Model training method and device

Abstract

The invention provides a model training method and device, and relates to the technical field of model training. The model training method comprises the steps that well drilling parameters of an explored well, well logging parameters of the explored well and well drilling parameters of a target drilled well are preprocessed; constructing a first speed model according to the preprocessed well logging parameters of the explored well; according to the preprocessed drilling parameters of the target drilling well and the preprocessed drilling parameters of the explored well, predicted drilling parameters are obtained; and according to the predicted drilling parameters and seismic data of the explored well and the target drilling well, performing primary iterative optimization on the first speed model. According to the method, the first speed model is iteratively optimized according to the seismic data and the drilling parameters of the explored well and the target drilling well, and the problem that the precision is not high due to the fact that a speed model constructed in the prior art is lack of data in an area where underground media are high in heterogeneity, the number of adjacent wells is rare and logging data is lacked can be solved.

Description

Technical Field

[0001] The present application relates to the field of model training technology, and in particular to a model training method and device. Background Art

[0002] In the drilling process, it is very important to build a velocity model. The constructed velocity model can help predict the velocity characteristics of underground rock formations. This prediction can improve the efficiency and safety of drilling, reduce drilling costs and avoid potential geological hazards by optimizing drilling parameters and adjusting the pressure and speed of the drill bit; at the same time, the velocity model also plays a key role in determining the stability of the wellbore, identifying the location of oil and gas reservoirs and designing appropriate wellbore trajectories.

[0003] At present, the construction of velocity models mainly adopts the model construction method combining well data and layer data. The velocity model is constructed based on existing logging data and seismic layer data. This method is simple and easy to use, but it is difficult to fully capture the geological changes near the well to be drilled in areas with strong heterogeneity of underground media, few adjacent wells, and lack of logging data, resulting in low prediction accuracy. Summary of the invention

[0004] The present application provides a model training method and device to solve the problem of low accuracy of velocity models constructed using existing technologies due to lack of data in areas with strong underground medium heterogeneity, few adjacent wells, and lack of logging data.

[0005] A first aspect of an embodiment of the present invention provides a model training method, comprising:

[0006] Preprocessing the drilling parameters of the explored wells, the logging parameters of the explored wells and the drilling parameters of the target wells, wherein the preprocessing includes data outlier processing and data smoothing filtering;

[0007] constructing a first velocity model according to the preprocessed logging parameters of the explored wells;

[0008] Obtaining predicted drilling parameters according to the pre-processed drilling parameters of the target well and the pre-processed drilling parameters of the explored well;

[0009] The first velocity model is optimized in a first-level iterative manner according to the predicted drilling parameters and the seismic data of the explored wells and the target wells.

[0010] In a possible design, the target optimization is any optimization in the first-level iterative optimization;

[0011] For the target optimization, the first velocity model is optimized in a first level iterative manner according to the predicted drilling parameters, the seismic data of the explored wells and the target wells, including:

[0012] Optimizing the first velocity model according to the seismic data of the explored wells and the target well, and using the optimization result as the second velocity model;

[0013] The second velocity model is optimized according to the predicted drilling parameters, and the optimization result is used as the first velocity model after the target optimization.

[0014] In a possible design, before optimizing the second velocity model according to the predicted drilling parameters and using the optimization result as the first velocity model after the target optimization, the method further includes:

[0015] Based on data analysis methods and machine learning algorithms, a mapping relationship between predicted drilling parameters and predicted velocity parameters is established;

[0016] For any optimization, the optimizing the second velocity model according to the predicted drilling parameters includes:

[0017] According to the predicted drilling parameter and the mapping relationship, the predicted speed parameter is obtained;

[0018] The second speed model is optimized according to the predicted speed parameter.

[0019] In a possible design, obtaining the predicted speed parameter according to the predicted drilling parameter and the mapping relationship includes:

[0020] The predicted speed parameter corresponding to the predicted drilling parameter is obtained through the prediction formula indicated by the mapping relationship; the prediction formula is:

[0021] V pred =F(P t+h )

[0022] Among them, the V pred is the prediction speed parameter, F() is the trained time series intelligent model, and P t+h is the predicted drilling parameter.

[0023] In a possible design, optimizing the second speed model according to the predicted speed parameter includes:

[0024] According to the predicted speed parameter and the second speed model, an optimized second speed model is obtained through a second speed model update formula; the second speed model update formula is:

[0025] V2=initial(V1,V pred )

[0026] Wherein, V2 is the optimized second velocity model, V1 is the second velocity model, and Vpred is the predicted speed parameter, and initial is the updating process of the second speed model.

[0027] In a possible design, optimizing the first velocity model according to the seismic data of the explored well and the target well, and using the optimization result as the second velocity model, includes:

[0028] According to the seismic data and the first velocity model, a second velocity model is obtained by using a first velocity model update formula; the first velocity model update formula is:

[0029] V1=seismic(V)

[0030] Wherein, V1 is the second velocity model, V is the first velocity model, and seismic is the updating process of the first velocity model.

[0031] In a possible design, the method of obtaining predicted drilling parameters according to the preprocessed drilling parameters of the target well and the preprocessed drilling parameters of the explored wells includes:

[0032] The forward prediction formula indicated by the time series prediction model is used to obtain predicted drilling parameters according to the pre-processed drilling parameters of the target well at multiple time points and the pre-processed drilling parameters of the explored wells; the forward prediction formula is:

[0033] P t+h =F(P t ,P t-1 ,...,P t-n ,H)

[0034] Among them, the P t+h is the predicted drilling parameter, F() is the time series prediction model, and P t is the drilling parameter of the target well at the current time point after preprocessing, the P t-1 is the drilling parameter of the target well at a past time point after preprocessing, the P t-n are the pre-processed drilling parameters of the target well at the past n time points, and H is the pre-processed drilling parameters of the explored well.

[0035] In a possible design, constructing a first velocity model according to the preprocessed logging parameters of the explored wells includes:

[0036] According to the pre-processed logging parameters of the explored well, a first velocity parameter is obtained, and the first velocity parameter is subjected to low-pass filtering through a low-pass filtering formula to obtain a second velocity parameter; the low-pass filtering formula is:

[0037]

[0038] Among them, the V low (f) is the second speed parameter, V(f) is the first speed parameter, and f c is the cutoff frequency;

[0039] An interpolated speed parameter is obtained according to the second speed parameter, and the first speed model is obtained according to the interpolated speed parameter.

[0040] In a possible design, obtaining the interpolated speed parameter according to the second speed parameter includes:

[0041] According to the second speed parameter, the interpolated speed parameter is obtained by an interpolation formula; the interpolation formula is:

[0042]

[0043] Wherein, V(x, y, z) is the velocity parameter after interpolation, n is the number of control points, and V j is the second speed parameter, the λ j is the interpolation weight, the j is obtained through a weight formula; the weight formula is:

[0044]

[0045] Among them, the γ(d ij ) is the semivariogram function between the control points in the velocity parameter curve, and the γ(d i0 ) is the semivariogram between the control point and the interpolation point in the velocity parameter curve, μ is the Lagrangian number, and n is the number of control points.

[0046] A second aspect of an embodiment of the present invention provides a model training device, comprising:

[0047] A preprocessing module, used to preprocess the drilling parameters of the explored wells, the logging parameters of the explored wells and the drilling parameters of the target wells, wherein the preprocessing includes data outlier processing and data smoothing filtering;

[0048] A first velocity model building module, used to build a first velocity model according to the pre-processed logging parameters of the explored well;

[0049] A forward exploration prediction module is used to obtain predicted drilling parameters based on the pre-processed drilling parameters of the target well and the pre-processed drilling parameters of the explored wells;

[0050] The velocity model iterative optimization module is used to perform first-level iterative optimization on the first velocity model according to the predicted drilling parameters, seismic data of the explored wells and the target wells.

[0051] A third aspect of an embodiment of the present invention provides an electronic device, comprising: a processor and a memory communicatively connected to the processor;

[0052] The memory stores computer-executable instructions;

[0053] The processor executes the computer-executable instructions stored in the memory to implement the model training method provided in the first aspect of the embodiment of the present invention.

[0054] A fourth aspect of an embodiment of the present invention provides a computer-readable storage medium, in which computer-executable instructions are stored to implement the model training method provided in the first aspect of an embodiment of the present invention.

[0055] A fifth aspect of an embodiment of the present invention provides a computer program product, including a computer program, which, when executed by a processor, implements the model training method provided in the first aspect of an embodiment of the present invention.

[0056] The present application provides a model training method and device, which includes: preprocessing the drilling parameters of an explored well, the logging parameters of the explored well, and the drilling parameters of a target well, wherein the preprocessing includes data outlier processing and data smoothing filtering; constructing a first velocity model according to the logging parameters of the explored well after preprocessing; obtaining predicted drilling parameters according to the preprocessed drilling parameters of the target well and the preprocessed drilling parameters of the explored well; and performing a first-level iterative optimization on the first velocity model according to the predicted drilling parameters, the seismic data of the explored well and the target well. Based on the above method, the following technical effects are achieved: on the basis of constructing the first velocity model, seismic data is added. In this process, seismic data mainly focuses on the heterogeneous characteristics of the formation. Through the introduction of seismic data, especially the support of amplitude information, a more detailed description of the formation changes can be made. Therefore, the first velocity model is corrected and optimized using seismic data, which can improve the prediction accuracy of the velocity model in areas with strong heterogeneity of underground media; on the basis of seismic data, drilling parameters are further introduced as constraints. There is a low-frequency correlation between drilling parameters and logging parameters, especially in the low-frequency components in the prediction results, which has a significant improvement effect. Therefore, by integrating these low-frequency information, even in areas with a small number of adjacent wells and a lack of logging data, it is still possible to predict the parameters within a certain range of the front formation and dynamically adjust the velocity model, thereby significantly improving the accuracy of the velocity model; the present application realizes the comprehensive utilization of drilling parameters, logging parameters and seismic data, gives full play to the advantages of various types of data, provides a rich source of information for the construction and optimization of the first velocity model, and improves the overall velocity model accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0058] Figure 1 Schematic diagram of the model training method provided in the embodiment of the present application Figure 1 ;

[0059] Figure 2 A comparison chart and probability distribution chart of well data before and after preprocessing provided in the embodiment of the present application;

[0060] Figure 3 Schematic diagram of the model training method provided in the embodiment of the present application Figure 2 ;

[0061] Figure 4A matrix diagram of the correlation between the longitudinal wave velocity and various drilling parameters provided in the embodiment of the present application;

[0062] Figure 5 A schematic diagram of a model building based on an intelligent timing network provided in an embodiment of the present application;

[0063] Figure 6 A schematic diagram of changes in the drilled, trained, and predicted parts during the gradual drilling process provided by an embodiment of the present application;

[0064] Figure 7 A schematic diagram of an original velocity model in the undrilled stage provided in an embodiment of the present application;

[0065] Figure 8 A schematic diagram of an original velocity model after optimization of seismic data in the undrilled stage provided in an embodiment of the present application;

[0066] Fig. 9 A schematic diagram of an original velocity model at the first stage of drilling provided in an embodiment of the present application;

[0067] Fig.10 A schematic diagram of an original velocity model after optimization of seismic data at the start-up stage of drilling provided in an embodiment of the present application;

[0068] Fig.11 A schematic diagram of an original velocity model in the second drilling stage provided in an embodiment of the present application;

[0069] Fig.12 A schematic diagram of an original velocity model after optimization of seismic data at the second drilling stage provided in an embodiment of the present application;

[0070] Fig.13 A schematic diagram of an original velocity model in the third drilling stage provided in an embodiment of the present application;

[0071] Fig.14 A schematic diagram of an original velocity model after optimization of seismic data at the third drilling stage provided in an embodiment of the present application;

[0072] Fig.15 A schematic diagram of the structure of a model training device provided in an embodiment of the present application;

[0073] Fig.16 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0074] In the embodiments of the present application, words such as "first" and "second" are used to distinguish between identical or similar items with substantially the same functions and effects. Those skilled in the art will understand that words such as "first" and "second" do not limit the quantity and execution order, and words such as "first" and "second" do not necessarily limit the difference. It should be noted that in the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "exemplary" or "for example" in this application should not be interpreted as being more preferred or more advantageous than other embodiments or design. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way. In the embodiments of the present application, "at least one" refers to one or more, and "more" refers to two or more.

[0075] It should be noted that the "at..." in the embodiment of the present application can be the instant when a certain situation occurs, or it can be a period of time after a certain situation occurs, and the embodiment of the present application does not specifically limit this. In addition, a model training method provided in the embodiment of the present application is only used as an example, and the model training method can also include more or less content.

[0076] In order to clearly describe the technical solutions of the embodiments of the present application, some terms and technologies involved in the embodiments of the present application are briefly introduced below:

[0077] Drilling parameters: Drilling parameters refer to various physical and engineering parameters used to describe and control drilling operations during drilling operations. These parameters include: the type and specifications of the drill bit, the properties and composition of the drilling fluid, the drilling speed and rotation speed, and the drilling pressure control parameters. This application uses the drilling parameters acquired in real time to obtain the corresponding predicted logging parameters, and performs synchronous prediction of the speed parameters based on the predicted logging parameters. This forward prediction enables the speed prediction to be effectively performed in the absence of actual logging data, providing a reference prediction result for the formation ahead.

[0078] Well logging parameters: Well logging parameters refer to the real-time or intermittent measurement of the formation in the well through logging tools to obtain the physical properties and structural information of the formation. This information includes: resistivity measurement (electrical logging), natural gamma radiation measurement, acoustic wave measurement (acoustic logging) and nuclear magnetic resonance measurement.

[0079] Seismic data: In the field of drilling, seismic data mainly refers to data that detects underground geological structures through seismic waves. It helps identify the type, thickness and structure of underground rock formations by analyzing the propagation speed and reflection characteristics of seismic waves in different rock formations. Seismic data mainly focuses on the heterogeneity of strata. The introduction of seismic data provides more abundant geological information for drilling predictions, especially the support of amplitude information, which can provide a more detailed description of stratum changes, thereby improving prediction accuracy. This application uses seismic data to optimize the constructed velocity model, which can improve the prediction accuracy of the velocity model in areas with strong heterogeneity of underground media.

[0080] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0081] The technical solution of the present invention is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present invention will be described below in conjunction with the accompanying drawings.

[0082] In order to clearly understand the technical solution of the present application, the solution of the prior art is first introduced in detail.

[0083] At present, the construction of velocity models mainly adopts the model construction method combining well data and layer data. The method of building models based on existing logging data and seismic layer data is simple and easy to use, but it is difficult to fully capture the geological changes near the well to be drilled in areas with strong heterogeneity of underground media, few adjacent wells, and lack of logging data, resulting in low prediction accuracy.

[0084] Therefore, for the velocity model constructed using existing technology, in areas with strong underground medium heterogeneity, a small number of adjacent wells, and a lack of logging data, the accuracy is low due to the lack of data. In the study, it was found that in order to solve this problem, ① firstly, the drilling parameters of the explored wells, the logging parameters of the explored wells, and the drilling parameters of the target wells are preprocessed: the data preprocessing operations used include outlier removal, data smoothing filtering, etc., to improve the quality and reliability of multi-information source data, laying the foundation for subsequent analysis and model construction; ② construct the first velocity model based on the logging parameters of the explored wells after preprocessing: combine the known acoustic velocity logging information of the explored wells to construct the first velocity model around the drilling well. This first velocity model does not combine seismic amplitude, waveform information and real-time drilling parameters, but mainly relies on low-frequency logging information and seismic layer data of explored wells to reflect the macroscopic characteristics of the formation; ③ The predicted drilling parameters are obtained based on the pre-processed drilling parameters of the target well and the pre-processed drilling parameters of the explored wells; ④ The seismic data and predicted drilling parameters of the explored wells and target wells are introduced as constraints to optimize the first velocity model.

[0085] Specific:

[0086] Preprocessing the drilling parameters of the explored wells, the logging parameters of the explored wells and the drilling parameters of the target wells, the preprocessing includes data outlier processing and data smoothing filtering;

[0087] constructing a first velocity model according to the preprocessed logging parameters of the explored wells;

[0088] Obtaining predicted drilling parameters according to the pre-processed drilling parameters of the target well and the pre-processed drilling parameters of the explored well;

[0089] The first velocity model is optimized by first-level iteration according to the predicted drilling parameters, seismic data of explored wells and target wells.

[0090] The model training method of the embodiment of the present invention constructs a first velocity model based on the principle of the above-mentioned working condition preset conditions, and then continuously iterates and optimizes the first velocity model according to the preprocessed seismic data of the explored wells and the target wells, and the drilling parameters of the explored wells and the target wells. By combining the seismic data of the explored wells and the target wells, and the drilling parameters of the explored wells and the target wells, the accuracy of the constructed first velocity model can be improved in areas where the underground medium heterogeneity is strong, the number of adjacent wells is scarce, and the logging data is lacking.

[0091] Based on the above creative findings, the technical solution of the present application is proposed.

[0092] The following introduces the application scenarios of the model training method provided by the embodiments of the present invention.

[0093] The first velocity model is constructed through the model training method of this scheme, and the first velocity model is optimized based on the seismic data and drilling parameters of the explored wells and target wells. It can be applied to: By constructing a velocity model, the pressure, temperature and rock properties of the formation can be predicted, helping researchers to select the appropriate drill bit type, drilling speed and mud parameters to improve drilling efficiency; during the drilling process, the velocity model is used to monitor the drilling parameters in real time, and the operation strategy is adjusted in time to reduce the occurrence of failures and accidents; the velocity model is used to analyze the changes in formation pressure and porosity, and the wellbore stability is determined to avoid wellbore instability.

[0094] The embodiments of the present application are introduced below in conjunction with the drawings in the specification.

[0095] Figure 1 Schematic diagram of the model training method provided in the embodiment of the present application Figure 1 , the model training method provided in this embodiment includes the following steps:

[0096] S101, pre-processing the drilling parameters of the explored wells, the logging parameters of the explored wells, and the drilling parameters of the target wells.

[0097] In this embodiment, the preprocessing includes data outlier processing and data smoothing filtering. The target drilling is drilling.

[0098] The sources of preprocessed data are 3D seismic data of the work area, seismic layer data, existing exploration well data (logging, drilling, logging, etc.) and the well locations and well trajectories of the wells to be drilled in the work area. It is mainly responsible for the preliminary processing of the collected drilling parameters and logging parameters to improve the data quality and the reliability of the inversion results. Specifically, the drilling parameters of the explored wells, the logging parameters of the explored wells and the drilling parameters of the target wells are preprocessed.

[0099] Specifically, seismic data is processed by seismic data processing personnel into high-quality post-stack seismic data, while the processing of target drilling data mainly includes data outlier processing and data smoothing filtering to improve the quality and reliability of multi-information source data. Through strict processing of outliers and smoothing of signals, the stability and accuracy of input data are ensured, laying the foundation for subsequent analysis and construction of velocity models.

[0100] The purpose of data outlier processing is to detect and process outliers in target drilling parameters, especially to remove errors introduced by operating errors or instrument noise. The method of data outlier processing can use conventional methods such as the 3σ principle or box plot method to clean data. These methods can effectively identify and remove data points that deviate from the normal range, thereby improving the overall quality of the data; the purpose of data smoothing filtering is to remove high-frequency noise in target drilling parameters and retain low-frequency components that reflect the macroscopic characteristics of the formation. Low-frequency components are crucial for the construction and prediction of velocity models and can effectively capture the main trends and structures of the formation, thereby improving the smoothness and stability of the data. The specific method of data smoothing filtering can use conventional filtering techniques such as sliding average method or low-pass filter. Through efficient preprocessing of data, a solid foundation can be provided for subsequent analysis and velocity model construction, and reliable data support can be laid for real-time adjustment and optimization of the model during drilling.

[0101] S102: construct a first velocity model according to the pre-processed logging parameters of the explored wells.

[0102] In this embodiment, based on the three-dimensional seismic layer data and combined with the acoustic velocity logging parameter information of the known explored wells, a first velocity model around the drilling well, i.e., the original velocity model, is constructed. This first velocity model does not combine the seismic amplitude, waveform information and real-time drilling parameters, but mainly relies on the low-frequency logging parameter information and seismic layer data of the explored wells to reflect the macroscopic characteristics of the formation.

[0103] S103, obtaining predicted drilling parameters according to the preprocessed drilling parameters of the target well and the preprocessed drilling parameters of the explored wells.

[0104] In this embodiment, the drilling parameters of the target well and the historical drilling parameters of the explored wells are combined to gradually predict the drilling parameters of the future drilling depth. This process is called forward prediction. The introduction of real-time drilling parameters optimizes the low-frequency information in the velocity model, and this forward prediction enables effective velocity prediction in the absence of actual logging data, providing reference prediction results for the front formation, thereby significantly improving the accuracy of the velocity model.

[0105] S104, performing first-level iterative optimization on the first velocity model according to the predicted drilling parameters and the seismic data of the explored wells and the target wells.

[0106] In this embodiment, compared with the velocity model construction method based on logging parameters and seismic layer data, the velocity prediction guided by seismic data optimizes and adjusts the velocity model by supplementing the amplitude information, structural information and waveform information of the seismic data, thereby improving the accuracy and reliability of the model and solving the impact of the heterogeneity of the underground medium to a certain extent.

[0107] The objective function of the model-driven inversion method is mainly composed of data fidelity terms and regularization constraints. The data fidelity term ensures the similarity between the synthetic seismic data and the actual seismic data, while the regularization constraint term is used to control the stability of the model and avoid overfitting. The objective function can be expressed as:

[0108]

[0109] Among them, Gz is the synthetic seismic data, G is the forward operator, z is the velocity model to be inverted, d is the actual seismic data, and λ is the regularization parameter, which is used to balance the weights of the data fidelity term and the regularization constraint term.

[0110] The first item is the data fidelity item, which ensures that the synthetic seismic data Gz is close to the actual seismic data d through the least squares method; the second item is the regularization constraint item, where the velocity model Ziinitial is used to improve the stability of the inversion and reduce the multi-solution.

[0111]

[0112] Among them, d i is the ith sampling point of the actual seismic data, Gz i For the i-th sampling point of the synthetic seismic data, the optimal velocity model z can be obtained by optimizing the objective function J(z). The optimization process usually uses the gradient descent method or other numerical optimization algorithms for iterative solution.

[0113] The amplitude information in the seismic data plays a vital role in this process. It can capture and correct the reflection characteristics of the strata, so that the elastic parameters of the inversion results are consistent with the actual seismic amplitude data. In addition, the introduction of structural information and waveform information can help accurately locate the spatial distribution and geological structure of the strata, ensuring that the inversion results are not only consistent with the actual situation in terms of geological morphology, but also can accurately reproduce the propagation path and arrival characteristics of seismic waves. Compared with simple well and layer extrapolation interpolation methods, model-driven inversion combines richer seismic data, which can not only reflect the macroscopic structure of the strata, but also reveal more subtle geological features.

[0114] After obtaining the wave impedance, the specific conversion relationship between impedance, velocity and density is obtained through the intersection analysis of the velocity and density in the well logging parameters of the explored wells. The wave impedance data can be accurately converted into velocity prediction results in different sections, that is, the velocity prediction results V guided by seismic. sei .

[0115] Figure 2The well data comparison chart and probability distribution chart before and after preprocessing provided for the embodiment of the present application mainly include logging data such as longitudinal wave velocity (Vp) and density (Den), as well as drilling parameters such as drilling pressure (Wob), standpipe pressure (SPP), Dc index, mechanical drilling speed (ROP) and the corresponding probability density distribution with depth. In the preprocessing process, the outliers and noise in the data are mainly cleaned up, and the high-frequency interference is eliminated by low-pass filtering. It can be seen that the data has been significantly improved in the single-channel visualization after preprocessing, especially the abnormal data is effectively processed, and the trend is smoother. At the same time, despite the processing, the overall distribution characteristics of the data remain unchanged, and the distribution curves are still close, which ensures that the authenticity of the original data is retained.

[0116] The present application provides a model training method and device, which includes: preprocessing the drilling parameters of an explored well, the logging parameters of the explored well, and the drilling parameters of a target well; constructing a first velocity model according to the preprocessed logging parameters of the explored well; obtaining predicted drilling parameters according to the preprocessed drilling parameters of the target well and the preprocessed drilling parameters of the explored well; and performing a first-level iterative optimization of the first velocity model according to the predicted drilling parameters, seismic data of the explored well and the target well. Based on the above method, the following technical effects are achieved: on the basis of constructing the first velocity model, seismic data is added. In this process, seismic data mainly focuses on the heterogeneous characteristics of the formation. Through the introduction of seismic data, especially the support of amplitude information, a more detailed description of the formation changes can be made. Therefore, the first velocity model is corrected and optimized using seismic data, which can improve the prediction accuracy of the velocity model in areas with strong heterogeneity of underground media; on the basis of seismic data, drilling parameters are further introduced as constraints. There is a low-frequency correlation between drilling parameters and logging parameters, especially in the low-frequency components in the prediction results, which has a significant improvement effect. Therefore, by integrating these low-frequency information, even in areas with a small number of adjacent wells and a lack of logging data, it is still possible to predict the parameters within a certain range of the front formation and dynamically adjust the velocity model, thereby significantly improving the accuracy of the velocity model; the present application realizes the comprehensive utilization of drilling parameters, logging parameters and seismic data, gives full play to the advantages of various types of data, provides a rich source of information for the construction and optimization of the first velocity model, and improves the overall velocity model accuracy.

[0117] Figure 3 Schematic diagram of the model training method provided in the embodiment of the present application Figure 2 .like Figure 3 As shown, the model training method provided in this embodiment is a further refinement of the model training method provided in the previous embodiment of the present invention. The model training method provided in this embodiment includes the following steps:

[0118] S201, pre-processing the drilling parameters of the explored wells, the logging parameters of the explored wells, and the drilling parameters of the target wells.

[0119] In this embodiment, the preprocessing includes data outlier processing and data smoothing filtering.

[0120] The implementation manner and effect of S201 are similar to the implementation manner and effect of S101 in the previous embodiment of the present invention, and will not be described in detail here.

[0121] S202, obtaining a first velocity parameter according to the pre-processed logging parameters of the explored well, and performing low-pass filtering on the first velocity parameter through a low-pass filtering formula to obtain a second velocity parameter; the low-pass filtering formula is:

[0122]

[0123] Among them, V low V(f) is the second speed parameter (i.e., the speed parameter after low-pass filtering), V(f) is the first speed parameter (i.e., the speed parameter without low-pass filtering), and f c is the cutoff frequency.

[0124] In this embodiment, low-frequency information can reflect the macroscopic characteristics and overall structure of the formation, while high-frequency information often contains noise and local irregularities. These high-frequency components are easily affected by geological anomalies, logging instrument noise or human operation errors, which may lead to the accumulation of velocity model errors. Low-frequency information plays a controlling role in the construction of the first velocity model and can provide key control points for the first velocity model to ensure that the low-frequency components of the first velocity model accurately reflect the basic characteristics of the formation.

[0125] In this embodiment, the corresponding velocity curve is obtained according to the well logging parameters of the explored well after preprocessing, and the velocity curve is low-pass filtered through a low-pass filtering formula to obtain the second velocity parameter, and it also includes: synthesizing seismic records using the seismic data in the well logging data of the explored well. The purpose of synthesizing seismic records is to generate seismic data through logging parameters, so as to correct and verify the actual seismic data and ensure the accuracy of the time-depth relationship. The structural interpretation results of the seismic data obtained from the seismic records are used to construct a stratigraphic model, and combined with the geometric characteristics of the geological structure, the stratigraphic interface and structural characteristics are defined, which can more accurately meet the geomorphic characteristics. The velocity curve is low-pass filtered to extract the second velocity parameter of the velocity curve, and the second velocity parameter here belongs to low-frequency information.

[0126] S203: Obtain an interpolated speed parameter according to the second speed parameter, and obtain a first speed model according to the interpolated speed parameter.

[0127] In this embodiment, a detailed stratigraphic layer model is combined with a velocity curve having control points, and interpolation is performed in three-dimensional space to obtain a first velocity model.

[0128] Optionally, in this embodiment, according to the second speed parameter, an interpolated speed parameter is obtained by an interpolation formula (this embodiment takes the Kriging interpolation formula as an example); the interpolation formula is:

[0129]

[0130] Among them, V(x,y,z) is the velocity parameter after interpolation, n is the number of control points, V j is the second speed parameter, λ j is the interpolation weight, λ j It is obtained through the weight formula; the weight formula is:

[0131]

[0132] Among them, γ(d ij ) is the semivariogram between the control points in the velocity parameter curve, γ(d i0 ) is the semivariogram between the control point and the interpolation point in the velocity parameter, μ is the Lagrangian number, and n is the number of control points.

[0133] S204: Obtain predicted drilling parameters according to the pre-processed drilling parameters of the target well and the pre-processed drilling parameters of the explored wells.

[0134] Optionally, in this embodiment, the forward prediction formula indicated by the time series prediction model is used to obtain predicted drilling parameters according to the preprocessed drilling parameters of the target well at multiple time points and the preprocessed drilling parameters of the explored wells; the forward prediction formula is:

[0135] P t+h =F(P t ,P t-1 ,...,P t-n ,H)

[0136] Among them, P t+h To predict drilling parameters, F() is the time series prediction model, P t is the drilling parameter of the target well at the current time point after preprocessing, P t-1 is the drilling parameter of the target well at a past time point after preprocessing, P t-n is the preprocessed drilling parameter of the target well at the past n time points, and H is the preprocessed drilling parameter of the explored well.

[0137] This prediction process can continuously add real-time data as drilling progresses, and gradually predict the formation drilling parameters ahead of the drill bit. The formula is:

[0138] P t+h+1 =F(P t+1 ,P t ,...,P t-n+1 ,H)

[0139] Among them, P t+h+1 To predict drilling parameters, F() is the time series prediction model, P t+1 is the drilling parameter of the target well at the current time point after preprocessing, P t is the drilling parameter of the target well at a past time point after preprocessing, P t-n+1 is the preprocessed drilling parameter of the target well at the past n time points, and H is the preprocessed drilling parameter of the explored well.

[0140] S205, optimizing the first velocity model according to the seismic data of the explored wells and the target well, and using the optimization result as the second velocity model.

[0141] In this embodiment, seismic data is added on the basis of constructing the first velocity model, and the first velocity model is corrected and optimized through the guidance of the seismic data. This process can more accurately capture subtle changes in the strata and further improve the accuracy of the first velocity model.

[0142] Optionally, in this embodiment, the second velocity model is obtained according to the seismic data and the first velocity model by using the first velocity model update formula; the first velocity model update formula is:

[0143] V1=seismic(V)

[0144] Among them, V1 is the second velocity model, V is the first velocity model, and seismic is the update process of the first velocity model.

[0145] S206. Establish a mapping relationship between the predicted drilling parameters and the predicted speed parameters based on the data analysis method and the machine learning algorithm.

[0146] In this embodiment, after completing the data preprocessing, the drilling parameters of the explored wells (such as drilling pressure, rotation speed, Dc index, mud density, etc.) are deeply analyzed to explore the correlation between them and key parameters such as seismic velocity and density. Through methods such as data cross plots, correlation calculations and principal component analysis, combined with machine learning algorithms (such as deep learning, regression analysis, etc.), key logging parameters such as velocity are inferred from known drilling parameters, and a mapping relationship between drilling parameters and velocity parameters is established. This step provides key support for the optimization of dynamic velocity prediction while drilling.

[0147] Data correlation analysis aims to analyze the correlation between drilling parameters and logging parameters such as seismic velocity and density, and extract features useful for velocity model construction. Through data correlation analysis, key drilling parameters that can reflect underground medium information can be identified, and these drilling parameters can be used to obtain predicted target logging parameters.

[0148] Analyze the drilling parameters of the explored wells to clarify the distribution characteristics (such as mean, variance, maximum, minimum, etc.) and physical meaning of each parameter. For example, the drilling pressure (Wob) reflects the pressure exerted by the drill bit on the formation. A higher drilling pressure usually means that the drill bit is drilling in a hard formation, while a lower drilling pressure may indicate that the drill bit is working in a soft formation. The rotation speed (Rpm) affects the wear and drilling speed of the drill bit. A higher rotation speed usually means a hard formation, while a lower rotation speed may indicate a soft formation. The mud weight (Mw) is used to balance the formation pressure and prevent blowouts and well collapse.

[0149] In particular, the Dc index, as an important indicator for evaluating the difficulty of formation drilling and the efficiency of the drill bit, is usually closely related to the physical properties of the rock, such as hardness and density. Harder rocks usually have higher speed and density, resulting in a higher Dc index, while softer rocks show a lower Dc index. By analyzing the correlation between the Dc index and the velocity parameters, an important reference can be provided for the construction of the velocity model.

[0150] Figure 4 A matrix schematic diagram of the correlation between the longitudinal wave velocity and various drilling parameters provided in the embodiment of the present application shows the correlation between the longitudinal wave velocity (Vp) and multiple drilling parameters, revealing the degree of influence of each parameter on the formation velocity. The numerical values ​​in the matrix represent the correlation coefficients between different parameters, and the depth of the color reflects the strength of the correlation. Among them, Wob is the drilling pressure, Rpm is the rotation speed, Torque is the torque, Spp is the riser pressure, Flow is the drilling fluid flow, Mw is the drilling fluid density, Temper is the temperature, Dc is the Dc index, Ecd is the equivalent circulating density, and ROP is the mechanical drilling rate. The correlation coefficient between the Dc index and the longitudinal wave velocity is the highest, which is 0.91, indicating that the Dc index may play an important role in the prediction of the longitudinal wave velocity.

[0151] In the process of data analysis, it is critical to choose the right analysis method. Common methods include data intersection plot analysis, correlation calculation and principal component analysis. Through data intersection plot analysis, the relationship between drilling parameters and speed parameters can be intuitively displayed; correlation calculation can quantify the correlation coefficient between each drilling parameter and speed parameter, so as to determine which parameters are closely related to the speed parameter; principal component analysis can extract the most important characteristic parameters through dimensionality reduction technology, simplify the complexity of the model, and improve the accuracy and efficiency of the prediction. For example, for the correlation analysis between the Dc index and the speed parameter, the correlation coefficient formula is used:

[0152]

[0153] Where ρ is the correlation coefficient between the Dc index and the velocity parameter, Cov(Dc, V) is the covariance, and σ is the standard deviation.

[0154] S207. Obtain predicted speed parameters according to the predicted drilling parameters and the mapping relationship.

[0155] After obtaining the predicted drilling parameters, the correlation analysis results of the explored well data are used to determine the relationship between the predicted drilling parameters and the predicted speed parameters. Artificial intelligence learning methods, especially the time series prediction method in deep learning, are used to fit the nonlinear relationship between drilling parameters and speed parameters. An accurate mapping model is formed by training a large amount of historical data. In the actual drilling process, the drilling parameters of the explored wells are obtained in real time and input into the trained intelligent model to predict the corresponding speed parameters in real time.

[0156] In this embodiment, during the actual drilling process, the drilling parameters of the target well are obtained in real time and input into the trained time series intelligent model to obtain the corresponding predicted speed parameters. The prediction formula is as follows:

[0157] V pred =F(Wob, Rpm, Mw, Rop...)

[0158] Here, the training and prediction parameters are mainly used as examples to predict drilling parameters such as bit weight (Wob), rotation speed (Rpm), mud weight (Mw) and mechanical drilling speed (Rop).

[0159] Optionally, in this embodiment, a prediction speed parameter corresponding to the drilling parameter is obtained through a prediction formula indicated by the mapping relationship; the prediction formula is:

[0160] V pred =F(P t+h )

[0161] Among them, V predis the predicted speed parameter, F() is the trained time series intelligent model, P t+h To predict drilling parameters.

[0162] By predicting velocity parameters, they can be further applied to the optimization process of the velocity model to achieve dynamic updating and optimization of the geological model.

[0163] Figure 5 A schematic diagram of a model based on an intelligent timing network provided in an embodiment of the present application. For the model building of "drilling parameters-speed parameters" based on an intelligent timing network, first, the input drilling parameters (including various drilling parameters, such as Ecd, Dc, Flow, Mw, Torque, etc.) are extracted through a temporal convolutional network (TCN), and then the time series features are processed through a long short-term memory network (LSTM), and finally a fully connected neural network (FNN) is used for regression analysis to output the corresponding speed parameters. This intelligent timing network architecture can effectively capture the complex nonlinear relationship between drilling parameters and speed parameters, and achieve accurate prediction and matching. Figure 6 A schematic diagram of changes in the drilled, trained, and predicted parts during the gradual drilling process provided in an embodiment of the present application.

[0164] S208. Optimize the second speed model according to the predicted speed parameters, and use the optimization result as the first speed model after the target optimization.

[0165] Optionally, in this embodiment, according to the predicted speed parameter and the second speed model, the optimized second speed model is obtained by the second speed model update formula; the second speed model update formula is:

[0166] V2=initial(V1,V pred )

[0167] Among them, V2 is the optimized second velocity model, V1 is the second velocity model, V pred To predict the speed parameters, initial is the update process of the second speed model.

[0168] The second velocity model is optimized by introducing predicted velocity parameters and the velocity distribution of the second velocity model is gradually updated, so that the second velocity model is closer to the actual geological conditions.

[0169] Especially within a specific range around the drilling well, we focus on optimizing and updating the model to ensure the prediction accuracy of the model in key areas. By continuously acquiring and applying real-time drilling parameters and combining seismic data, we continuously update the velocity model. The iterative optimization process enables the model to timely reflect new information obtained during the drilling process, making the model gradually closer to the actual geological conditions and reducing prediction errors.

[0170] Figure 7 A schematic diagram of the original velocity model in the undrilled stage provided in an embodiment of the present application, Figure 8 A schematic diagram of an original velocity model after optimization of seismic data in the undrilled stage provided in an embodiment of the present application, Fig. 9 A schematic diagram of the original velocity model at the start of drilling provided in an embodiment of the present application, Fig.10 A schematic diagram of an original velocity model after optimization of seismic data at the start-up stage of drilling provided in an embodiment of the present application, Fig.11 A schematic diagram of the original velocity model in the second drilling stage provided in the embodiment of the present application, Fig.12 A schematic diagram of an original velocity model after optimization of seismic data in the second drilling phase provided in an embodiment of the present application, Fig.13 This is a schematic diagram of the original velocity model in the third drilling stage provided in the embodiment of the present application. Fig.14 A schematic diagram of the original velocity model after optimization of seismic data during the third drilling stage provided in an embodiment of the present application.

[0171] When the verification well has not been drilled, the low-frequency velocity model mainly relies on the interpolation of the velocity of adjacent wells, which may cause certain errors in areas with complex structures. Figure 7 The difference between the validation well projection and the migration results around the well can be seen in: 1) Figure 7 The low velocity area of ​​2.3s is deeply dislocated; 2) Figure 7 The low-velocity area at 2.7s is overestimated, and some layers that should appear as low-velocity areas show higher velocities; 3) Figure 7 The low-velocity layer near 3.0s is too deep. The velocity prediction results of well-seismic fusion before drilling ( Figure 8 ) captures more details in the high frequency part, but its reflection of the layer is still not accurate enough. As the drilling progresses, from stage 1 to stage 3 ( Figure 9-14 ), the low-frequency model was gradually updated by combining drilling data and seismic information, especially in the key areas of 2.3s and 2.7s, the misalignment of low velocity values ​​was corrected, and the characteristics of high- and low-velocity areas gradually emerged, which was more consistent with the actual geological structure. In stage 4, the adjustment effect of the model in the deep 3.0s area was significant.

[0172] The technical effect of this embodiment is as follows: on the basis of constructing the first velocity model, seismic data is added. In this process, seismic data mainly focuses on the heterogeneous characteristics of the strata. Through the introduction of seismic data, especially the support of amplitude information, a more detailed description of the strata changes can be made. Therefore, the first velocity model is corrected and optimized using seismic data, which can improve the prediction accuracy of the velocity model in areas with strong heterogeneity of underground media; on the basis of seismic data, drilling parameters are further introduced as constraints. There is a low-frequency correlation between drilling parameters and logging parameters, especially in the low-frequency components of the prediction results, which has a significant improvement effect. Therefore, by integrating these low-frequency information, even in areas with a small number of adjacent wells and a lack of logging data, it is still possible to predict the parameters within a certain range of the front strata, and dynamically adjust the velocity model, thereby significantly improving the accuracy of the velocity model; this application realizes the comprehensive utilization of drilling parameters, logging parameters and seismic data, gives full play to the advantages of various types of data, and provides a basis for the construction and optimization of the first velocity model. The present invention provides a rich source of information and improves the accuracy of the overall velocity model. Through data analysis methods and intelligent time series models, the implicit relationship between the real-time measured drilling parameters and velocity parameters is revealed and used to optimize the velocity model. The present invention realizes dynamic forward prediction in complex drilling scenarios without logging parameters by deeply exploring the low-frequency correlation between drilling parameters and logging parameters. The technology combines the real-time generated drilling information with historical logging data to effectively improve the accuracy and reliability of formation velocity prediction, especially in areas with complex geological conditions and strong heterogeneity, ensuring that the model can reflect formation changes in a timely manner, providing key data support for the optimization of subsequent velocity models, and significantly reducing geological hazards. The present invention forms a highly coordinated velocity model optimization and prediction system by comprehensively utilizing multi-source data in drilling scenarios, including seismic data, drilling parameters and logging parameters. The system significantly improves the prediction accuracy of geological characteristics and the reliability of the model when dealing with strong heterogeneity of underground media and difficult to accurately predict with traditional methods.

[0173] The embodiment of the present invention can divide the electronic device or the main control device into functional modules according to the above method example. For example, each functional module can be divided according to each function, or two or more functions can be integrated into one processing unit. The above integrated unit can be implemented in the form of hardware or in the form of software functional modules. It should be noted that the division of modules in the embodiment of the present invention is schematic and is only a logical function division. There may be other division methods in actual implementation.

[0174] Fig.15 This is a schematic diagram of the structure of the model training device provided in the embodiment of the present application. Fig.15 As shown, in this embodiment, the simulation training device can be located in an electronic device. The simulation training device includes:

[0175] A preprocessing module 301 is used to preprocess the drilling parameters of the explored wells, the logging parameters of the explored wells and the drilling parameters of the target wells, wherein the preprocessing includes data outlier processing and data smoothing filtering;

[0176] A first velocity model building module 302 is used to build a first velocity model according to the pre-processed logging parameters of the explored wells;

[0177] Prediction module 303: used to obtain predicted drilling parameters according to the pre-processed drilling parameters of the target well and the pre-processed drilling parameters of the explored wells;

[0178] The velocity model iterative optimization module 304 is used to perform a first-level iterative optimization on the first velocity model according to the predicted drilling parameters and the seismic data of the explored wells and the target wells.

[0179] The model training device provided in this embodiment can be executed Figure 1 The technical solution of the model training method embodiment shown in the figure has the same implementation principle and technical effect as Figure 1 The model training method embodiments shown are similar and will not be described in detail here.

[0180] At the same time, the model training device provided by the present invention further refines the model training device based on the model training device provided by the previous embodiment.

[0181] Optionally, in this embodiment, the first velocity model building module 302 is specifically used for:

[0182] According to the pre-processed logging parameters of the explored well, the first velocity parameter is obtained, and the first velocity parameter is subjected to low-pass filtering through a low-pass filtering formula to obtain a second velocity parameter; the low-pass filtering formula is:

[0183]

[0184] Among them, V low (f) is the second speed parameter, V(f) is the first speed parameter, f c is the cutoff frequency;

[0185] An interpolated speed parameter is obtained according to the second speed parameter, and a first speed model is obtained according to the interpolated speed parameter.

[0186] Optionally, when obtaining the interpolated speed parameter according to the second speed parameter, the first speed model building module 302 obtains the interpolated speed parameter according to the second speed parameter through an interpolation formula; the interpolation formula is:

[0187]

[0188] Among them, V(x,y,z) is the velocity parameter after interpolation, n is the number of control points, V j is the second speed parameter, λ j is the interpolation weight, λ j It is obtained through the weight formula; the weight formula is:

[0189]

[0190] Among them, γ(d ij ) is the semivariogram between the control points in the velocity parameter curve, γ(d i0 ) is the semivariogram between the control point and the interpolation point in the velocity parameter curve, μ is the Lagrangian number, and n is the number of control points.

[0191] Optionally, in this embodiment, the forward prediction module 303 is specifically used for:

[0192] The predicted drilling parameters are obtained according to the pre-processed drilling parameters of the target well and the pre-processed drilling parameters of the explored well.

[0193] Optionally, the forward prediction module 303 obtains predicted drilling parameters according to the preprocessed drilling parameters of the target well at multiple time points and the preprocessed drilling parameters of the explored wells through the forward prediction formula indicated by the time series prediction model; the forward prediction formula is:

[0194] P t+h =F(P t ,P t-1 ,...P t-n ,H)

[0195] Among them, P t+h To predict drilling parameters, F() is the time series prediction model, P t is the drilling parameter of the target well at the current time point after preprocessing, P t-1 is the drilling parameter of the target well at a past time point after preprocessing, P t-n is the preprocessed drilling parameter of the target well at the past n time points, and H is the preprocessed drilling parameter of the explored well.

[0196] Optionally, in this embodiment, the velocity model iterative optimization module 304 is specifically used for:

[0197] The first velocity model is optimized by first-level iteration according to the predicted drilling parameters, seismic data of explored wells and target wells.

[0198] Optionally, when the velocity model iterative optimization module 304 performs a first-level iterative optimization on the first velocity model according to the predicted drilling parameters, the seismic data of the explored wells and the target wells, the target optimization is any optimization in the first-level iterative optimization;

[0199] For the target optimization, the first velocity model is optimized in a first-level iteration based on the predicted drilling parameters, the seismic data of the explored wells and the target wells, including:

[0200] Optimizing the first velocity model according to the seismic data of the explored wells and the target wells, and using the optimization result as the second velocity model;

[0201] The second velocity model is optimized according to the predicted drilling parameters, and the optimization result is used as the first velocity model after the target optimization.

[0202] Optionally, when optimizing the second velocity model according to the predicted drilling parameters, the velocity model iteration optimization module 304 includes:

[0203] According to the predicted drilling parameters and the mapping relationship, the predicted speed parameters are obtained;

[0204] The second speed model is optimized according to the predicted speed parameters and the speed prediction result.

[0205] Optionally, when the velocity model iteration optimization module 304 obtains the predicted velocity parameters according to the predicted drilling parameters and the mapping relationship, the predicted velocity parameters corresponding to the drilling parameters are obtained by using the prediction formula indicated by the mapping relationship; the prediction formula is:

[0206] V pred =F(P t+h )

[0207] Among them, V pred is the predicted speed parameter, F() is the trained time series intelligent model, P t+h To predict drilling parameters.

[0208] Optionally, when optimizing the second velocity model according to the predicted velocity parameters, the velocity model iteration optimization module 304 obtains the optimized second velocity model through a second velocity model update formula according to the predicted velocity parameters and the second velocity model; the second velocity model update formula is:

[0209] V2=initial(V1,V pred )

[0210] Among them, V2 is the optimized second velocity model, V1 is the second velocity model, V pred To predict the speed parameters, initial is the update process of the second speed model.

[0211] Optionally, when the velocity model iteration optimization module 304 optimizes the first velocity model according to the seismic data of the explored well and the target well, and uses the optimization result as the second velocity model, the second velocity model is obtained by updating the first velocity model according to the seismic data and the first velocity model through the first velocity model updating formula; the first velocity model updating formula is:

[0212] V1=seismic(V)

[0213] Among them, V1 is the second velocity model, V is the first velocity model, and seismic is the update process of the first velocity model.

[0214] Optionally, in this embodiment, the simulation training device further includes:

[0215] Data correlation analysis module: used to establish a mapping relationship between predicted drilling parameters and predicted speed parameters based on data analysis methods and machine learning algorithms.

[0216] The simulation training device provided in this embodiment can be executed Figure 1 and Figure 3 The technical solution of the simulation training method embodiment shown in the figure has the same implementation principle and technical effect as Figure 1 and Figure 3 The simulation training method embodiments shown are similar and will not be described in detail here.

[0217] According to an embodiment of the present invention, the present invention also provides a simulation training electronic device, a computer-readable storage medium and a computer program product.

[0218] Fig.16 A schematic diagram of the structure of an electronic device provided by an embodiment of the present application. The simulation training electronic device is intended for various electronic devices that can perform the simulation training method, such as microcomputers, single-chip microcomputers, and other suitable computers. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0219] like Fig.16 As shown, the electronic device includes: at least one processor 401 and a memory 402. The electronic device also includes a communication component 403. The processor 401, the memory 402 and the communication component 403 are connected via a bus 404.

[0220] During the specific implementation process, at least one processor 401 executes the computer execution instructions stored in the memory 402, so that at least one processor 401 executes the model training method executed by the electronic device side as above.

[0221] The specific implementation process of the processor 401 can be found in the above-mentioned model training method embodiment. Its implementation principle and technical effects are similar, and will not be repeated here in this embodiment.

[0222] In the above embodiment, it should be understood that the processor 401 can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. The general-purpose processor 401 can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the invention can be directly embodied as being executed by a hardware processor, or executed by a combination of hardware and software modules in the processor.

[0223] The memory 402 may include a high-speed RAM memory, and may also include a non-volatile storage NVM, such as at least one disk storage.

[0224] The bus 404 may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus 404 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, the bus 404 in the drawings of the present application is not limited to only one bus or one type of bus.

[0225] The above-mentioned functions implemented by the electronic device and the main control device introduce the scheme provided by the embodiment of the present invention. It can be understood that in order to implement the above-mentioned functions, the electronic device or the main control device includes a hardware structure and / or software module corresponding to the execution of each function. In combination with the units and algorithm steps of each example described in the embodiment disclosed in the embodiment of the present invention, the embodiment of the present invention can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the technical solution of the embodiment of the present invention.

[0226] The present application also provides a computer-readable storage medium, which stores computer execution instructions. When a processor executes the computer execution instructions, the above model training method is implemented.

[0227] The computer-readable storage medium mentioned above can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The readable storage medium can be any available medium that can be accessed by a general or special-purpose computer.

[0228] An exemplary readable storage medium is coupled to a processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. The readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located in an application specific integrated circuit (ASIC). The processor and the readable storage medium can also exist as discrete components in an electronic device or a main control device.

[0229] The memory 402 is a non-transient computer-readable storage medium provided by the present invention. The non-transient computer-readable storage medium of the present invention stores computer instructions, and the computer instructions are used to enable a computer to execute the simulation training method provided by the present invention.

[0230] The memory 402 is a non-transient computer-readable storage medium that can be used to store non-transient software programs, non-transient computer executable programs and modules, such as program instructions / modules corresponding to the simulation training method in the embodiment of the present invention (for example, Fig.15 The processor 401 executes various functional applications and data processing by running the non-transient software programs, instructions and modules stored in the memory 402, that is, the model training method in the above method embodiment is implemented.

[0231] At the same time, this embodiment also provides a computer program product. When the instructions in the computer program product are executed by a processor, the model training method of the above embodiment can be executed.

[0232] Those skilled in the art will readily appreciate other implementations of the embodiments of the present invention after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses, or adaptations of the embodiments of the present invention that follow the general principles of the embodiments of the present invention and include common knowledge or customary technical means in the art that are not disclosed in the embodiments of the present invention. The specification and examples are to be regarded as exemplary only, and the true scope and spirit of the embodiments of the present invention are indicated by the claims.

[0233] It should be understood that the embodiments of the present invention are not limited to the precise structures described above and shown in the drawings, and various modifications and changes may be made without departing from the scope thereof. The scope of the embodiments of the present invention is limited only by the appended claims.

Claims

1. A model training method, characterized in that: include: Preprocessing the drilling parameters of the explored wells, the logging parameters of the explored wells and the drilling parameters of the target wells, wherein the preprocessing includes data outlier processing and data smoothing filtering; constructing a first velocity model according to the preprocessed logging parameters of the explored wells; Obtaining predicted drilling parameters according to the pre-processed drilling parameters of the target well and the pre-processed drilling parameters of the explored well; The first velocity model is optimized in a first-level iterative manner according to the predicted drilling parameters and the seismic data of the explored wells and the target wells.

2. The model training method according to claim 1, characterized in that: The target optimization is any optimization in the first-level iterative optimization; For the target optimization, the first velocity model is optimized in a first level iterative manner according to the predicted drilling parameters, the seismic data of the explored wells and the target wells, including: Optimizing the first velocity model according to the seismic data of the explored wells and the target well, and using the optimization result as the second velocity model; The second velocity model is optimized according to the predicted drilling parameters, and the optimization result is used as the first velocity model after the target optimization.

3. The model training method according to claim 2, characterized in that: Before optimizing the second velocity model according to the predicted drilling parameters and using the optimization result as the first velocity model after the target optimization, the method further includes: Based on data analysis methods and machine learning algorithms, a mapping relationship between predicted drilling parameters and predicted velocity parameters is established; For any optimization, the optimizing the second velocity model according to the predicted drilling parameters includes: According to the predicted drilling parameter and the mapping relationship, the predicted speed parameter is obtained; The second speed model is optimized according to the predicted speed parameter.

4. The model training method according to claim 3, characterized in that: The step of obtaining the predicted speed parameter according to the predicted drilling parameter and the mapping relationship includes: The predicted speed parameter corresponding to the predicted drilling parameter is obtained through the prediction formula indicated by the mapping relationship; the prediction formula is: V pred =F(P t+h ) Among them, the V pred is the prediction speed parameter, F() is the trained time series intelligent model, and P t+h is the predicted drilling parameter.

5. The model training method according to claim 4, characterized in that: Optimizing the second speed model according to the predicted speed parameter includes: According to the predicted speed parameter and the second speed model, an optimized second speed model is obtained through a second speed model update formula; the second speed model update formula is: V2=initial(V1,V pred ) Wherein, V2 is the optimized second velocity model, V1 is the second velocity model, and V pred is the predicted speed parameter, and initial is the updating process of the second speed model.

6. The model training method according to claim 2, characterized in that: The step of optimizing the first velocity model according to the seismic data of the explored well and the target well, and using the optimization result as the second velocity model, comprises: According to the seismic data and the first velocity model, a second velocity model is obtained by using a first velocity model update formula; the first velocity model update formula is: V1=seismic(V) Wherein, V1 is the second velocity model, V is the first velocity model, and seismic is the updating process of the first velocity model.

7. The model training method according to claim 1, characterized in that: The method of obtaining predicted drilling parameters according to the pre-processed drilling parameters of the target well and the pre-processed drilling parameters of the explored wells includes: The forward prediction formula indicated by the time series prediction model is used to obtain predicted drilling parameters according to the pre-processed drilling parameters of the target well at multiple time points and the pre-processed drilling parameters of the explored wells; the forward prediction formula is: P t+h =F(P t ,P t-1 ,...,P t-n ,H) Among them, the P t+h is the predicted drilling parameter, F() is the time series prediction model, and P t is the drilling parameter of the target well at the current time point after preprocessing, the P t-1 is the drilling parameter of the target well at a past time point after preprocessing, the P t-n are the pre-processed drilling parameters of the target well at the past n time points, and H is the pre-processed drilling parameters of the explored well.

8. The model training method according to claim 1, characterized in that: The step of constructing a first velocity model according to the pre-processed logging parameters of the explored wells comprises: According to the pre-processed logging parameters of the explored well, a first velocity parameter is obtained, and the first velocity parameter is subjected to low-pass filtering through a low-pass filtering formula to obtain a second velocity parameter; the low-pass filtering formula is: Among them, the V low (f) is the second speed parameter, V(f) is the first speed parameter, and f c is the cutoff frequency; An interpolated speed parameter is obtained according to the second speed parameter, and the first speed model is obtained according to the interpolated speed parameter.

9. The model training method according to claim 8, characterized in that: The step of obtaining the interpolated speed parameter according to the second speed parameter includes: According to the second speed parameter, the interpolated speed parameter is obtained by an interpolation formula; the interpolation formula is: Wherein, V(x, y, z) is the velocity parameter after interpolation, n is the number of control points, and V j is the second speed parameter, the λ j is the interpolation weight, the j is obtained through a weight formula; the weight formula is: Among them, the γ(d ij ) is the semivariogram function between the control points in the velocity parameter curve, and the γ(d i0 ) is the semivariogram between the control point and the interpolation point in the velocity parameter curve, μ is the Lagrangian number, and n is the number of control points.

10. A model training device, characterized in that: include: A preprocessing module, used to preprocess the drilling parameters of the explored wells, the logging parameters of the explored wells and the drilling parameters of the target wells, wherein the preprocessing includes data outlier processing and data smoothing filtering; A first velocity model building module, used to build a first velocity model according to the pre-processed logging parameters of the explored well; A forward exploration prediction module is used to obtain predicted drilling parameters based on the pre-processed drilling parameters of the target well and the pre-processed drilling parameters of the explored wells; The velocity model iterative optimization module is used to perform first-level iterative optimization on the first velocity model according to the predicted drilling parameters, seismic data of the explored wells and the target wells.

Images