Curve reconstruction model training and curve reconstruction method and device, equipment and medium

Abstract

The embodiment of the invention discloses a curve reconstruction model training method and device, a curve reconstruction method and device, electronic equipment and a storage medium. The method comprises the following steps: training required target sample logging data according to a logging curve reconstruction model, and executing a logging data spatio-temporal feature extraction task; according to the well logging data spatial-temporal feature extraction task, obtaining sample well logging data after spatial-temporal feature extraction, and controlling the well logging curve reconstruction model to execute a well logging curve reconstruction training task; and according to the well logging data spatial-temporal feature extraction task and the well logging curve reconstruction training task, performing parameter adjustment on the well logging curve reconstruction model to obtain a convergent well logging curve reconstruction model. By adopting the technical scheme of the embodiment of the invention, the attention mechanism is fused into the two-dimensional convolutional neural network, so that the ability of the deep learning network to capture the self-correlation and cross-correlation feature information of the logging curve is enhanced, and the precision and generalization ability of the deep learning network to reconstruct the acoustic wave and density logging curve are improved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of petroleum geophysical exploration, and in particular to a curve reconstruction model training and curve reconstruction method, device, equipment and medium. Background Art

[0002] In the quantitative interpretation of seismic data, density and acoustic time-difference logging curves are two key curves in seismic rock physics and seismic inversion technology. These two curves can provide a reliable time-depth relationship to calibrate depth-domain logging data and time-domain seismic data. They are curves of rock elasticity and physical property information that are essential to rock physics modeling technology.

[0003] However, in actual logging, the density and acoustic logging curves are abnormal or missing due to the influence of boreholes, logging instruments and human factors at the well site, which brings certain difficulties to the subsequent comprehensive interpretation of logging and seismic exploration. In addition, the existing empirical model method, multivariate fitting method and rock physics modeling method not only have the problem of low accuracy in reconstructing the target curve, but also have the problem of difficulty in handling the simultaneous reconstruction of two curves. Summary of the invention

[0004] The embodiment of the present invention provides a curve reconstruction model training and curve reconstruction method, device, electronic device and storage medium to improve the accuracy and generalization ability of acoustic wave and density logging curves.

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

[0006] Reconstruct the target sample logging data required for model training based on the logging curves and perform the task of extracting spatiotemporal features of the logging data;

[0007] According to the spatiotemporal feature extraction task of the logging data, sample logging data after spatiotemporal feature extraction is obtained, and the logging curve reconstruction model is controlled to perform the logging curve reconstruction training task; wherein the sample logging data after spatiotemporal feature extraction is the target sample logging data required for the training of the logging curve reconstruction model; the logging curve reconstruction training task is used to train the curve reconstruction model to reconstruct the logging curve;

[0008] According to the well logging data spatiotemporal feature extraction task and the well logging curve reconstruction training task, the well logging curve reconstruction model is parameter adjusted to obtain a converged well logging curve reconstruction model.

[0009] In a second aspect, an embodiment of the present invention further provides a curve reconstruction method, wherein the curve reconstruction model is obtained by using the training method of the curve reconstruction model described in any embodiment of the present invention, and the curve reconstruction method comprises:

[0010] Determine the well logging data for the area to be explored;

[0011] The logging data is input into the curve reconstruction model to obtain the acoustic wave and density reconstruction logging curve of the area to be explored.

[0012] In a third aspect, an embodiment of the present invention further provides a curve reconstruction model training device, comprising:

[0013] The spatiotemporal feature extraction module is used to reconstruct the target sample logging data required for model training based on the logging curves and perform the spatiotemporal feature extraction task of the logging data;

[0014] A curve reconstruction training module is used to obtain sample logging data after spatiotemporal feature extraction according to the spatiotemporal feature extraction task of logging data, and control the logging curve reconstruction model to perform the logging curve reconstruction training task; wherein the sample logging data after spatiotemporal feature extraction is the target sample logging data required for the training of the logging curve reconstruction model; the logging curve reconstruction training task is used to train the curve reconstruction model to reconstruct the logging curve;

[0015] The parameter adjustment module is used to adjust the parameters of the well logging curve reconstruction model according to the well logging data spatiotemporal feature extraction task and the well logging curve reconstruction training task to obtain a converged well logging curve reconstruction model.

[0016] In a fourth aspect, an embodiment of the present invention further provides a curve reconstruction device, wherein the curve reconstruction model is obtained by using the training method of the curve reconstruction model described in any embodiment of the present invention, and the curve reconstruction device comprises:

[0017] A well logging data determination module, used to determine the well logging data of the area to be explored;

[0018] The logging curve reconstruction module is used to input the logging data into the curve reconstruction model to obtain the acoustic wave and density reconstructed logging curve of the area to be explored.

[0019] In a fifth aspect, an embodiment of the present invention further provides an electronic device, the electronic device comprising:

[0020] at least one processor; and

[0021] a memory communicatively connected to the at least one processor; wherein,

[0022] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the one or more processors implement the curve reconstruction model training and curve reconstruction method described in any embodiment of the present invention.

[0023] In a sixth aspect, an embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the curve reconstruction model training and curve reconstruction method described in any embodiment of the present invention.

[0024] The embodiment of the present invention provides a curve reconstruction model training and curve reconstruction method, device, electronic device and storage medium, which performs the task of extracting the spatiotemporal features of logging data according to the target sample logging data required for the logging curve reconstruction model training; obtains the sample logging data after spatiotemporal feature extraction according to the logging data spatiotemporal feature extraction task, and controls the logging curve reconstruction model to perform the logging curve reconstruction training task; wherein the sample logging data after spatiotemporal feature extraction is the target sample logging data required for the logging curve reconstruction model training; the logging curve reconstruction training task is used to train the curve reconstruction model to reconstruct the logging curve; according to the logging data spatiotemporal feature extraction task and the logging curve reconstruction training task, the logging curve reconstruction model is parameter adjusted to obtain a converged logging curve reconstruction model. The technical solution of the embodiment of the present invention is adopted to integrate the attention mechanism into the two-dimensional convolutional neural network, so as to enhance the ability of the deep learning network to capture the autocorrelation and cross-correlation feature information of the logging curve, so as to improve the accuracy and generalization ability of the deep learning network to reconstruct the acoustic wave and density logging curves. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Other features, objects and advantages of the present invention will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings. The drawings are only for the purpose of illustrating preferred embodiments and are not to be considered as limiting the present invention. Also, the same reference symbols are used throughout the drawings to represent the same parts. In the drawings:

[0026] Figure 1 is a flow chart of a curve reconstruction model training method provided in an embodiment of the present invention;

[0027] Figure 2 It is a schematic diagram of curve reconstruction model training and reconstructed curve prediction provided in an embodiment of the present invention;

[0028] Figure 3 is a schematic diagram of spatial feature weight distribution of well logging data provided in an embodiment of the present invention;

[0029] Figure 4 is a schematic diagram of the time series feature weight distribution of well logging data provided in an embodiment of the present invention;

[0030] Figure 5 is a schematic diagram of a fusion network structure based on an attention mechanism provided in an embodiment of the present invention;

[0031] Figure 6 It is a schematic diagram of the intersection of acoustic wave, density and sample logging data provided in an embodiment of the present invention;

[0032] Figure 7 is a schematic diagram of autocorrelation coefficient of well logging data provided in an embodiment of the present invention;

[0033] Figure 8 is a flow chart of a curve reconstruction method provided in an embodiment of the present invention;

[0034] Fig. 9 It is a schematic diagram comparing prediction results of acoustic waves and density under different networks provided in an embodiment of the present invention;

[0035] Fig.10 It is a schematic diagram of comparison between an acoustic wave and a density expansion curve before and after correction provided in an embodiment of the present invention;

[0036] Fig.11 is a structural schematic diagram of a curve reconstruction model training device provided in an embodiment of the present invention;

[0037] Fig.12 is a structural schematic diagram of a curve reconstruction device provided in an embodiment of the present invention;

[0038] Fig.13 It is a structural schematic diagram of an electronic device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0039] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0040] Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flow charts. Although the flow charts describe the operations (or steps) as sequential processes, many of the operations (or steps) therein can be implemented in parallel, concurrently or simultaneously. In addition, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but can also have additional steps not included in the accompanying drawings. The process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0041] Among them, the acquisition, storage, use and processing of data in the technical solution of this application are in compliance with the relevant provisions of national laws and regulations.

[0042] Figure 11 is a flow chart of a curve reconstruction model training method provided in an embodiment of the present invention. This embodiment is applicable to the case of training a well logging curve reconstruction model. The method of this embodiment can be executed by a curve reconstruction model training device, which can be implemented in hardware and / or software. The device can be configured in a curve reconstruction model training server. The method specifically includes the following steps:

[0043] S110, reconstructing target sample logging data required for model training based on logging curves, and performing a task of extracting spatiotemporal features of logging data.

[0044] Among them, the task of extracting spatiotemporal features of logging data is used to extract spatiotemporal features of logging data involved in logging curve reconstruction during the training process of the logging curve reconstruction model.

[0045] The spatiotemporal feature extraction may refer to obtaining the weight values ​​of the logging data in the temporal attention and the spatial attention. The logging data includes but is not limited to natural gamma (GR), spontaneous potential (SP), neutron porosity (CNL) and formation resistivity (RT).

[0046] In the model training stage of the well logging curve reconstruction model, initial sample well logging data corresponding to the well logging curve before reconstruction may be preprocessed to obtain sample well logging data for training the well logging curve reconstruction model.

[0047] As an optional but non-limiting implementation, before performing the task of extracting spatiotemporal features of logging data based on the target sample logging data required for the logging curve reconstruction model training, the curve reconstruction model training method further includes but is not limited to steps A1-A2:

[0048] Step A1: Obtain initial sample logging data for well logging curve reconstruction model training.

[0049] Step A1: preprocessing the initial sample logging data to obtain target sample logging data required for executing the logging curve reconstruction model training task;

[0050] The initial logging data is obtained, and the initial data is used to establish initial sample logging data and label logging data through quality control data; and the initial sample logging data is standardized and normalized, and divided into a training set and a test set. Figure 2 As shown, the training set is input into the network model for curve reconstruction model training; the test set is input into the optimal network model to reconstruct the acoustic wave and density curve. The initial logging data includes but is not limited to logging data, drilling data and seismic data, and the preprocessing includes but is not limited to standardization and normalization of the logging data.

[0051] Optionally, in the embodiment of the present invention, StandardScaler and MinMaxScaler functions may be used to perform standardization and normalization processing on the initial sample logging data to eliminate the difference between the dimensions of different initial sample logging data. The standardization and normalization may be expressed as:

[0052]

[0053]

[0054] Among them, X i is the initial sample logging data, X m is the mean of the initial sample logging data, X σ is the variance of the initial sample logging data, Y i is the standardized logging data, Y max and Y min are the maximum and minimum values ​​of the standardized data, Y i ′ is the normalized logging data.

[0055] As an optional but non-limiting implementation, the sample logging data required for the logging curve reconstruction model training and the task of extracting spatiotemporal features of the logging data are performed, including but not limited to steps B1-B2:

[0056] Step B1: according to the first sample test data in the target sample logging data required for executing the logging curve reconstruction model training task, perform the spatial attention feature extraction task on the first sample logging data.

[0057] Step B2: Based on the second sample test data in the target sample logging data required for executing the logging curve reconstruction model training task, the temporal attention feature extraction task is performed on the second sample logging data.

[0058] The well logging data spatiotemporal feature extraction task includes a spatial attention feature extraction task and a temporal attention feature extraction task, and the spatial attention feature extraction task and the temporal attention feature extraction task are used to improve the sensitivity of the well logging curve reconstruction model to the target spatiotemporal features. The well logging data spatiotemporal feature extraction task can integrate the temporal attention mechanism and the spatial attention mechanism into a unified spatiotemporal attention convolutional neural network model to identify sample well logging data and obtain distinctive target sample well logging data from a large amount of sample well logging data.

[0059] Optional, see Figure 3, perform the spatial attention feature extraction task on the first sample logging data, and determine the weight distribution of the first sample logging data in the spatial attention layer. When the label is sound wave, the weight distribution of the spatial attention layer is CNL, RT, GR, SP from large to small; when the label is density, the weight distribution of the spatial attention layer is GR, CNL, RT, SP from large to small; but the weight distribution of the network without the attention mechanism does not have this rule.

[0060] See also Figure 4 , perform the temporal attention feature extraction task on the second sample logging data, and determine the weight distribution of the second sample logging data in the temporal attention layer. Select 15 sampling points in the conventional logging data as the second sample logging data, and use the acoustic wave (AC) or density (DEN) located in the middle of the sample length as the label. When predicting AC or DEN, the second sample logging data located at the label position is given the largest weight ratio by the temporal attention layer. As the distance on both sides of the label increases, the overall trend of the attention weight decreases.

[0061] S120, obtaining sample logging data after spatiotemporal feature extraction according to the logging data spatiotemporal feature extraction task, and controlling the logging curve reconstruction model to execute the logging curve reconstruction training task.

[0062] The sample logging data after spatiotemporal feature extraction is the target sample logging data required for the well logging curve reconstruction model training; the well logging curve reconstruction training task is used to train the curve reconstruction model's ability to reconstruct the well logging curve.

[0063] As an optional but non-limiting implementation, see Figure 5 The logging curve reconstruction model includes an input layer, two two-dimensional convolutional layers, a spatial attention layer, a temporal attention layer and a fully connected layer; wherein the two-dimensional convolutional layer is used to extract the spatiotemporal characteristics of the logging data, the spatial attention layer and the temporal attention layer are used to improve the sensitivity of the logging curve reconstruction model to the target spatiotemporal characteristics, and the fully connected layer is used to improve the nonlinear ability of the logging curve reconstruction model.

[0064] When analyzing the correlation between the reconstructed logging curve and the sample logging curve, the target sample logging data is divided into two categories under the constraint of the wellbore curve, namely, the logging data of the non-expanded section and the logging data of the expanded section. Figure 6The cross-plot of the reconstruction curve of the non-expanded section and the sample logging data curve analyzes the cross-correlation between the reconstruction curve and the sample logging curve, and the reconstruction curve includes the acoustic reconstruction curve and the density reconstruction curve. The correlations between AC and the target sample logging data are as follows: neutron porosity (CNL), resistivity (RT), gamma (GR), and natural potential (SP). The coefficients of determination of DEN are as follows: gamma (GR), neutron porosity (CNL), resistivity (RT), and natural potential (SP). See Figure 7 , the autocorrelation function (ACF) is used to analyze the autocorrelation between the reconstructed curve and the sample logging data. The autocorrelation decreases with the increase of the lag distance.

[0065] The existing deep learning network model has the ability to establish a complex nonlinear mapping relationship between input and output, but does not emphasize the extraction of the autocorrelation and cross-correlation features of the logging curve, which reduces the reconstruction accuracy and generalization ability of the deep learning network. The present invention integrates the attention mechanism into the two-dimensional convolutional neural network to enhance the ability of the deep learning network to capture the autocorrelation and cross-correlation feature information of the logging curve, thereby improving the accuracy of the deep learning network in reconstructing the acoustic wave and density logging curves.

[0066] S130, adjusting parameters of the well logging curve reconstruction model according to the well logging data spatiotemporal feature extraction task and the well logging curve reconstruction training task to obtain a converged well logging curve reconstruction model.

[0067] The parameters of the logging curve reconstruction model are adjusted to obtain a converged logging curve reconstruction model. In the embodiment of the present invention, mean square error is used as a loss function and the Adam optimization algorithm is used to perform network optimization until the loss error converges.

[0068] As an optional but non-limiting implementation, the well logging curve reconstruction model is parameter adjusted according to the well logging data spatiotemporal feature extraction task and the well logging curve reconstruction training task to obtain a converged well logging curve reconstruction model, including but not limited to steps C1-C3:

[0069] Step C1: Determine a first loss function value corresponding to the well logging data spatiotemporal feature extraction training task and a second loss function value corresponding to the well logging curve reconstruction training task.

[0070] Step C2: The first loss function value and the second loss function value are fused according to the loss weights of their respective training tasks.

[0071] Step C3: adjusting parameters of the well logging curve reconstruction model according to the loss function value obtained by the loss fusion until the well logging curve reconstruction model converges.

[0072] By adopting the above method, the loss function score of the logging curve reconstruction training task can be fused with the loss function scores of each auxiliary training task at the beginning, and the total loss function score after fusion can be used to train the logging curve reconstruction model as a whole. Since it is fused at the beginning, it can deeply integrate the sample logging data obtained due to the deviation in the extraction of spatiotemporal features of the logging data and the deviation in the sample logging data on the reconstruction of the logging curve, thereby minimizing the problem of low model generalization caused by weak diversity of field data.

[0073] In an optional solution of the embodiment of the present invention, the coefficient of determination (R 2 ) quantitatively evaluates the well logging curve reconstruction model, and the evaluation calculation method is:

[0074]

[0075] Among them, y i represents the true label, represents the mean of the true values, represents the predicted value of the network, and n represents the number of samples i. Among them, the coefficient of determination (R 2 ) value is larger, the better the prediction ability of the logging curve reconstruction model is.

[0076] The embodiment of the present invention provides a curve reconstruction model training method, which performs a well logging data spatiotemporal feature extraction task based on target sample well logging data required for well logging curve reconstruction model training; obtains sample well logging data after spatiotemporal feature extraction based on the well logging data spatiotemporal feature extraction task, and controls the well logging curve reconstruction model to perform the well logging curve reconstruction training task; wherein the sample well logging data after spatiotemporal feature extraction is the target sample well logging data required for well logging curve reconstruction model training; the well logging curve reconstruction training task is used to train the curve reconstruction model's ability to reconstruct well logging curves; according to the well logging data spatiotemporal feature extraction task and the well logging curve reconstruction training task, the well logging curve reconstruction model is parameter adjusted to obtain a converged well logging curve reconstruction model. The technical solution of the embodiment of the present invention is adopted to integrate the attention mechanism into the two-dimensional convolutional neural network, so as to enhance the ability of the deep learning network to capture the autocorrelation and cross-correlation feature information of the well logging curve, so as to improve the accuracy and generalization ability of the deep learning network to reconstruct the acoustic wave and density well logging curves.

[0077] Figure 8 : is a flow chart of a curve reconstruction method provided in an embodiment of the present invention. This embodiment is applicable to the case of reconstructing a well logging curve. The method of this embodiment can be executed by a curve reconstruction device, which can be implemented in hardware and / or software. The device can be configured in a curve reconstruction server. The method specifically includes the following steps:

[0078] S810: Determine the logging data of the area to be explored.

[0079] The logging data include but are not limited to natural gamma, natural potential, resistivity and neutron porosity.

[0080] S820, inputting the logging data into the curve reconstruction model to obtain the acoustic wave and density reconstruction logging curve of the area to be explored.

[0081] Among them, the present invention analyzes the prediction results of spatiotemporal attention convolutional neural network (STACNN), two-dimensional convolutional neural network, gated recurrent unit and multivariate fitting respectively. The network structure is shown in Table 1. All networks use Adam with momentum as the optimization algorithm, which combines the advantages of AdaGrad and RMSProp algorithms and has strong advantages in processing large-scale data, parameter optimization and non-steady-state targets. At the same time, a Dropout layer is added to randomly discard neurons to increase the generalization of the network.

[0082] Table 1: STACNN, 2D convolutional neural network and gated recurrent unit network structure

[0083]

[0084] Among them, Conv = 2D convolution layer + ReLU activation function; SATT = spatial attention layer; TATT = temporal attention layer; Dropout = 0.25.

[0085] In an optional solution of the embodiment of the present invention, the present invention compares the performance effects of different deep learning network reconstructions. Fig. 9 , the well logging data of gd-13 well was selected to train the two-dimensional convolutional neural network, gated recurrent unit and STACNN network. Overall, the prediction results of AC and DEN of STACNN network were better than those of two-dimensional convolutional neural network and gated recurrent unit network, and compared with the prediction results of multivariate fitting, the prediction effects of two-dimensional convolutional neural network, gated recurrent unit and STACNN network were greatly improved (see Fig. 9 This shows that the deep learning network has obvious advantages in capturing the complex nonlinear relationship between the reconstructed curve and other curves.

[0086] In another optional scheme of the embodiment of the present invention, the present invention conducts a well diameter expansion test to verify the reconstruction effect of the deep learning network, and conducts an experiment on the non-expanded section of the gd-2 well to train and correct the error curve of the expanded section. The well diameter curve is selected as a constraint condition to divide the sample set into expanded and non-expanded parts. The network mentioned in the article is trained with the sample set of the non-expanded part, and the AC and DEN of all samples are predicted respectively. Fig.10 The middle borehole diameter curve shows that the borehole quality in this section is poor, with a large section of expansion (see Fig.10 In addition, there are also some small-scale expansion phenomena.

[0087] The values ​​of AC and DEN logging curves before and after correction are slightly different in a small range of expanded well sections, but the values ​​are quite different in a large range of expanded well sections (see Fig.10 According to the sedimentary evolution law of the strata, the physical properties of the same lithological segment with the same sedimentary background are basically similar. The curve shape changes more gently after correction, while the sound wave before correction has obvious periodic jump phenomenon due to the influence of well diameter expansion. In addition, compared with the synthetic record before expansion correction, the synthetic record after expansion correction is more consistent with the real seismic record, especially in Fig.10 At the 3640ms position, the relative relationship of the amplitude of the corrected synthetic record is basically consistent with the real seismic record. When there are errors in both the AC and DEN logging curves, the reconstructed curve obtained by the logging curve reconstruction model determined by adding the attention mechanism proposed in the embodiment of the present invention is more obvious.

[0088] Therefore, the analysis results of the above embodiments show that conventional logging curve reconstruction methods are difficult to capture the strong nonlinear relationship between logging curves. The present invention proposes a CNN fusion network based on the spatiotemporal attention mechanism to reconstruct the acoustic wave and density logging curves, and attempts to use geophysical expertise to explain the "black box" of the deep learning network, that is, the weight analysis of the deep learning network shows that the weight distribution of the attention layer of the deep learning network is consistent with the autocorrelation and cross-correlation characteristics of the logging curve, and verifies that adding an attention layer to the deep learning network can improve the network's sensitivity to important spatiotemporal features. The mutual verification effect of well seismic data shows that the CNN fusion network based on the spatiotemporal attention mechanism proposed in the embodiment of the present invention is better than multivariate linear fitting, gated recurrent unit and two-dimensional convolutional neural network in the correction and missing reconstruction of acoustic wave and density logging curves.

[0089] Fig.11 : is a structural schematic diagram of a curve reconstruction model training device provided in an embodiment of the present invention. The technical solution of this embodiment can be applied to the case of training the logging curve reconstruction model to improve the logging curve reconstruction accuracy. The device can be implemented by software and / or hardware, and is generally integrated on any electronic device with network communication function, including but not limited to: servers, computers, personal digital assistants and other devices. Fig.11 As shown, the curve reconstruction model training device provided in this embodiment may include the following: a spatiotemporal feature extraction module 1110, a curve reconstruction training module 1120 and a parameter adjustment module 1130, wherein:

[0090] The spatiotemporal feature extraction module 1110 is used to perform the spatiotemporal feature extraction task of the logging data according to the target sample logging data required for the logging curve reconstruction model training;

[0091] The curve reconstruction training module 1120 is used to obtain sample logging data after spatiotemporal feature extraction according to the spatiotemporal feature extraction task of the logging data, and control the logging curve reconstruction model to perform the logging curve reconstruction training task; wherein the sample logging data after spatiotemporal feature extraction is the target sample logging data required for the training of the logging curve reconstruction model; the logging curve reconstruction training task is used to train the curve reconstruction model to reconstruct the logging curve;

[0092] The parameter adjustment module 1130 is used to adjust the parameters of the well logging curve reconstruction model according to the well logging data spatiotemporal feature extraction task and the well logging curve reconstruction training task to obtain a converged well logging curve reconstruction model.

[0093] Based on the above embodiment, optionally, the spatiotemporal feature extraction module includes:

[0094] Acquire initial sample logging data for training the logging curve reconstruction model;

[0095] Preprocessing the initial sample logging data to obtain target sample logging data required for performing the logging curve reconstruction model training task;

[0096] The preprocessing includes standardization and normalization of the logging data.

[0097] Based on the above embodiment, optionally, the spatiotemporal feature extraction module further includes:

[0098] According to first sample test data in the target sample logging data required for performing the logging curve reconstruction model training task, performing a spatial attention feature extraction task on the first sample logging data;

[0099] According to the second sample test data in the target sample logging data required for executing the logging curve reconstruction model training task, the temporal attention feature extraction task is performed on the second sample logging data;

[0100] Among them, the spatial attention feature extraction task and the temporal attention feature extraction task are used to improve the sensitivity of the logging curve reconstruction model to the target spatiotemporal characteristics.

[0101] Based on the above embodiment, optionally, the curve reconstruction training module includes:

[0102] The logging curve reconstruction model includes an input layer, two two-dimensional convolutional layers, a spatial attention layer, a temporal attention layer and a fully connected layer; the two-dimensional convolutional layer is used to extract the spatiotemporal characteristics of the logging data, the spatial attention layer and the temporal attention layer are used to improve the sensitivity of the logging curve reconstruction model to the target spatiotemporal characteristics, and the fully connected layer is used to improve the nonlinear ability of the logging curve reconstruction model. Based on the above embodiment, optionally, the parameter adjustment module includes:

[0103] Determine a first loss function value corresponding to the well logging data spatiotemporal feature extraction training task and a second loss function value corresponding to the well logging curve reconstruction training task;

[0104] The first loss function value and the second loss function value are subjected to loss fusion according to the loss weights of the respective training tasks;

[0105] According to the loss function value obtained by the loss fusion, the parameters of the well logging curve reconstruction model are adjusted until the well logging curve reconstruction model converges.

[0106] The curve reconstruction model training device provided in the embodiment of the present invention can execute the curve reconstruction model training method provided in any embodiment of the present invention mentioned above, and has the corresponding functions and beneficial effects of executing the curve reconstruction model training method. For detailed process, please refer to the relevant operations of the curve reconstruction model training method in the above embodiment.

[0107] Fig.12 : is a schematic diagram of the structure of a curve reconstruction device provided in an embodiment of the present invention. The technical solution of this embodiment can be applied to the use of logging curve reconstruction models. The device can be implemented by software and / or hardware and is generally integrated on any electronic device with network communication function, including but not limited to: servers, computers, personal digital assistants and other devices. Fig.12 As shown, the curve reconstruction device provided in this embodiment may include the following: a logging data determination module 1210 and a logging curve reconstruction module 1220; wherein,

[0108] The well logging data determination module 1210 is used to determine the well logging data of the area to be explored;

[0109] The logging curve reconstruction module 1220 is used to input the logging data into the curve reconstruction model to obtain the acoustic wave and density reconstructed logging curve of the area to be explored.

[0110] The curve reconstruction device provided in the embodiment of the present invention can execute the curve reconstruction method provided in any embodiment of the present invention, and has the corresponding functions and beneficial effects of executing the curve reconstruction method. For detailed process, please refer to the relevant operations of the curve reconstruction method in the above embodiment.

[0111] Fig.13 1 is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. The electronic device 10 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. 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 required herein.

[0112] like Fig.13 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0113] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0114] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as curve reconstruction model training and curve reconstruction method.

[0115] In some embodiments, the curve reconstruction model training and curve reconstruction method can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the curve reconstruction model training and curve reconstruction method described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to perform the curve reconstruction model training and curve reconstruction method in any other appropriate manner (for example, by means of firmware).

[0116] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0117] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0118] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0119] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0120] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0121] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0122] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0123] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A curve reconstruction model training method, characterized in that: The method comprises: Reconstruct the target sample logging data required for model training based on the logging curves and perform the task of extracting spatiotemporal features of the logging data; According to the spatiotemporal feature extraction task of the logging data, sample logging data after spatiotemporal feature extraction is obtained, and the logging curve reconstruction model is controlled to perform the logging curve reconstruction training task; wherein the sample logging data after spatiotemporal feature extraction is the target sample logging data required for the training of the logging curve reconstruction model; the logging curve reconstruction training task is used to train the curve reconstruction model to reconstruct the logging curve; According to the well logging data spatiotemporal feature extraction task and the well logging curve reconstruction training task, the well logging curve reconstruction model is parameter adjusted to obtain a converged well logging curve reconstruction model.

2. The method according to claim 1, characterized in that Before performing the task of extracting spatiotemporal features of logging data based on target sample logging data required for logging curve reconstruction model training, the curve reconstruction model training method further includes: Acquire initial sample logging data for training the logging curve reconstruction model; Preprocessing the initial sample logging data to obtain target sample logging data required for performing the logging curve reconstruction model training task; The preprocessing includes standardization and normalization of the logging data.

3. The method according to claim 2, characterized in that The target sample logging data required for the training of the model reconstruction based on the logging curves and the task of extracting the spatiotemporal features of the logging data are performed, including: According to first sample test data in the target sample logging data required for performing the logging curve reconstruction model training task, performing a spatial attention feature extraction task on the first sample logging data; According to the second sample test data in the target sample logging data required for executing the logging curve reconstruction model training task, the temporal attention feature extraction task is performed on the second sample logging data; Among them, the spatial attention feature extraction task and the temporal attention feature extraction task are used to improve the sensitivity of the logging curve reconstruction model to the target spatiotemporal characteristics.

4. The method according to claim 3, characterized in that The logging curve reconstruction model includes an input layer, two two-dimensional convolutional layers, a spatial attention layer, a temporal attention layer and a fully connected layer; wherein the two-dimensional convolutional layer is used to extract the spatiotemporal characteristics of the logging data, the spatial attention layer and the temporal attention layer are used to improve the sensitivity of the logging curve reconstruction model to the target spatiotemporal characteristics, and the fully connected layer is used to improve the nonlinear ability of the logging curve reconstruction model.

5. The method according to claim 1, characterized in that The step of adjusting parameters of the well logging curve reconstruction model according to the well logging data spatiotemporal feature extraction task and the well logging curve reconstruction training task to obtain a converged well logging curve reconstruction model includes: Determine a first loss function value corresponding to the well logging data spatiotemporal feature extraction training task and a second loss function value corresponding to the well logging curve reconstruction training task; The first loss function value and the second loss function value are subjected to loss fusion according to the loss weights of the respective training tasks; According to the loss function value obtained by the loss fusion, the parameters of the well logging curve reconstruction model are adjusted until the well logging curve reconstruction model converges.

6. A curve reconstruction method, characterized in that: A curve reconstruction model obtained by using the curve reconstruction model training method described in any one of claims 1 to 5, wherein the curve reconstruction method comprises: Determine the well logging data for the area to be explored; The logging data is input into the curve reconstruction model to obtain the acoustic wave and density reconstruction logging curve of the area to be explored.

7. A curve reconstruction model training device, characterized in that: The curve reconstruction model training device comprises: The spatiotemporal feature extraction module is used to reconstruct the target sample logging data required for model training based on the logging curves and perform the spatiotemporal feature extraction task of the logging data; A curve reconstruction training module is used to obtain sample logging data after spatiotemporal feature extraction according to the spatiotemporal feature extraction task of logging data, and control the logging curve reconstruction model to perform the logging curve reconstruction training task; wherein the sample logging data after spatiotemporal feature extraction is the target sample logging data required for the training of the logging curve reconstruction model; the logging curve reconstruction training task is used to train the curve reconstruction model to reconstruct the logging curve; The parameter adjustment module is used to adjust the parameters of the well logging curve reconstruction model according to the well logging data spatiotemporal feature extraction task and the well logging curve reconstruction training task to obtain a converged well logging curve reconstruction model.

8. A curve reconstruction device, characterized in that: The curve reconstruction device comprises: A well logging data determination module, used to determine the well logging data of the area to be explored; The logging curve reconstruction module is used to input the logging data into the curve reconstruction model to obtain the acoustic wave and density reconstructed logging curve of the area to be explored.

9. An electronic device, characterized in that: include: one or more processors; A storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the curve reconstruction model training method described in any one of claims 1 to 5 or the curve reconstruction method described in claim 6.

10. A storage medium containing computer executable instructions, characterized in that: The computer executable instructions are used to execute the curve reconstruction model training method described in any one of claims 1 to 5 or the curve reconstruction method described in claim 6 when executed by a computer processor.

Images