Specifically, some of the field data needs to be identified manually or by machine to form editable structured data; outliers and missing values ​​in the original data documents are preprocessed, including data cleaning, data integration, data standardization and data reduction, to finally obtain preprocessed data.

[0073] Step S2: Solve the transient wave equation using the method of characteristics to calculate the wellbore leakage pressure and obtain the theoretical value of the wellbore leakage pressure;

[0074] Step S3: Using the theoretical value of wellbore leakage pressure as a label and the historical data of the model input as the model input, train multiple neural networks; by evaluating the average deviation of the predictions of the multiple neural networks after training, obtain the optimal neural network;

[0075] In some embodiments, in step S3, the plurality of neural networks include:

[0076] Convolutional neural networks, generative adversarial neural networks, attention mechanism neural networks, fully connected neural networks, recurrent neural networks, and transformer neural networks.

[0077] The process of training multiple neural networks, using the theoretical value of wellbore leakage pressure as the label and historical data as the model input, includes:

[0078] The corresponding loss functions are constructed by applying the mean square error of the theoretical value of wellbore leakage pressure and the predicted value of each neural network.

[0079] The loss function is then applied to train the corresponding neural network.

[0080] Step S4: Using the measured wellbore leakage pressure as the label, and the measured data as the model input, the optimal neural network is trained again to obtain the trained optimal neural network.

[0081] Step S5: Apply the optimal neural network to predict the formation leakage pressure at the new well location and generate a pressure profile.

[0082] Example 4:

[0083] Example 4 discloses a wellbore leakage pressure prediction method based on transient fluctuation theory and neural network algorithm, the method comprising:

[0084] Step S1: Structure the well history data, well logging data, well logging data, and seismic data to obtain structured data; preprocess the structured data to obtain the model input historical data;

[0085] In step S1, the well history data, well logging data, well logging data and seismic data are structured to obtain structured data; the structured data is preprocessed to obtain the model input historical data.

[0086] In some embodiments, in step S1, the structured data includes:

[0087] Formation Poisson's ratio, formation elastic modulus, lithological density, effective porosity, sonic transit time, permeability, resistivity, water saturation, pore pressure, spontaneous potential, and spontaneous gamma.

[0088] Specifically, some of the field data needs to be identified manually or by machine to form editable structured data; outliers and missing values ​​in the original data documents are preprocessed, including data cleaning, data integration, data standardization and data reduction, to finally obtain preprocessed data.

[0089] Step S2: Solve the transient wave equation using the method of characteristics to calculate the wellbore leakage pressure and obtain the theoretical value of the wellbore leakage pressure;

[0090] Step S3: Using the theoretical value of wellbore leakage pressure as a label and the historical data of the model input as the model input, train multiple neural networks; by evaluating the average deviation of the predictions of the multiple neural networks after training, obtain the optimal neural network;

[0091] In some embodiments, in step S3, the plurality of neural networks include:

[0092] Convolutional neural networks, generative adversarial neural networks, attention mechanism neural networks, fully connected neural networks, recurrent neural networks, and transformer neural networks.

[0093] The process of training multiple neural networks, using the theoretical value of wellbore leakage pressure as the label and historical data as the model input, includes:

[0094] The corresponding loss functions are constructed by applying the mean square error of the theoretical value of wellbore leakage pressure and the predicted value of each neural network.

[0095] The loss function is then applied to train the corresponding neural network.

[0096] In step S4, the measured wellbore leakage pressure is used as the label, and the measured data is used as the model input. The optimal neural network is then trained again to obtain the trained optimal neural network.

[0097] Specifically, the optimal neural network is applied to predict the wellbore leakage pressure of drilled wells, and the prediction is compared with the measured wellbore leakage pressure to obtain the mean square error value.

[0098] The mean squared error is then used to further validate and train the optimal neural network, gradually approximating the measured value to obtain the trained optimal neural network.

[0099] Step S5: Apply the optimal neural network to predict the formation leakage pressure at the new well location and generate a pressure profile.

[0100] In summary, the proposed solution effectively combines mathematical models, field measured data, and predicted data variation patterns. It integrates multiple data structures from rock mechanics, seismic analysis, and well logging, ensuring reliable prediction of leakage pressure under the constraints of measured data. Furthermore, the predicted profile can provide a reference for on-site design and construction, effectively reducing the probability of complex accidents and ensuring the safe and efficient drilling process in ultra-deep wells.

[0101] The second aspect of this invention discloses a wellbore leakage pressure prediction system based on transient fluctuation theory and neural network algorithm.

[0102] Example 5:

[0103] Figure 2 This is a structural diagram of a wellbore leakage pressure prediction system based on transient fluctuation theory and neural network algorithm according to Embodiment 5 of the present invention; as shown. Figure 2 As shown, the system includes:

[0104] The first processing module 101 is configured to: structure the well history data, well logging data, well logging data and seismic data to obtain structured data; and preprocess the structured data to obtain historical data for model input.

[0105] The second processing module 102 is configured to use the method of characteristics to solve the transient wave equation, calculate the wellbore leakage pressure, and obtain the theoretical value of the wellbore leakage pressure.

[0106] The third processing module 103 is configured to train multiple neural networks using the theoretical value of the wellbore leakage pressure as a label and the historical data of the model input as the model input; and to obtain the optimal neural network by evaluating the average deviation of the predictions of the multiple neural networks after training.

[0107] The fourth processing module 104 is configured to use the measured wellbore leakage pressure as a label, the measured data as the model input, and retrain the optimal neural network to obtain the trained optimal neural network.

[0108] The fifth processing module 105 is configured to use the optimal neural network to predict the formation leakage pressure at the new well location and generate a pressure profile.

[0109] Example 6:

[0110] Example 6 discloses a structural diagram of a wellbore leakage pressure prediction system based on transient fluctuation theory and neural network algorithm; the system includes:

[0111] The first processing module 101 is configured to: structure the well history data, well logging data, well logging data and seismic data to obtain structured data; and preprocess the structured data to obtain historical data for model input.

[0112] According to a system based on a second aspect of the present invention, the first processing module 101 is specifically configured such that the structured data includes:

[0113] Formation Poisson's ratio, formation elastic modulus, lithological density, effective porosity, sonic transit time, permeability, resistivity, water saturation, pore pressure, spontaneous potential, and spontaneous gamma.

[0114] Specifically, some of the field data needs to be identified manually or by machine to form editable structured data; outliers and missing values ​​in the original data documents are preprocessed, including data cleaning, data integration, data standardization and data reduction, to finally obtain preprocessed data.

[0115] The second processing module 102 is configured to use the method of characteristics to solve the transient wave equation, calculate the wellbore leakage pressure, and obtain the theoretical value of the wellbore leakage pressure.

[0116] The third processing module 103 is configured to train multiple neural networks using the theoretical value of the wellbore leakage pressure as a label and the historical data of the model input as the model input; and to obtain the optimal neural network by evaluating the average deviation of the predictions of the multiple neural networks after training.

[0117] The fourth processing module 104 is configured to use the measured wellbore leakage pressure as a label, the measured data as the model input, and retrain the optimal neural network to obtain the trained optimal neural network.

[0118] The fifth processing module 105 is configured to use the optimal neural network to predict the formation leakage pressure at the new well location and generate a pressure profile.

[0119] Example 7:

[0120] Example 7 discloses a structural diagram of a wellbore leakage pressure prediction system based on transient fluctuation theory and neural network algorithm; the system includes:

[0121] The first processing module 101 is configured to: structure the well history data, well logging data, well logging data and seismic data to obtain structured data; and preprocess the structured data to obtain historical data for model input.

[0122] According to a system based on a second aspect of the present invention, the first processing module 101 is specifically configured such that the structured data includes:

[0123] Formation Poisson's ratio, formation elastic modulus, lithological density, effective porosity, sonic transit time, permeability, resistivity, water saturation, pore pressure, spontaneous potential, and spontaneous gamma.

[0124] Specifically, some of the field data needs to be identified manually or by machine to form editable structured data; outliers and missing values ​​in the original data documents are preprocessed, including data cleaning, data integration, data standardization and data reduction, to finally obtain preprocessed data.

[0125] The second processing module 102 is configured to use the method of characteristics to solve the transient wave equation, calculate the wellbore leakage pressure, and obtain the theoretical value of the wellbore leakage pressure.

[0126] The third processing module 103 is configured to train multiple neural networks using the theoretical value of the wellbore leakage pressure as a label and the historical data of the model input as the model input; and to obtain the optimal neural network by evaluating the average deviation of the predictions of the multiple neural networks after training.

[0127] According to a system based on a second aspect of the present invention, the third processing module 103 is specifically configured such that the plurality of neural networks include:

[0128] Convolutional neural networks, generative adversarial neural networks, attention mechanism neural networks, fully connected neural networks, recurrent neural networks, and transformer neural networks.

[0129] The process of training multiple neural networks, using the theoretical value of wellbore leakage pressure as the label and historical data as the model input, includes:

[0130] The corresponding loss functions are constructed by applying the mean square error of the theoretical value of wellbore leakage pressure and the predicted value of each neural network.

[0131] The loss function is then applied to train the corresponding neural network.

[0132] The fourth processing module 104 is configured to use the measured wellbore leakage pressure as a label, the measured data as the model input, and retrain the optimal neural network to obtain the trained optimal neural network.

[0133] The fifth processing module 105 is configured to use the optimal neural network to predict the formation leakage pressure at the new well location and generate a pressure profile.

[0134] Example 8:

[0135] Example 8 discloses a structural diagram of a wellbore leakage pressure prediction system based on transient fluctuation theory and neural network algorithm; the system includes:

[0136] The first processing module 101 is configured to: structure the well history data, well logging data, well logging data and seismic data to obtain structured data; and preprocess the structured data to obtain historical data for model input.

[0137] According to a system based on a second aspect of the present invention, the first processing module 101 is specifically configured such that the structured data includes:

[0138] Formation Poisson's ratio, formation elastic modulus, lithological density, effective porosity, sonic transit time, permeability, resistivity, water saturation, pore pressure, spontaneous potential, and spontaneous gamma.

[0139] Specifically, some of the field data needs to be identified manually or by machine to form editable structured data; outliers and missing values ​​in the original data documents are preprocessed, including data cleaning, data integration, data standardization and data reduction, to finally obtain preprocessed data.

[0140] The second processing module 102 is configured to use the method of characteristics to solve the transient wave equation, calculate the wellbore leakage pressure, and obtain the theoretical value of the wellbore leakage pressure.

[0141] The third processing module 103 is configured to train multiple neural networks using the theoretical value of the wellbore leakage pressure as a label and the historical data of the model input as the model input; and to obtain the optimal neural network by evaluating the average deviation of the predictions of the multiple neural networks after training.

[0142] According to a system based on a second aspect of the present invention, the third processing module 103 is specifically configured such that the plurality of neural networks include:

[0143] Convolutional neural networks, generative adversarial neural networks, attention mechanism neural networks, fully connected neural networks, recurrent neural networks, and transformer neural networks.

[0144] The process of training multiple neural networks, using the theoretical value of wellbore leakage pressure as the label and historical data as the model input, includes:

[0145] The corresponding loss functions are constructed by applying the mean square error of the theoretical value of wellbore leakage pressure and the predicted value of each neural network.

[0146] The loss function is then applied to train the corresponding neural network.

[0147] The fourth processing module 104 is configured to use the measured wellbore leakage pressure as a label, the measured data as the model input, and retrain the optimal neural network to obtain the trained optimal neural network.

[0148] According to the system of the second aspect of the present invention, the fourth processing module 104 is specifically configured to apply an optimal neural network to predict the wellbore leakage pressure of the drilled well, and compare and analyze it with the measured value of the wellbore leakage pressure to obtain the mean square error value.

[0149] The mean squared error is then used to further validate and train the optimal neural network, gradually approximating the measured value to obtain the trained optimal neural network.

[0150] The fifth processing module 105 is configured to use the optimal neural network to predict the formation leakage pressure at the new well location and generate a pressure profile.

[0151] A third aspect of this invention discloses an electronic device. The electronic device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the steps of the wellbore leakage pressure prediction method based on transient fluctuation theory and neural network algorithm according to any one of the first aspects of this invention.

[0152] Figure 3 This is a structural diagram of an electronic device according to an embodiment of the present invention, such as... Figure 3As shown, the electronic device includes a processor, memory, communication interface, display screen, and input device connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, Near Field Communication (NFC), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input device can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the device's casing, or an external keyboard, touchpad, or mouse.

[0153] Those skilled in the art will understand that Figure 3 The structure shown is merely a structural diagram of the part related to the technical solution of this disclosure and does not constitute a limitation on the electronic device to which the solution of this application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0154] A fourth aspect of this invention discloses a computer-readable storage medium. The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a wellbore leakage pressure prediction method based on transient fluctuation theory and neural network algorithm, as described in any of the first aspects of this invention.

[0155] Please note that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments have been described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification. The above embodiments only illustrate several implementation methods of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the invention patent. It should be pointed out that for those skilled in the art, several modifications and improvements can be made without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for predicting lost circulation pressure in a wellbore based on transient fluctuation theory and neural network algorithm, characterized in that, The method comprises: Step S1, structuring well history data, logging data, drilling data and seismic data to obtain structured data; preprocessing the structured data to obtain model input historical data; Step S2, solving the transient wave equation by using the method of characteristics to calculate the wellbore leakage pressure to obtain the theoretical value of the wellbore leakage pressure; Step S3, taking the theoretical value of the wellbore leakage pressure as a label and the model input historical data as model input to train multiple neural networks; obtaining the optimal neural network by evaluating the average deviation of the prediction of the trained multiple neural networks; Step S4, applying the measured value of the wellbore leakage pressure as a label and the measured model input data as model input to retrain the optimal neural network to obtain the trained optimal neural network; Step S5, applying the optimal neural network to predict the formation leakage pressure at a new well site and generate a pressure profile.

2. The method of claim 1, wherein, In the step S1, the structured data comprises: Formation Poisson's ratio, formation elastic modulus, lithology density, effective porosity, acoustic time difference, permeability, resistivity, water saturation, pore pressure, spontaneous potential and natural gamma.

3. The method of claim 2, wherein, In the step S1, the structured data is edited by artificial editing or machine recognition to form editable original structured data; the abnormal values and missing values in the original structured data are preprocessed, and the preprocessing comprises data cleaning, data integration, data standardization and data reduction, and finally the structured data is obtained.

4. The method of claim 1, wherein, In the step S3, the multiple neural networks comprise: Convolutional neural network, generative adversarial neural network, attention mechanism neural network, fully connected neural network, recurrent neural network and transformer neural network.

5. The method of claim 4, wherein, In the step S3, taking the theoretical value of the wellbore leakage pressure as a label and the model input historical data as model input to train multiple neural networks comprises: Applying the mean square error of the theoretical value of the wellbore leakage pressure and the prediction value of each neural network to construct a corresponding loss function; Training the corresponding neural network by using the loss function.

6. The method of claim 5, wherein, In the step S4, the optimal neural network is applied to predict the wellbore leakage pressure of the drilled well, and compared with the measured value of the wellbore leakage pressure to obtain a mean square error value.

7. The method of claim 6, wherein, In the step S4, the mean square error value is further used to verify and train the optimal neural network to gradually approach the measured value to obtain the trained optimal neural network.

8. A wellbore lost circulation pressure prediction system based on transient fluctuation theory and neural network algorithm, characterized in that, The system comprises: A first processing module configured to structure well history data, logging data, drilling data and seismic data to obtain structured data; preprocessing the structured data to obtain model input historical data; A second processing module configured to solve the transient wave equation by using the method of characteristics to calculate the wellbore leakage pressure to obtain the theoretical value of the wellbore leakage pressure; A third processing module configured to take the theoretical value of the wellbore leakage pressure as a label and the model input historical data as model input to train multiple neural networks; obtaining the optimal neural network by evaluating the average deviation of the prediction of the trained multiple neural networks; The fourth processing module is configured to apply the wellbore leakage pressure measured value as a label, apply the model input measured data as a model input, retrain the optimal neural network, and obtain a trained optimal neural network. The fifth processing module is configured to apply the optimal neural network to predict the formation leakage pressure at a new well location and generate a pressure profile.

9. The system for predicting the pressure of a lost circulation zone in a wellbore based on transient fluctuation theory and neural network algorithm of claim 8, wherein, The structured data includes: Formation Poisson's ratio, formation elastic modulus, lithology density, effective porosity, acoustic travel time, permeability, resistivity, water saturation, pore pressure, spontaneous potential, and natural gamma.

10. The wellbore lost circulation pressure prediction system based on transient fluctuation theory and neural network algorithm of claim 9, wherein, The structured data is edited by an artificial or machine-recognized and forms editable original structured data; the abnormal values and missing values in the original structured data are preprocessed, and the preprocessing includes data cleaning, data integration, data standardization, and data reduction, and finally the structured data is obtained.

11. The wellbore lost circulation pressure prediction system based on transient fluctuation theory and neural network algorithm of claim 8, wherein, The plurality of neural networks includes: Convolutional neural network, generative adversarial neural network, attention mechanism neural network, fully connected neural network, recurrent neural network, and transformer neural network.

12. The wellbore lost circulation pressure prediction system based on transient fluctuation theory and neural network algorithm of claim 11, wherein, The training of the plurality of neural networks includes: Applying the mean square error of the wellbore leakage pressure theoretical value and the prediction value of each neural network to construct a corresponding loss function; Applying the loss function to train the corresponding neural network.

13. The wellbore lost circulation pressure prediction system based on transient fluctuation theory and neural network algorithm of claim 12, wherein, Applying the optimal neural network to predict the wellbore leakage pressure of the drilled well and comparing and analyzing with the measured value of the wellbore leakage pressure to obtain a mean square error value; Applying the mean square error value to further verify and train the optimal neural network, gradually approaching the measured value, and obtaining a trained optimal neural network.

14. An electronic device, comprising: The electronic device includes a memory and a processor, the memory stores a computer program, and the processor executes the computer program to realize the steps of the wellbore leakage pressure prediction method based on the transient wave theory and neural network algorithm in any one of claims 1 to 7.

15. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program is executed by the processor to realize the steps of the wellbore leakage pressure prediction method based on the transient wave theory and neural network algorithm in any one of claims 1 to 7.

Images