Formation pressure prediction model training method, prediction method and device

Abstract

The invention discloses a formation pressure prediction model training method and device and a prediction method and device. The prediction model training method comprises the following steps: acquiring formation pressure data of at least one drilled adjacent well of a to-be-predicted well in a target block and logging data of formation pressure influence parameters of the drilled adjacent well; based on the logging data and the formation pressure data of the drilled adjacent well, obtaining an optimal influence parameter set of which the correlation between the drilled adjacent well and the formation pressure reaches a preset condition; taking the drilled adjacent well, the logging data containing the preferable influence parameters in the preferable influence parameter set and the corresponding formation pressure data as sample data, and constructing a sample data set; and based on the sample data set, training a pre-constructed formation pressure prediction model to obtain a trained formation pressure prediction model. According to the method, the sample data set is constructed through the logging data of the multiple optimal influence parameters in the optimal influence parameter set and the formation pressure data to train the model, and the prediction precision of the formation pressure prediction model can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of testing methods, and particularly to a method and device for training a formation pressure prediction model and a prediction method. Background Art

[0002] Since the 1960s, the prediction of formation pressure has been playing a very important role in the process of oil exploration. The accurate grasp of formation pressure not only has an important effect on the safety of drilling and shortening the drilling cycle, but also has an obvious impact on improving the drilling rate, cementing, protecting the oil and gas reservoir, and reducing the drilling cost. In some existing areas, the reservoir burial depth of multiple wells exceeds 7,000 m. Affected by multiple factors such as structure, hydrocarbon generation, fault, and weak diagenesis, multiple sets of high-pressure brine layers and high-pressure oil layers are developed in the reservoir in local areas. In a formation environment with high temperature and high pressure and formation pressure exceeding 2.1 g / cm3, it is inevitable to encounter high temperature and high pressure and geological trap environments during the drilling operation, making it difficult to accurately predict and detect the formation pore pressure, which seriously hinders the efficient development of ultra-deep oil and gas resources.

[0003] Due to the complex geological structure and diverse causes of abnormal pressure in some areas, the existing method of obtaining formation pressure through numerical calculation based on well logging data of a certain formation pressure influencing parameter has poor application effects. Therefore, in the prior art, machine learning methods are gradually adopted for predicting formation pore pressure, such as: a method of predicting formation pore pressure of high-pressure formations by using a BP neural network combined with well logging data, a method of establishing a supervised learning formation pressure prediction model by using SVR and LSTM (time series) for formation pressure prediction, a method of predicting formation pressure by combining prestack inversion and an SVM to correct the in-situ stress model, and a method of predicting formation pressure by combining a random forest (RF) model and well logging data in the normal pressure section, etc. Summary of the Invention

[0004] The inventors of the present application found that in the existing method of predicting formation pore pressure through machine learning, when selecting training data, data of a certain specific formation pressure influencing parameter is usually selected as sample data, and the considered formation pressure influencing parameter is relatively single. Therefore, when predicting formation pressure, there is a problem of low accuracy of the prediction result of formation pressure.

[0005] In view of the above problems, the present invention is proposed to provide a method and device for training a formation pressure prediction model and a prediction method that overcome the above problems or at least partially solve the above problems.

[0006] In a first aspect, an embodiment of the present invention provides a method for training a formation pressure prediction model, including:

[0007] Obtain the formation pressure data of at least one drilled adjacent well of the target block to be predicted and the logging data of the formation pressure influence parameters of the drilled adjacent well;

[0008] Based on the logging data and the formation pressure data of the drilled adjacent well, obtain a set of preferred influence parameters whose correlation with the formation pressure of the drilled adjacent well reaches a preset condition;

[0009] Take the drilled adjacent well, the logging data including the preferred influence parameters in the set of preferred influence parameters, and its corresponding formation pressure data as sample data to construct a sample data set;

[0010] Based on the sample data set, train a pre-constructed formation pressure prediction model to obtain a trained formation pressure prediction model.

[0011] In an optional embodiment, the obtaining a set of preferred influence parameters whose correlation with the formation pressure of the drilled adjacent well reaches a preset condition based on the logging data and the formation pressure data of the drilled adjacent well includes:

[0012] Based on the correlation between the formation pressure data of the abnormal formation pressure section of the drilled adjacent well and each influence parameter, obtain a first set of preferred influence parameters; the first set of preferred influence parameters includes: engineering parameters under drilling conditions, drilling fluid related parameters and gas logging parameters;

[0013] Through the grey relational analysis method, calculate the correlation coefficient between the logging data of each influence parameter in the first set of preferred influence parameters and the formation pressure data, and obtain a second set of preferred influence parameters whose correlation with the formation pressure of the drilled adjacent well reaches a preset condition by judging whether the magnitude of the correlation coefficient meets the preset condition.

[0014] In an optional embodiment, after obtaining a set of preferred influence parameters whose correlation with the formation pressure of each drilled adjacent well reaches a preset condition, it further includes:

[0015] Normalize the logging data of the preferred influence parameters in the set of preferred influence parameters respectively to obtain the normalized logging data of the preferred influence parameters;

[0016] Correspondingly, the sample data is: the normalized logging data of the preferred influence parameters and its corresponding formation pressure data.

[0017] In an optional embodiment, the pre-constructed formation pressure prediction model is: a support vector regression model established based on the preferred influence parameters in the set of preferred influence parameters.

[0018] In an alternative embodiment, during the training process of the formation pressure prediction model, the whale algorithm is used to optimize the model parameters of the formation pressure prediction model; the model parameters include: a penalty factor and a slack variable.

[0019] In an alternative embodiment, the kernel function selected for the support vector regression model is the radial basis kernel function.

[0020] In an alternative embodiment, the engineering data under the drilling state includes: well depth, drilling time, drilling pressure, rotational speed, and torque;

[0021] The drilling fluid related parameters include: drilling fluid inlet flow rate, drilling fluid outlet flow rate, flow rate difference between the drilling fluid inlet flow rate and the outlet flow rate, drilling fluid inlet density, drilling fluid outlet density, density difference between the drilling fluid outlet density and the inlet density, drilling fluid inlet temperature, drilling fluid outlet temperature, temperature difference between the drilling fluid inlet temperature and the outlet temperature, drilling fluid inlet conductivity, drilling fluid outlet conductivity, conductivity difference between the drilling fluid inlet conductivity and the outlet conductivity;

[0022] The gas logging parameters include: total hydrocarbon content in gas logging.

[0023] In a second aspect, an embodiment of the present invention provides a formation pressure prediction method, which is characterized by including:

[0024] Obtain real-time logging data of formation pressure influence parameters of a target well, where the formation pressure influence parameters are preferred influence parameters that are pre-obtained and whose correlation with the formation pressure meets a preset requirement;

[0025] Input the real-time logging data into the trained formation pressure prediction model, and through the formation pressure prediction model, output the formation pressure prediction result of the well to be predicted in the target block;

[0026] Among them, the formation pressure prediction model is obtained through the above-mentioned formation pressure prediction model training method.

[0027] In an alternative embodiment, the formation pressure prediction method provided by the embodiment of the present invention further includes:

[0028] Perform normalization processing on the obtained logging data to obtain the normalized logging data;

[0029] Correspondingly,

[0030] The step of inputting the logging data into the trained formation pressure prediction model is to input the normalized logging data into the trained formation pressure prediction model.

[0031] In a third aspect, an embodiment of the present invention provides a training device for a formation pressure prediction model, including: a first acquisition module, a parameter processing module, a construction module, and a training module;

[0032] The first acquisition module is configured to acquire formation pressure data of at least one drilled adjacent well of a well to be predicted in a target block and logging data of formation pressure influence parameters of the drilled adjacent well;

[0033] The parameter processing module is configured to obtain a set of preferred influence parameters whose correlation with the formation pressure of the drilled adjacent well reaches a preset condition based on the logging data and the formation pressure data of the drilled adjacent well;

[0034] The construction module is configured to use the drilled adjacent well, the logging data including the preferred influence parameters in the set of preferred influence parameters, and its corresponding formation pressure data as sample data to construct a sample data set;

[0035] The training module is configured to train a pre-constructed formation pressure prediction model based on the sample data set to obtain a trained formation pressure prediction model.

[0036] In a fourth aspect, an embodiment of the present invention provides a formation pressure prediction device, including: a second acquisition module and a prediction module;

[0037] The second acquisition module is configured to acquire real-time logging data of formation pressure influence parameters of a target well, where the formation pressure influence parameters are preferred influence parameters pre-acquired with a correlation with the formation pressure meeting a preset requirement;

[0038] The prediction module is configured to input the real-time logging data into the trained formation pressure prediction model, and output a formation pressure prediction result of a well to be predicted in a target block through the formation pressure prediction model;

[0039] Wherein, the formation pressure prediction model is obtained through the above-mentioned formation pressure prediction model training method.

[0040] In a fifth aspect, an embodiment of the present invention provides a computer storage medium, characterized in that computer-executable instructions are stored in the computer storage medium, and when the computer-executable instructions are executed by a processor, the above-mentioned formation pressure prediction model training method and the above-mentioned formation pressure prediction method are implemented.

[0041] In a sixth aspect, an embodiment of the present invention further provides a computer device, including: a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the program, the above-mentioned formation pressure prediction model training method and the above-mentioned formation pressure prediction method are implemented.

[0042] The beneficial effects of the above technical solutions provided by the embodiments of the present invention at least include:

[0043] The formation pressure prediction model training method provided by the embodiments of the present invention predicts the formation pressure of un-drilled wells through the logging data of drilled wells in the target block; when selecting the sample data for model training, based on the logging data of the formation pressure influence parameters of at least one drilled adjacent well of the well to be predicted and its formation pressure data obtained, a preferred influence parameter set with a formation pressure correlation reaching a preset condition is obtained, and the logging data containing the preferred influence parameters in the preferred influence parameter set and its corresponding formation pressure data are used as sample data to construct a sample data set, and the pre-constructed formation pressure prediction model is trained to obtain a trained formation pressure prediction model and applied to actual formation pressure prediction. When determining the sample data, this method fully considers various influence parameters of the formation pressure, and constructs a sample data set based on multiple influence parameters in the obtained preferred influence parameter set, so that the formation pressure predicted by the finally obtained trained formation pressure prediction model is more consistent with the actual pressure. Therefore, when predicting the formation pressure of the well to be predicted, the accuracy of the prediction result can be improved, and it has practical value.

[0044] Other features and advantages of the present invention will be described in the following description, and part of them will become obvious from the description, or be understood by implementing the present invention. The objectives and other advantages of the present invention can be achieved and obtained through the structures specifically pointed out in the written description, claims, and drawings.

[0045] The technical solutions of the present invention will be further described in detail below through the drawings and embodiments. Description of the Drawings

[0046] The drawings are used to provide a further understanding of the present invention, and constitute a part of the description. They are used together with the embodiments of the present invention to explain the present invention, and do not constitute a limitation to the present invention. In the drawings:

[0047] Figure 1 It is a schematic flow chart of the formation pressure prediction model training method in the embodiments of the present invention;

[0048] Figure 2 It is a comparison chart of the formation pressure data predicted by the formation pressure prediction model trained by using the formation pressure prediction model training method of the embodiments of the present invention and the measured formation pressure data;

[0049] Figure 3 It is a schematic flow chart of the formation pressure prediction method in the embodiments of the present invention;

[0050] Figure 4Schematic diagram of the formation pressure prediction model training device in the embodiments of the present invention;

[0051] Figure 5 Schematic diagram of the formation pressure prediction device in the embodiments of the present invention. Detailed implementation manners

[0052] Hereinafter, exemplary embodiments of the present disclosure will be described in more detail with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully conveyed to those skilled in the art.

[0053] In order to accurately grasp the formation pressure of the un-drilled well and solve the problem that the prediction accuracy of the formation pressure porosity pressure by the machine learning method in the prior art is not high, the embodiments of the present invention provide a formation pressure prediction model training method, a prediction method and a device.

[0054] The formation pressure prediction model training method provided by the embodiments of the present invention, its flowchart is referred to Figure 1 as shown, and includes the following steps:

[0055] Step S101: Obtain the formation pressure data of at least one drilled adjacent well of the well to be predicted in the target block and the logging data of the formation pressure influence parameters of the drilled adjacent well;

[0056] Step S102: Based on the logging data and the formation pressure data of the formation pressure influence parameters of the drilled adjacent well, obtain a preferred influence parameter set in which the correlation between the drilled adjacent well and the formation pressure reaches a preset condition;

[0057] Step S103: Use the drilled adjacent well, the logging data including the preferred influence parameters in the preferred influence parameter set, and its corresponding formation pressure data as sample data to construct a sample data set;

[0058] Step S104: Based on the sample data set, train the pre-constructed formation pressure prediction model to obtain a trained formation pressure prediction model.

[0059] Optionally, when selecting the drilled adjacent wells of the well to be predicted, the specific number of selected wells is not specifically limited in the embodiments of the present invention; it is known that the more adjacent wells are selected, the more similar the formation pressure system of the adjacent wells is to the well to be predicted, and the more accurate the result obtained when analyzing the formation pressure of the well to be predicted after training the pre-constructed formation pressure prediction model.

[0060] Among them, in step S101, formation pressure data of adjacent drilled wells and well logging data of their corresponding formation pressure influence parameters are obtained. Specifically, it can be obtained according to relevant measured well logging data of the drilled wells, and the formation pressure data of adjacent drilled wells and the well logging data of their corresponding formation pressure influence parameters obtained in the embodiments of the present invention are specifically data after removing outliers in the relevant well logging data.

[0061] In an optional embodiment, in step S102, based on the well logging data and formation pressure data of adjacent drilled wells, a set of preferred influence parameters with the correlation between adjacent drilled wells and formation pressure reaching a preset condition is obtained, including:

[0062] Based on the correlation between the formation pressure data of the abnormal formation pressure section of adjacent drilled wells and each influence parameter, a first set of preferred influence parameters is obtained; the first set of preferred influence parameters includes: engineering parameters under drilling conditions, drilling fluid related parameters, and gas logging parameters.

[0063] By using the grey relational analysis method, the correlation coefficient between the well logging data of each influence parameter and the formation pressure data in the first set of preferred influence parameters is calculated. By judging whether the magnitude of the correlation coefficient meets the preset condition, a second set of preferred influence parameters with the correlation between adjacent drilled wells and formation pressure reaching the preset condition is obtained.

[0064] Specifically, the formation pressure data includes: formation pressure data under normal pressure conditions, and formation pressure under abnormal pressure (high pressure, ultra-high pressure) conditions. Moreover, the correlation between formation pressure and each influence parameter can be more reflected in the abnormal formation pressure section. Therefore, after obtaining the formation pressure data of adjacent drilled wells, the relevant data affected by engineering factors and causing pressure changes can be removed first, and each formation pressure influencing factor can be initially screened through the formation pressure data of the abnormal formation pressure section to obtain a first set of influence parameters.

[0065] After obtaining the first set of influence parameters, in order to quantitatively analyze the influence degree of each influence parameter on formation pore pressure and reduce the interference of redundant parameters on formation pressure prediction, the embodiments of the present invention also use grey relational analysis to calculate the correlation magnitude between the well logging data of each influence parameter and the formation pressure. The specific calculation formula is as follows:

[0066]

[0067]

[0068] In the formula, γ(X 0 , X i ) is the correlation coefficient; X 0 is the comparison sequence; x 0 (k) is the k-th data of the data set; X i is the correlation sequence |x0 (k)-x i (k)| is the k-th data X 0 and X i the absolute value of the difference; min i min k |x 0 (k)-x i (k)|, max i max k |x 0 (k)-x i (k)| are the two-level minimum difference and the maximum difference respectively; α is the discrimination coefficient (the value range is (0, 1), and the greater the difference in the correlation coefficient, the stronger the discrimination ability).

[0069] Among them, the comparison sequence X 0 is the formation pressure data set of the adjacent wells that have been drilled; the correlation sequence X i is the logging data set of the influencing factors of the formation pressure of each completed well that has been drilled; the two-level minimum difference min i min k |x 0 (k)-x i (k)| is the difference between the minimum value of the formation pressure data of the adjacent wells that have been drilled and the minimum value of the logging data of the influencing factors of each formation pressure, and similarly, the two-level maximum difference max i max k |x 0 (k)-x i (k)| is the difference between the maximum value of the formation pressure data of the adjacent wells that have been drilled and the maximum value of the logging data of the influencing factors of each formation pressure.

[0070] And the value range of the discrimination coefficient α is (0, 1), and the size of its selection mainly affects the size of the difference between the correlation coefficients finally solved. In specific applications, it can be determined according to actual needs, and the embodiments of the present invention do not make specific limitations on this.

[0071] Specifically, in the first optimized influence parameter set, the engineering data in the drilling state includes: well depth, drilling time, drilling pressure, rotation speed, torque;

[0072] The drilling fluid related parameters include: drilling fluid inlet flow rate, drilling fluid outlet flow rate, flow rate difference between the drilling fluid inlet flow rate and the outlet flow rate, drilling fluid inlet density, drilling fluid outlet density, density difference between the drilling fluid outlet density and the inlet density, drilling fluid inlet temperature, drilling fluid outlet temperature, temperature difference between the drilling fluid inlet temperature and the outlet temperature, drilling fluid inlet conductivity, drilling fluid outlet conductivity, difference between the drilling fluid inlet conductivity and the outlet conductivity;

[0073] The gas logging parameters include: gas logging total hydrocarbon content.

[0074] From the relevant parameters in the first set of optimized influence parameters, it can be seen that when selecting the formation pressure influence parameters in the embodiments of the present invention, various aspects of influence parameters are fully considered, especially the drilling fluid related parameters with a relatively high correlation with abnormal formation pressure conditions.

[0075] It should be noted that for the preset conditions satisfied by the correlation coefficient when determining the second set of preferred influence parameters by the grey correlation method, they can be selected according to actual needs. For example, the obtained correlation coefficient greater than 0.6 can be selected as the preset condition, or the correlation coefficient greater than 0.8 can be selected as the preset condition; correspondingly, under different preset conditions, the preferred influence parameters in the second set of preferred influence parameters that meet the preset conditions will also change accordingly. Therefore, the logging data of the preferred influence parameters for constructing the sample data set is not specifically limited in the embodiments of the present invention.

[0076] Optionally, in the embodiments of the present invention, with the resolution coefficient α selected as 0.5, the formation pressure influence parameters with a correlation coefficient greater than 0.6 are used as the optimized influence parameters of the second set of preferred influence parameters, and the obtained optimized influence parameters are as follows:

[0077] Among them, the engineering parameters include: well depth, drilling time, drilling pressure, rotary speed, torque;

[0078] The drilling fluid related parameters include: drilling fluid inlet density, drilling fluid outlet density, drilling fluid inlet temperature, drilling fluid outlet temperature, temperature difference between the drilling fluid inlet temperature and the outlet temperature,

[0079] The gas logging parameters include: total hydrocarbon content of gas logging.

[0080] In an optional embodiment, for the formation pressure prediction model training method provided by the embodiments of the present invention, after obtaining the set of preferred influence parameters with a correlation between each drilled adjacent well and the formation pressure reaching the preset condition in step S102, it further includes: respectively performing normalization processing on the logging data of the preferred influence parameters in the set of preferred influence parameters to obtain the normalized logging data of the preferred influence parameters;

[0081] Correspondingly, the sample data is: the normalized logging data of the preferred influence parameters and their corresponding formation pressure data.

[0082] In order to eliminate the dimension and reduce the influence of different orders of magnitude between various influence parameters on the subsequent model training results, after determining the set of influence parameters with a correlation between each drilled adjacent well and the formation pressure reaching the preset condition, the logging data of each influence parameter in the set of influence parameters can be normalized. The embodiments of the present invention do not specifically limit the method for normalizing relevant data; for example: the Max-Min normalization method can be used to normalize it. At this time, the normalization function is:

[0083]

[0084] In the formula, represents the result after normalization of the i-th parameter data in the dataset; X i is the i-th input data, MAX(X i ) is the maximum data in the input dataset, and MIN(X i ) is the minimum data in the input dataset; where the dataset in the formula is the logging data of the influencing parameters to be normalized.

[0085] Furthermore, the formation pressure prediction model pre-constructed in the embodiment of the present invention is: a support vector regression model established according to the preferred influencing parameters in the preferred influencing parameter set.

[0086] In an optional embodiment, in the model training method of the embodiment of the present invention, in step S103, during the training process of the formation pressure prediction model, the whale algorithm is used to optimize the model parameters of the formation pressure prediction model to make the formation pressure prediction model optimal; the model parameters include: penalty factor and relaxation variable.

[0087] Compared with the current formation pressure prediction neural network model, during the model training process, the parameter settings such as weight thresholds are all from random values or multiple comparison values. Using the whale algorithm to optimize the model parameters of the formation pressure prediction model can make the model reach the global optimum, and thus make the prediction accuracy of the trained model higher. Moreover, since the penalty factor and the relaxation variable are two factors that have a greater impact on the support vector regression model, making the penalty factor and the relaxation variable reach the optimum through the whale algorithm can further improve the accuracy of the support vector regression model while improving the optimization speed of the model.

[0088] The following takes the optimization process of the whale algorithm (WOA) as an example to optimize the penalty factor C of the support vector regression model, and the optimization process is exemplarily described as the following three steps:

[0089] The first step: Encircle the prey (each possibility of the penalty factor C is a prey). Identify the position of the prey and surround it. The mathematical expression for updating its position is:

[0090]

[0091] In the formula, A is the convergence factor; t is the current iteration number; Y*(t) is the optimal position vector of the whale; Y is the position of the whale; a is the convergence parameter, and linearly decreases from 2 to 0 during the iteration; r 1 、r 2is a random number uniformly distributed in [0, 1].

[0092] Step 2: Hunting. Humpback whales adopt the predation methods of blowing bubbles or shrinking the encirclement. The position update between the whales and the prey is expressed as a logarithmic spiral equation:

[0093]

[0094] where: D′ is the distance between the search object and the current optimal solution; l is a random number in [0, 1]; p is the predation mechanism probability, which is a random number in [0, 1]; b is the logarithmic spiral shape parameter.

[0095] Step 3: Prey search. The formula is:

[0096]

[0097] where: Y′(t) is the position of the random object; D″ is the distance between the search object and the random object. When the humpback whale deviates from the prey, it will look for a more suitable prey to enhance the global search ability.

[0098] Specifically, using the whale algorithm to optimize the slack variable ξ of the formation pressure prediction model i The specific process of optimization can refer to the optimization process of the above penalty factor C, and this embodiment of the present invention will not elaborate on it in detail.

[0099] Furthermore, using the whale algorithm to optimize the model parameters of the formation pressure prediction model to make the formation pressure prediction model optimal, including:

[0100] After completing one optimization iteration process of the model parameters of the formation pressure prediction model using the whale algorithm, based on the optimized model parameters, the formation pressure prediction model is trained, and it is judged whether to update the relevant model parameters of the pressure prediction model according to the result of the model training: if the result of the model training is better than the result of the model training under the previous set of model parameters, the model parameters are updated; if the result of the model training is worse than the result of the model training under the previous set of model parameters, this set of model parameters is discarded; then the process of optimizing and iterating the model parameters using the whale algorithm is repeated until the formation pressure prediction model reaches the optimal.

[0101] In an optional embodiment, the kernel function selected by the support vector regression model of this embodiment of the present invention can preferably be a radial basis kernel function, and its formula is as follows:

[0102]

[0103] where, k(x, x i ) is the kernel function, and x i , x j are the training set samples.

[0104] In an optional embodiment, the model training method provided by the embodiments of the present invention is used, and a sample data set is constructed by using the relevant logging data and formation pressure data of the adjacent well HT101 that has been drilled in Well LH1 to be predicted in a certain area. The constructed formation pressure prediction model is trained to obtain the formation pressure prediction model of the well to be predicted, and this model is used to predict Well LH1. A partial comparison diagram of the predicted formation pressure result and the measured formation pressure result of this well is as Figure 2 shown. It can be seen from the figure that the coincidence degree of the predicted formation pressure and the measured formation pressure is very high, and the prediction accuracy can reach more than 90%.

[0105] The formation pressure prediction model training method provided by the embodiments of the present invention analyzes the variation law between the logging data of the formation pressure influence parameters screened and the formation pressure curve to predict the variation law of the abnormal formation pressure of the undrilled well; when knowing the logging data of the formation pressure influence parameters and the formation pressure data of multiple adjacent wells of the well to be predicted, for the logging data of multiple adjacent wells, the sensitivity parameters with stronger correlation with the formation pressure are screened out from the logging parameters, and a support vector regression model is established based on the screened parameters. During the continuous correction of the model, the whale algorithm is used to optimize the model parameters in the support vector regression model, and then a non-linear mapping relationship between the sensitivity parameters and the formation pressure is constructed, and the corrected model is applied to the formation pressure prediction of the target well to qualitatively evaluate the downhole formation pressure. In this model training method, when determining the sample data for model training, all the influence parameters of the formation pressure are fully considered, and a sample data set is constructed based on multiple influence parameters in the obtained optimal influence parameter set, so that the formation pressure predicted by the finally obtained trained formation pressure prediction model is more consistent with the actual pressure. Therefore, when predicting the formation pressure of the well to be predicted, it can improve the accuracy of the prediction result and has practical value.

[0106] Based on the same inventive concept, the embodiments of the present invention also provide a formation pressure prediction method, and its flowchart is referred to Figure 3 shown, and includes the following steps:

[0107] Step S201: Obtain the real-time logging data of the formation pressure influence parameters of the target well, where the formation pressure influence parameters are the optimal influence parameters pre-obtained with the correlation with the formation pressure meeting the preset requirements;

[0108] Step S202: Input the real-time logging data into the trained formation pressure prediction model, and through the formation pressure prediction model, output the formation pressure prediction result of the well to be predicted in the target block;

[0109] Among them, the formation pressure prediction model is obtained by the above-mentioned formation pressure prediction model training method.

[0110] Optionally, the formation pressure prediction method provided by the embodiments of the present invention further includes: performing normalization processing on the real-time logging data of the formation pressure influence parameters obtained to obtain the normalized real-time logging data;

[0111] Correspondingly,

[0112] Input the real-time logging data into the trained formation pressure prediction model, that is, input the normalized real-time logging data into the trained formation pressure prediction model.

[0113] The method provided by the embodiments of the present invention uses the trained formation pressure prediction model to perform real-time pressure prediction. From the relevant content of the above formation pressure prediction model training method, it can be seen that the formation pressure prediction model used in the embodiments of the present invention has high prediction accuracy, can well evaluate the prediction of the formation pressure of the un-drilled well section, and has practical value.

[0114] Based on the same inventive concept, the embodiments of the present invention further provide a training device for a formation pressure prediction model. Referring to Figure 4 as shown, it includes: a first acquisition module, a parameter processing module, a construction module, and a training module;

[0115] The first acquisition module 11 is used to acquire the formation pressure data of at least one drilled adjacent well of the well to be predicted in the target block and the logging data of the formation pressure influence parameters of the drilled adjacent well;

[0116] The parameter processing module 12 is used to obtain a set of preferred influence parameters with a preset correlation between the drilled adjacent well and the formation pressure based on the logging data and formation pressure data of the drilled adjacent well;

[0117] The construction module 13 is used to use the drilled adjacent well, the logging data including the preferred influence parameters in the set of preferred influence parameters, and its corresponding formation pressure data as sample data to construct a sample data set;

[0118] The training module 14 trains the pre-constructed formation pressure prediction model based on the sample data set to obtain a trained formation pressure prediction model.

[0119] Regarding the training device for the formation pressure prediction model in the above embodiments, the specific manners in which each module performs operations have been described in detail in the embodiments of the training method for the formation pressure prediction model, and will not be elaborated here.

[0120] Based on the same inventive concept, the embodiments of the present invention further provide a formation pressure prediction device. Referring to Figure 5 as shown, it includes: a second acquisition module and a prediction module;

[0121] The second acquisition module 21 is configured to acquire real-time logging data of formation pressure influence parameters of a target well, where the formation pressure influence parameters are preferably influence parameters that are pre-acquired and whose correlation with the formation pressure meets a preset requirement;

[0122] The prediction module 22 is configured to input the real-time logging data into a trained formation pressure prediction model, and output a formation pressure prediction result of a well to be predicted in a target block through the formation pressure prediction model;

[0123] Wherein, the formation pressure prediction model is obtained by the above-mentioned formation pressure prediction model training method.

[0124] Regarding the formation pressure prediction device in the above embodiments, the specific manners in which each module performs operations have been described in detail in the embodiments related to the formation pressure prediction method, and will not be elaborated herein.

[0125] Based on the same inventive concept, an embodiment of the present invention further provides a computer storage medium, where computer-executable instructions are stored in the computer storage medium, and when the computer-executable instructions are executed by a processor, the above-mentioned formation pressure prediction model training method and the above-mentioned formation pressure prediction method are implemented.

[0126] Based on the same inventive concept, an embodiment of the present invention further provides a computer device, including: a memory, a processor, and a computer program stored on the memory and executable on the processor, where when the processor executes the program, the above-mentioned formation pressure prediction model training method and the above-mentioned formation pressure prediction method are implemented.

[0127] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) containing computer-usable program code.

[0128] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate for implementing in the process Figure 1 one process or multiple processes and / or blocksFigure 1 means for the functions specified in one or more boxes.

[0129] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means that implement the functions specified in one Figure 1 or more processes and / or boxes Figure 1 means for the functions specified in one or more boxes.

[0130] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in one Figure 1 or more processes and / or boxes Figure 1 means for the functions specified in one or more boxes.

[0131] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these changes and modifications.

Claims

1. A method for training a formation pressure prediction model, characterized in that, it includes: Obtain the formation pressure data of at least one drilled adjacent well of the well to be predicted in the target block and the logging data of the formation pressure influence parameters of the drilled adjacent well; Based on the logging data and the formation pressure data of the drilled adjacent well, obtain a set of preferred influence parameters whose correlation with the formation pressure of the drilled adjacent well reaches a preset condition; Take the drilled adjacent well, the logging data including the preferred influence parameters in the set of preferred influence parameters, and its corresponding formation pressure data as sample data to construct a sample data set; Based on the sample data set, train a pre-constructed formation pressure prediction model to obtain a trained formation pressure prediction model.

2. The method for training a formation pressure prediction model according to claim 1, characterized in that, Based on the logging data and the formation pressure data of the drilled adjacent well, obtain a set of preferred influence parameters whose correlation with the formation pressure of the drilled adjacent well reaches a preset condition; including: Based on the correlation between the formation pressure data of the abnormal formation pressure section of the drilled adjacent well and each influence parameter, obtain a first set of preferred influence parameters; the first set of preferred influence parameters includes: engineering parameters under drilling conditions, drilling fluid related parameters and gas logging parameters; Through the grey relational analysis method, calculate the correlation coefficient between the logging data of each influence parameter in the first set of preferred influence parameters and the formation pressure data, and obtain a second set of preferred influence parameters whose correlation with the formation pressure of the drilled adjacent well reaches a preset condition by judging whether the magnitude of the correlation coefficient meets the preset condition.

3. The method for training a formation pressure prediction model according to claim 1, characterized in that, After obtaining a set of preferred influence parameters whose correlation with the formation pressure of each drilled adjacent well reaches a preset condition, it further includes: Normalize the logging data of the preferred influence parameters in the set of preferred influence parameters respectively to obtain the normalized logging data of the preferred influence parameters; Correspondingly, the sample data is: the normalized logging data of the preferred influence parameters and its corresponding formation pressure data.

4. The method for training a formation pressure prediction model according to claim 1, characterized in that, The pre-constructed formation pressure prediction model is: a support vector regression model established based on the preferred influence parameters in the set of preferred influence parameters.

5. The method for training a formation pressure prediction model according to claim 4, characterized in that, During the training process of the formation pressure prediction model, use the whale algorithm to optimize the model parameters of the formation pressure prediction model; the model parameters include: penalty factor and slack variable.

6. The method for training a formation pressure prediction model according to claim 5, characterized in that, The kernel function selected by the support vector regression model is a radial basis kernel function.

7. The method for training a formation pressure prediction model according to any one of claims 2-6, characterized in that, The engineering data under drilling conditions includes: well depth, drilling time, drilling pressure, rotation speed, torque; The relevant parameters of the drilling fluid include: the inlet flow rate of the drilling fluid, the outlet flow rate of the drilling fluid, the flow rate difference between the inlet and outlet flow rates of the drilling fluid, the inlet density of the drilling fluid, the outlet density of the drilling fluid, the density difference between the outlet and inlet densities of the drilling fluid, the inlet temperature of the drilling fluid, the outlet temperature of the drilling fluid, the temperature difference between the inlet and outlet temperatures of the drilling fluid, the inlet conductivity of the drilling fluid, the outlet conductivity of the drilling fluid, and the conductivity difference between the inlet and outlet conductivities of the drilling fluid; The gas logging parameters include: the total hydrocarbon content of gas logging.

8. A formation pressure prediction method, characterized in that, it includes: Obtain the real-time logging data of the formation pressure influence parameters of the target well, where the formation pressure influence parameters are the preferred influence parameters obtained in advance and having a correlation with the formation pressure meeting the preset requirements; Input the real-time logging data into the trained formation pressure prediction model, and through the formation pressure prediction model, output the formation pressure prediction result of the well to be predicted in the target block; Among them, the formation pressure prediction model is obtained by the formation pressure prediction model training method described in any one of claims 1-7.

9. The formation pressure prediction method according to claim 7, characterized in that, it further includes: Perform normalization processing on the obtained logging data to obtain the normalized logging data; Correspondingly, The step of inputting the logging data into the trained formation pressure prediction model is to input the normalized logging data into the trained formation pressure prediction model.

10. A training device for a formation pressure prediction model, characterized in that, it includes: A first acquisition module, a parameter processing module, a construction module and a training module; The first acquisition module is used to acquire the formation pressure data of at least one drilled adjacent well of the well to be predicted in the target block and the logging data of the formation pressure influence parameters of the drilled adjacent well; The parameter processing module is used to obtain a set of preferred influence parameters whose correlation with the formation pressure of the drilled adjacent well reaches the preset condition based on the logging data and the formation pressure data of the drilled adjacent well; The construction module is used to use the drilled adjacent well, the logging data including the preferred influence parameters in the set of preferred influence parameters, and its corresponding formation pressure data as sample data to construct a sample data set; The training module is used to train the pre-constructed formation pressure prediction model based on the sample data set to obtain a trained formation pressure prediction model.

11. A formation pressure prediction device, characterized in that, it includes: A second acquisition module and a prediction module; The second acquisition module is used to acquire the real-time logging data of the formation pressure influence parameters of the target well, where the formation pressure influence parameters are the preferred influence parameters obtained in advance and having a correlation with the formation pressure meeting the preset requirements; The prediction module is used to input the real-time logging data into the trained formation pressure prediction model, and through the formation pressure prediction model, output the formation pressure prediction result of the well to be predicted in the target block; Among them, the formation pressure prediction model is obtained by the formation pressure prediction model training method described in any one of claims 1-7.

12. A computer storage medium, characterized in that, the computer storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by a processor, the formation pressure prediction model training method according to any one of claims 1-7 and the formation pressure prediction method according to any one of claims 8-9 are implemented.

13. A computer device, characterized in that, comprising: a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the program, the formation pressure prediction model training method according to any one of claims 1-7 and the formation pressure prediction method according to any one of claims 8-9 are implemented.

Images