Horizontal well formation pressure artificial intelligence prediction method and system

Abstract

The present application belongs to the field of oil and gas exploration, and discloses a horizontal well formation pressure artificial intelligence prediction method and system, which comprises: horizontal well logging data sensitive response feature construction; original logging response feature data preprocessing; softening treatment of pressure coefficient; pressure prediction model training based on neural network regression; and horizontal well pressure coefficient prediction. The present application is a cross-border innovation of the existing artificial intelligence method in the field of oil and gas exploration. Through comparative analysis of the measured pressure coefficient and the predicted pressure coefficient, the rationality and obvious advantages of the present application can be proved. As an effective horizontal well pressure coefficient prediction method, the present application significantly improves the performance accuracy of the traditional formation pressure prediction method, makes the formation pressure prediction method widely applicable to different lithology conditions, ensures the performance stability of the formation pressure prediction method in the application of horizontal wells, and accurately solves the 'pain points' of the traditional formation pressure prediction method.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of oil and gas exploration, and particularly relates to a horizontal well formation pressure artificial intelligence prediction method and system. BACKGROUND

[0002] Formation pressure refers to the pressure generated by the action of geological fluids such as formation water, oil, natural gas, etc. in rock pores. Under normal circumstances, when the formation pressure exceeds the static column pressure, the formation will exhibit an abnormally high pressure phenomenon. In the field of oil and gas geological exploration and oil and gas well engineering, abnormally high pressure formation will not only seriously affect the wellbore stability during drilling, but also increase the risk of drilling accidents such as well kick and blowout. Therefore, accurate prediction of formation pressure can effectively prevent wellbore instability, improve drilling efficiency, save drilling cost, and provide an important basis for the rational design of drilling fluid density and well structure. In the traditional formation pressure prediction method, Eaton and Bowers method is mainly used for straight well formation pressure prediction, and the effectiveness of these methods depends on the establishment of normal compaction trend line. In continuous formation, the establishment of normal compaction trend line usually depends on expert experience, which will make the prediction result of formation pressure have strong subjectivity and large error.

[0003] From the perspective of oil exploration and development, the application of horizontal wells is more conducive to achieving high and stable production, further increasing the urgency of the study of horizontal well formation pressure prediction methods. Although various horizontal well formation pressure prediction methods have been proposed by different researchers, the above methods cannot escape the dependence on vertical depth. The direct way to solve the above problems is to explore the relationship between the well pressure coefficient and the sensitive response characteristics of logging data, and use the relevant response characteristics in the logging process to achieve the purpose of predicting the formation pressure of horizontal wells.

[0004] Through the above analysis, the problems and defects of the prior art are:

[0005] (1) The effectiveness of the traditional formation pressure prediction method depends on the establishment of normal compaction trend line, and in continuous formation, the establishment of normal compaction trend line usually depends on expert experience, which makes the prediction result of formation pressure have strong subjectivity and large error;

[0006] (2) The existing horizontal well formation pressure prediction method cannot escape the dependence on vertical depth;

[0007] (3) The existing method is difficult to apply to different lithology conditions, and its performance is unstable in the application of horizontal wells. SUMMARY

[0008] In view of the problems existing in the prior art, the present application provides a horizontal well formation pressure artificial intelligence prediction method and system.

[0009] The application is achieved by a horizontal well formation pressure artificial intelligence prediction method, which comprises the following steps:

[0010] Step one, sensitive response feature construction of horizontal well logging data, combined with measured formation pressure in the research area and adjacent area, natural gas charging abundance caused by oil and gas charging difference, uplift damage dispersion, structural difference, and stress load difference response to formation pressure difference, according to the difference of genetic mechanism and pressure distribution characteristics, the sensitive response parameters of horizontal well formation pressure are researched, and the original logging response features such as vertical depth (TVD) and acoustic time difference (AC) are extracted;

[0011] Step two, original logging response feature data preprocessing, the feature data corresponding to the mudstone part in the original logging response feature is removed;

[0012] Step three, softening processing of pressure coefficient, softening processing of known horizontal well pressure coefficient;

[0013] Step four, pressure prediction model training based on neural network regression, using a neural network model, under the condition of the same sand group, the logging response features and pressure coefficient , the neural network regression model obtains the weight between different network layer logic units , and the neural network regression model for horizontal well pressure prediction is trained;

[0014] Step five, horizontal well pressure coefficient prediction, after the vertical depth (TVD), acoustic time difference (AC), neutron porosity (CNL), density (DEN), natural gamma (GR), resistivity (RT), porosity (POR), and permeability (PERM) and other logging response features of unknown logging are collected, and after data preprocessing as in step 2, the neural network regression model trained as in step 4 is used to effectively predict the pressure coefficient of unknown horizontal well.

[0015] Further, the original logging response features specifically include: vertical depth (TVD), acoustic time difference (AC), neutron porosity (CNL), density (DEN), natural gamma (GR), resistivity (RT), porosity (POR), and permeability (PERM).

[0016] Further, the mudstone part corresponding to the original logging response feature in the sand and mud interbedded formation is removed, and the original logging response feature needs to be normalized according to the following rules: ;

[0017] Among them, is the maximum value data of various logging response features, the minimum value data of various logging response characteristics, represent the "maximum-minimum normalized" logging response characteristics, and the normalization of the logging response characteristics is strictly performed after the mudstone data is removed.

[0018] Further, the softening processing of the known horizontal well pressure coefficient is specifically: ;

[0019] wherein, is the original pressure coefficient of the known horizontal well, represents the horizontal well pressure coefficient after softening processing.

[0020] Further, the neural network model performs data preprocessing on the logging response characteristics including the vertical depth (TVD), acoustic time difference (AC), neutron porosity (CNL), density (DEN), natural gamma (GR), resistivity (RT), porosity (POR), and permeability (PERM), and obtains the logging response characteristics As the input of the neural network logic unit, the prediction label of the neural network regression value adopts the horizontal well pressure coefficient value after softening processing .

[0021] Another object of the present application is to provide a horizontal well formation pressure artificial intelligence prediction system, which comprises:

[0022] An input module is configured to construct a horizontal well logging data sensitive response characteristic;

[0023] A data preprocessing module is configured to remove the feature data corresponding to the mudstone part in the original logging response characteristic;

[0024] A coefficient softening module is configured to perform softening processing on the known horizontal well pressure coefficient as the prediction label of the neural network regression value;

[0025] A training module is configured to train a pressure prediction model based on neural network regression;

[0026] A prediction module is configured to predict the pressure coefficient of an unknown horizontal well based on the neural network regression model obtained by training.

[0027] Another object of the present application is to provide a computer device, which comprises a memory and a processor, wherein the memory stores a computer program, and the computer program is executed by the processor to make the processor execute the steps of the horizontal well formation pressure artificial intelligence prediction method.

[0028] Another object of the present application is to provide a computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the steps of the horizontal well formation pressure artificial intelligence prediction method.

[0029] Another object of the present application is to provide an information data processing terminal for implementing the horizontal well formation pressure artificial intelligence prediction system.

[0030] In combination with the above technical solutions and solved technical problems, the technical solution of the present application has the following advantages and positive effects:

[0031] Firstly, the present application eliminates the feature data corresponding to the mudstone part in the original logging response characteristics, effectively avoiding the interference of the logging response characteristics corresponding to the mudstone part in the horizontal well section on the overall formation pressure prediction result.

[0032] The present application softens the known horizontal well pressure coefficient, enhancing the solvability of the regression prediction model.

[0033] The present application is based on a neural network regression model, which effectively predicts the pressure coefficient of unknown horizontal wells.

[0034] Secondly, the horizontal well formation pressure artificial intelligence prediction method and system proposed by the present application significantly improves the performance accuracy of the traditional formation pressure prediction method, making the formation pressure prediction method widely applicable to different lithology conditions and ensuring the performance stability of the formation pressure prediction method in the application of horizontal wells, accurately solving the "pain points" of the traditional formation pressure prediction method.

[0035] Thirdly, as the creative auxiliary evidence of the present application, it is also reflected in the following important aspects:

[0036] (1) The expected income and commercial value of the technical solution of the present application after transformation are:

[0037] Accurate prediction of formation pressure can effectively prevent wellbore instability, improve drilling efficiency and save drilling cost, and the application of horizontal wells is more conducive to achieving high and stable production. The horizontal well formation pressure artificial intelligence prediction method and system proposed by the present application can overcome the large result error of the traditional formation pressure prediction method, making the prediction method applicable to different lithology conditions and having great application value in the field of oil and gas exploration.

[0038] (2) The technical solution of the present application fills the technical gap in the industry at home and abroad:

[0039] The application provides a general stratum pressure prediction method: a horizontal well stratum pressure artificial intelligence prediction technology, which is based on a neural network regression model and effectively predicts the pressure coefficient of an unknown horizontal well, which is a cross-border innovation of an existing artificial intelligence method in the field of oil and gas exploration.

[0040] (3) Does the technical solution of the application solve the technical problems that people have long desired to solve but have always failed to successfully solve?

[0041] The existing horizontal well stratum pressure prediction method cannot get rid of the dependence on vertical depth, and the application introduces an artificial intelligence method into the field of oil and gas exploration, effectively solving this technical problem.

[0042] (4) Does the technical solution of the application overcome technical bias?

[0043] As an effective horizontal well pressure coefficient prediction method, the application significantly improves the performance accuracy of the traditional stratum pressure prediction method, makes the stratum pressure prediction method widely applicable to different lithology conditions, ensures the performance stability of the stratum pressure prediction method in the application of horizontal wells, and accurately solves the 'pain points' of the traditional stratum pressure prediction method. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 is a flow chart of the horizontal well stratum pressure artificial intelligence prediction method provided by the embodiment of the application.

[0045] Figure 2 is a neural network model framework diagram provided by the embodiment of the application. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solution and advantages of the application clearer and more apparent, the application will be further described in detail below with reference to the embodiments. It should be understood that the specific embodiments described herein are only used to explain the application and do not limit the application.

[0047] As shown in Figure 1 The horizontal well stratum pressure artificial intelligence prediction method provided by the embodiment of the application includes the following steps:

[0048] S101, constructing a horizontal well logging data sensitive response feature;

[0049] S102, preprocessing of original logging response feature data;

[0050] S103, softening processing of the pressure coefficient;

[0051] S104, training of a pressure prediction model based on neural network regression;

[0052] S105, prediction of the horizontal well pressure coefficient.

[0053] The horizontal well formation pressure artificial intelligence prediction method provided by the technical solution comprises the following detailed steps:

[0054] S101: Sensitivity response feature construction of horizontal well logging data

[0055] In this step, appropriate logging tools can be selected for logging, and logging curves that need to be recorded can be determined according to different geological conditions and oil production schemes. Then, by processing and analyzing the logging curves, sensitivity response features such as porosity, permeability, and saturation are obtained as input features of the subsequent prediction model.

[0056] S102: Raw logging response feature data preprocessing

[0057] In this step, the raw logging data needs to be preprocessed, including data cleaning, outlier processing, missing value filling, etc., to ensure the quality and integrity of the data. At the same time, data normalization and standardization operations can be performed to facilitate subsequent modeling.

[0058] S103: Softening processing of pressure coefficient

[0059] In this step, the original pressure data can be softened to eliminate the influence of noise and random errors and improve the accuracy and stability of the prediction model. Softening processing can use methods such as moving average and exponential smoothing.

[0060] S104: Pressure prediction model training based on neural network regression

[0061] In this step, a suitable neural network structure and algorithm can be selected to train the preprocessed data to obtain a horizontal well formation pressure prediction model, which includes an input layer, a hidden layer, and an output layer. During the training process, cross-validation and regularization methods can be used to avoid overfitting and underfitting.

[0062] S105: Horizontal well pressure coefficient prediction

[0063] In this step, the trained neural network model can be used to predict new logging data to obtain the coefficient results of the horizontal well formation pressure. The prediction results can be used to guide oilfield development and production, improve recovery and economic benefits.

[0064] In summary, the technical solution of the present application provides a horizontal well formation pressure prediction method based on artificial intelligence. Through the steps of constructing sensitivity response features, preprocessing data, softening pressure coefficients, training neural network models, and predicting pressure coefficients, accurate and efficient formation pressure prediction can be achieved.

[0065] The horizontal well formation pressure artificial intelligence prediction system provided by the embodiment of the present application comprises:

[0066] The input module is used for constructing a sensitive response feature of the horizontal well logging data;

[0067] The data preprocessing module is used for eliminating feature data corresponding to a mudstone part in the original logging response feature;

[0068] The coefficient softening module is used for softening the known horizontal well pressure coefficient as a prediction label of the neural network regression value;

[0069] The training module is used for training a pressure prediction model based on neural network regression;

[0070] The prediction module is used for predicting the pressure coefficient of an unknown horizontal well based on the neural network regression model obtained through training.

[0071] The horizontal well formation pressure artificial intelligence prediction system provided by the embodiment of the present application can be further specifically refined as:

[0072] 1. The input module: the module can include data acquisition, data cleaning, data processing and the like, aims to obtain logging data of the horizontal well, and constructs a sensitive response feature, which is used for subsequent model training and prediction.

[0073] 2. The data preprocessing module: the module can adopt various data processing methods, such as filtering, noise reduction, abnormal value processing and the like, eliminates feature data corresponding to a mudstone part in the original logging response feature, so as to improve the accuracy and reliability of model training and prediction.

[0074] 3. The coefficient softening module: the module can convert the pressure coefficient of the horizontal well into a prediction label of the neural network regression model based on the pressure coefficient of the known horizontal well by using certain softening processing methods, such as smoothing processing, weighting processing and the like, so as to improve the accuracy and stability of model prediction.

[0075] 4. The training module: the module can adopt a machine learning method based on neural network regression, trains a model for predicting the pressure coefficient of the horizontal well, including network structure design, parameter adjustment, loss function selection and the like, so as to improve the prediction precision and generalization ability of the model.

[0076] 5. The prediction module: the module can predict the pressure coefficient of an unknown horizontal well based on the neural network regression model obtained through training, including input data preprocessing, model prediction output, post-processing and the like, so as to realize accurate prediction and optimized management of the pressure of the horizontal well.

[0077] The horizontal well formation pressure artificial intelligence prediction method provided by the embodiment of the present application specifically comprises:

[0078] (1) Sensitivity response characteristics of horizontal well logging data construction, combined with the measured formation pressure in the research area and adjacent area, the response of natural gas injection abundance caused by the difference of oil and gas injection, uplift damage dispersion, tectonic difference, the difference of formation pressure caused by the difference of ground stress load, according to the difference of genetic mechanism and pressure distribution characteristics, the sensitivity response parameters of horizontal well formation pressure are studied, and the original logging response characteristics such as TVD and AC are extracted.

[0079] (2) Data preprocessing of original logging response characteristics, assuming that the prediction area is sand and mud interbedded formation, and the sandstone of small layer is the main lithology, the original logging response characteristic values extracted in different lithology will show obvious difference, and the logging response characteristics corresponding to the mudstone part in the horizontal well section will directly interfere with the prediction result of the whole formation pressure of the prediction method. Therefore, according to the size relationship between GR value and GR threshold value in logging response, the characteristic data corresponding to mudstone part in original logging response characteristics is removed.

[0080] (3) Softening treatment of pressure coefficient, unlike the pressure coefficient of vertical well, the pressure coefficient of horizontal well usually does not show obvious change, which will reduce the solvability of regression prediction model. In order to enhance the solvability of regression prediction model, when using neural network regression model to fit the nonlinear relationship between logging response characteristics and pressure coefficient Y, the known horizontal well pressure coefficient needs to be softened.

[0081] (4) Pressure prediction model training based on neural network regression, using the neural network model as shown in Figure 2 , with the logging response characteristics and pressure coefficient under the same sand group condition, the neural network regression model gets the weight between different network layer logic units, so as to train the neural network regression model for horizontal well pressure prediction.

[0082] (5) Prediction of horizontal well pressure coefficient, after data preprocessing of logging response characteristics such as TVD, AC, CNL, DEN, GR, RT, POR and PERM of unknown logging as step 2, using the neural network regression model trained in step 4, the unknown horizontal well can be effectively predicted.

[0083] The original logging response features provided by the embodiment of the present application specifically include: true vertical depth (TVD), acoustic time (AC), neutron porosity (CNL), density (DEN), natural gamma (GR), resistivity (RT), porosity (POR) and permeability (PERM).

[0084] The embodiment of the present application provides that in the sand and mud interbedded formation, the feature data corresponding to the mudstone part in the original logging response features is removed, different logging response features have their own dimensions and dimension units, in order to avoid the values with large data levels in the response features from interfering with the predicted pressure coefficient of the horizontal well, the original logging response features need to be normalized according to the following rules:

[0085] Among them, is the maximum value data of various logging response features, is the minimum value data of various logging response features, represents the logging response features after the "maximum-minimum normalization". The normalization processing of the logging response features is strictly performed after the mudstone data is removed.

[0086] The embodiment of the present application provides softening processing for the known horizontal well pressure coefficient, specifically:

[0087] Among them, is the original pressure coefficient of the known horizontal well. represents the horizontal well pressure coefficient after the softening processing, which means that a thousandth percentile error with a normal distribution is applied on the corresponding pressure coefficient.

[0088] The neural network model provided by the embodiment of the present application, after the data preprocessing of the logging response features such as true vertical depth (TVD), acoustic time (AC), neutron porosity (CNL), density (DEN), natural gamma (GR), resistivity (RT), porosity (POR) and permeability (PERM), the obtained logging response features are taken as the input of the neural network logic unit, and the predicted label of the neural network regression value adopts the horizontal well pressure coefficient value after the softening processing .

[0089] ​​​The following takes a certain oil field in the west of China as an example to simulate the application of the formation pressure prediction of the horizontal well to further illustrate the present application. According to step 1, the present application will combine the measured formation pressure of the research area and the adjacent area, adopt the natural gas charging abundance caused by the charging difference of oil and gas, the lifting damage dispersion, the structural action difference, the response of the ground stress load difference to the formation pressure difference, and extract the vertical depth (TVD), the acoustic time difference (AC), the neutron porosity (CNL), the density (DEN), the gamma (GR), the resistivity (RT), the porosity (POR), the permeability (PERM) and other original logging response characteristics according to the local conditions. In combination with step 2, the present application will eliminate the feature data corresponding to the mudstone part in the original logging response characteristics, and perform the "maximum-minimum normalization" on the remaining logging response feature data, so as to avoid the interference of the abnormal data in the original logging response characteristics to the prediction result of the formation pressure of the horizontal well. According to step 3, the present application also needs to apply a thousandth error of a normal distribution to the pressure coefficient of the horizontal well, so as to soften the pressure coefficient of the horizontal well, and enhance the solvability of the regression prediction model. Step 4 needs to combine the logging response characteristics after the data preprocessing of step 2 and the pressure coefficient of the horizontal well after the softening processing of step 3, train a horizontal well pressure prediction model according to the neural network model as shown in the formula (1). Figure 2 Through step 4, a nonlinear horizontal well formation pressure prediction model under the same sand group condition can be obtained. Step 5 needs to collect the corresponding original logging response characteristics as in step 1, and after the data preprocessing as shown in step 2, the trained horizontal well formation pressure prediction model can be used to predict the pressure of the unknown well. According to the actual operation result of the present application, the prediction performance of the pressure coefficient of the horizontal well is shown through the comparative analysis of the measured pressure coefficient and the predicted pressure coefficient, and the effectiveness of the present application in the oil and gas exploration field is verified.

[0090] The engineering management overall construction progress data real-time input statistical method provided by the application embodiment is applied to a computer device, the computer device comprising a memory and a processor, the memory storing a computer program, and the computer program being executed by the processor to enable the processor to execute the steps of the engineering management overall construction progress data real-time input statistical method.

[0091] The engineering management overall construction progress data real-time input statistical method provided by the application embodiment is applied to an information data processing terminal, and the information data processing terminal is used to realize the engineering management overall construction progress data real-time input statistical system.

[0092] The engineering management overall construction progress data real-time input statistical method provided by the application embodiment has the following advantages and technical effects:

[0093] 1. Higher prediction accuracy: The neural network regression model is used for horizontal well pressure prediction, which can more accurately predict the formation pressure of horizontal wells compared with traditional statistical methods and empirical formulas.

[0094] 2. Comprehensive data processing method: The method uses a variety of logging response characteristics, including vertical depth, acoustic time difference, neutron porosity, density, natural gamma, resistivity, porosity, and permeability, which can reflect the properties of the formation and improve the prediction accuracy.

[0095] 3. Strong scalability: The method is based on a neural network regression model, which can further improve prediction accuracy and generalization ability by adding more logging response characteristics and data samples.

[0096] 4. Wide applicability: The method is suitable for horizontal well formation pressure prediction under various genetic mechanisms and geological conditions, and has certain universality.

[0097] It should be noted that the embodiments of the present application can be realized by hardware, software or a combination of software and hardware. The hardware part can be realized by special logic; the software part can be stored in the memory and executed by appropriate instruction execution system, such as microprocessor or special designed hardware. Those skilled in the art can understand that the above devices and methods can be realized by computer executable instructions and / or included in processor control code, such as carrier medium, such as magnetic disk, CD or DVD-ROM, programmable memory, such as read-only memory (firmware), or data carrier, such as optical or electronic signal carrier. The device and its modules of the present application can be realized by hardware circuit, such as ultra large scale integrated circuit or gate array, semiconductor, such as logic chip, transistor, or programmable hardware device, such as field programmable gate array, programmable logic device, etc. It can also be realized by software executed by various types of processors, or by a combination of the above hardware circuit and software, such as firmware. The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any modification, equivalent replacement and improvement within the technical range disclosed by the present application, which is within the spirit and principle of the present application, should be covered within the protection scope of the present application.

Claims

1. A horizontal well formation pressure artificial intelligence prediction method, characterized in that, Comprise: Step one, the sensitive response characteristics of horizontal well logging data are constructed, combined with the measured formation pressure in the research area and adjacent area, the response of natural gas charging abundance caused by the differences in oil and gas charging, uplift damage dispersion and tectonic action, and the response of the difference in ground stress load to the difference in formation pressure, according to the differences in genetic mechanism and pressure distribution characteristics, the sensitive response parameters of horizontal well formation pressure are researched, and the vertical depth and acoustic time difference original logging response characteristics are extracted; Step two, the original logging response characteristic data is pretreated, and the characteristic data corresponding to the mudstone part in the original logging response characteristic is removed; Step three, the softening treatment of the pressure coefficient is carried out on the known horizontal well pressure coefficient; Step four, pressure prediction model training based on neural network regression, adopts a neural network model to obtain the logging response characteristics under the same sand group conditions and the pressure coefficient of the horizontal well after softening treatment The neural network regression model obtains the weights between the logical units of different network layers , and the neural network regression model for horizontal well pressure prediction is trained; Step five, the horizontal well pressure coefficient is predicted, after the vertical depth, acoustic time difference (AC), neutron porosity, density, natural gamma, resistivity, porosity and permeability of the unknown well logging are collected and pretreated as in step two, the neural network regression model trained in step 4 is used to effectively predict the pressure coefficient of the unknown horizontal well; In formations with interbedded sand and mud, the characteristic data corresponding to the mudstone portion in the original well logging response characteristics will be removed. The original well logging response characteristics need to be processed according to the following rules. Normalization is performed: ; wherein, is the maximum value data of each type of logging response feature, is the minimum value data of each type of logging response feature, represents the logging response feature, and the normalization processing of the logging response feature is strictly performed after eliminating the shale data. The known horizontal well pressure coefficient softening treatment, in particular is: ; wherein, is the original pressure coefficient of the known horizontal well, represents the pressure coefficient of the horizontal well after softening treatment.

2. The horizontal well formation pressure artificial intelligence prediction method of claim 1, wherein, The original logging response characteristics specifically include: vertical depth, acoustic time difference, neutron porosity, density, natural gamma, resistivity, porosity and permeability.

3. The horizontal well formation pressure artificial intelligence prediction method of claim 1, wherein, The neural network model, after data preprocessing of the logging response characteristics of vertical depth, acoustic time difference, neutron porosity, density, natural gamma, resistivity, porosity and permeability, obtains logging response characteristics As the input of the neural network logic unit, the predicted label of the neural network regression value adopts the softened pressure coefficient .

4. A horizontal well formation pressure artificial intelligence prediction system for implementing the horizontal well formation pressure artificial intelligence prediction method according to any one of claims 1 to 3, characterized by, The horizontal well formation pressure artificial intelligence prediction system comprises: An input module for constructing the sensitive response characteristics of horizontal well logging data; A data preprocessing module for removing the characteristic data corresponding to the mudstone part in the original logging response characteristic; A coefficient softening module for softening the known horizontal well pressure coefficient as a prediction label of neural network regression value; A training module for training a pressure prediction model based on neural network regression; A prediction module for predicting the pressure coefficient of the unknown horizontal well based on the neural network regression model trained.

5. A computer device, comprising: The computer device comprises a memory and a processor, the memory stores a computer program, and the computer program is executed by the processor to make the processor execute the steps of the horizontal well formation pressure artificial intelligence prediction method in any one of claims 1-3.

6. A computer readable storage medium storing a computer program, the computer program being executed by a processor to make the processor execute the steps of the horizontal well formation pressure artificial intelligence prediction method in any one of claims 1-3.

7. An information data processing terminal, characterized by The information data processing terminal is used to realize the horizontal well formation pressure artificial intelligence prediction system in claim 4.

Images