Q-value estimation and earthquake attenuation compensation method based on fusion network

Abstract

The present application relates to and discloses a Q value estimation and seismic attenuation compensation method based on a fusion network, the method comprising: obtaining seismic data and well logging data in a study area, and processing the seismic data and well logging data to obtain low-frequency Q value data and attenuated seismic data; inputting the low-frequency Q value data and attenuated seismic data into a Q value estimation model to obtain a Q value estimation result; inputting the attenuated seismic data and the Q value estimation result into an attenuation compensation model to obtain an attenuation compensation result. The present application effectively avoids the cumulative error of the traditional two-step method, and while realizing high-resolution processing of seismic data, provides reliable Q value information, meeting the requirements of high-precision seismic exploration and fine reservoir characterization.

Description

Technical Field

[0001] The present application relates to the technical field of seismic exploration, and specifically refers to a Q value estimation and seismic attenuation compensation method based on a fusion network. Background Art

[0002] With the deepening of exploration and development, the geological conditions of exploration and development are becoming more and more complex, the depth of exploration targets is getting deeper and deeper, and the requirements for exploration accuracy are getting higher and higher. High-resolution seismic data is the prerequisite for achieving high-precision reservoir parameter prediction. However, due to the non-perfect elasticity of the underground medium, the energy of the seismic wave is irreversibly converted into heat energy during the propagation process, causing the amplitude of the seismic wave to attenuate; in addition, due to the different energy absorption attenuation rates of different frequency components of the seismic wave, as the propagation distance of the seismic wave increases, the phase of the seismic wavelet is also distorted. Among them, the attenuation caused by the non-perfect elasticity of the underground medium is an inherent characteristic of the medium, which can be quantitatively described by the quality factor Q. In addition, the Q value itself is also closely related to parameters such as lithology, saturation, and porosity, and can be used for reservoir identification and hydrocarbon detection.

[0003] For the above reasons, the resolution of seismic records will be reduced, increasing the risk of subsequent seismic data interpretation and reservoir characterization and evaluation. Therefore, in order to improve the resolution of seismic records, it is necessary to absorb and compensate for the attenuation of seismic waves.

[0004] Seismic inverse Q filtering is a data processing technology that improves seismic resolution. It uses the inverse process of wave propagation to achieve seismic wave attenuation compensation and phase correction. If the Q value is known, the seismic data can be inversely Q-filtered to enhance the high-frequency energy of the seismic signal, thereby improving the vertical resolution of the seismic data. However, the inverse Q filtering method based on the wave propagation principle requires a huge time cost. Moreover, the inverse Q filtering method belongs to the category of inverse problem solving. The energy compensation of high-frequency components is extremely unstable, and there is often a problem of insufficient deep energy compensation. In addition, when performing inverse Q filtering, the existing inverse Q filtering algorithms all require a known quality factor Q as a prerequisite, and the accuracy of the Q value directly affects the effect of inverse Q filtering. The quality factor Q is also an important parameter of lithologic reservoirs and can be used for source rock indication and oil and gas reservoir distribution description. However, due to unclear physical mechanisms and seismic noise, the Q value solution process is unstable and the accuracy and resolution of the Q value calculation results are limited. Q value estimation is both a hot topic and a difficult point in research.

[0005] At present, Q-value estimation and seismic attenuation compensation are usually two independent processes. First, the Q-value is extracted using a Q-value estimation method (e.g., spectrum ratio method, rise time method, center frequency offset, etc.); then, the seismic attenuation compensation is achieved using an inverse Q-filter method. Since most traditional Q-value estimation methods rely on artificially set analysis time windows, there is a large uncertainty in the Q-value estimation results, and their calculation accuracy cannot meet the requirements of fine characterization of actual reservoirs. At the same time, it leads to error accumulation in the inverse Q-filter method, making the final seismic attenuation compensation results unable to meet the requirements of seismic structural interpretation and reservoir evaluation. Summary of the invention

[0006] Based on the above technical problems, the present application provides a Q value estimation and seismic attenuation compensation method based on a fusion network. The present application effectively avoids the cumulative error of the traditional two-step method, and while realizing high-resolution processing of seismic data, provides reliable Q value information, meeting the requirements of high-precision seismic exploration and fine reservoir characterization.

[0007] In order to solve the above technical problems, the technical solutions adopted in this application are as follows:

[0008] A Q value estimation and seismic attenuation compensation method based on a fusion network, comprising:

[0009] Obtain seismic data and well logging data in the study area, and process the seismic data and well logging data to obtain low-frequency Q value data and attenuation seismic data;

[0010] Inputting low-frequency Q value data and attenuated seismic data into a Q value estimation model to obtain a Q value estimation result;

[0011] The attenuation seismic data and the Q value estimation result are input into the attenuation compensation model to obtain the attenuation compensation result.

[0012] A seismic attenuation compensation and Q value synchronous estimation device based on a fusion network, comprising:

[0013] A data acquisition module is used to acquire seismic data and well logging data in the study area, and process the seismic data and well logging data to obtain low-frequency Q value data and attenuation seismic data;

[0014] A Q value estimation module, the Q value estimation module is used to input low-frequency Q value data and attenuated seismic data into a Q value estimation model to obtain a Q value estimation result;

[0015] The attenuation compensation module is used to input the attenuation seismic data and the Q value estimation result into the attenuation compensation model to obtain the attenuation compensation result.

[0016] A computer device includes a memory and a processor. The memory stores a computer program. When the computer program is executed by the processor, the processor executes the steps of the Q value estimation and seismic attenuation compensation method based on a fusion network.

[0017] A computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the processor executes the steps of the Q value estimation and seismic attenuation compensation method based on a fusion network.

[0018] Compared with the prior art, the beneficial effects of this application are:

[0019] The above method, device, computer equipment and storage medium

[0020] 1. This application uses deep learning methods to estimate the seismic Q value and attenuation compensation, avoiding the problems of low accuracy and poor stability under the assumptions of traditional physical models;

[0021] 2. This application uses random simulation to construct a training data set to ensure the robustness of deep learning network training;

[0022] 3. The coupling loss function of earthquake Q value estimation and attenuation compensation constructed in this application realizes earthquake Q value estimation and attenuation compensation simultaneously, avoiding the cumulative error caused by the two-step method. This scheme can effectively improve the accuracy of earthquake Q value estimation and attenuation compensation;

[0023] 4. This application introduces low-frequency Q-value data constructed based on geological background information during the network training and prediction stages, which can greatly improve the stability of seismic attenuation compensation and Q-value estimation, while significantly improving the generalization performance of the seismic attenuation compensation and Q-value estimation network.

[0024] 5. The earthquake Q value estimation model and attenuation compensation model constructed in this application can be directly used for earthquake Q value estimation and attenuation compensation after training, and have higher computational efficiency than traditional model-driven methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. Among them:

[0026] Figure 1 It is a flowchart of the Q value estimation and seismic attenuation compensation method based on the fusion network.

[0027] Figure 2 A flowchart of a training method for the Q-value estimation model and the attenuation compensation model is provided.

[0028] Figure 3Flow chart of the method for obtaining accurate Q-value training data.

[0029] Figure 4 Schematic diagram of the method flow for obtaining low-frequency Q-value training data.

[0030] Figure 5 Schematic diagram of the method flow for obtaining unattenuated earthquake training data.

[0031] Figure 6 Schematic diagram of the framework of Q-value estimation and seismic attenuation compensation method based on fusion network.

[0032] Figure 7 Some example images of training data.

[0033] Figure 8 Some example plots of noise-free low-frequency Q-value data and attenuated seismic data used as input to the model.

[0034] Fig. 9 Some example plots of noisy (SNR = 2) low-frequency Q-value data and attenuated seismic data used as model input.

[0035] Fig.10 This is the output result of the Q value estimation and seismic attenuation compensation method based on the fusion network in the noise-free condition.

[0036] Fig.11 This is the output result of the Q value estimation and seismic attenuation compensation method based on the fusion network under noisy conditions (signal-to-noise ratio is 2). DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solution and advantages of the embodiments of the present disclosure clearer, the technical solution of the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings of the embodiments of the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, not all of the embodiments. Based on the described embodiments of the present disclosure, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present disclosure.

[0038] It should be understood that the "system", "device", "unit" and / or "module" used in this specification is a method for distinguishing different components, elements, parts, parts or assemblies at different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.

[0039] As shown in this specification and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "comprise" and "include" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0040] Flowcharts are used in this specification to illustrate the operations performed by the system according to the embodiments of this specification. It should be understood that the preceding or following operations are not necessarily performed precisely in order. Instead, the steps may be processed in reverse order or simultaneously. At the same time, other operations may also be added to these processes, or one or more operations may be removed from these processes.

[0041] See also Figure 1 In some embodiments, a Q value estimation and seismic attenuation compensation method based on a fusion network includes:

[0042] S101, obtaining seismic data and well logging data in a study area, and processing the seismic data and well logging data to obtain low-frequency Q value data and attenuation seismic data;

[0043] Specifically, for the acquisition of low-frequency Q value data, the velocity model of each survey line in the logging data in the time domain can be obtained by interpolating and extrapolating the logging data according to the structural pattern; then the velocity model is substituted into the relationship function between the seismic Q value and the velocity to obtain the calculation result, and finally the calculation result is smoothed to obtain the low-frequency Q value data.

[0044] Specifically, for the acquisition of attenuated seismic data, generally during the seismic data acquisition stage, the attenuated seismic data can be directly acquired.

[0045] S102, inputting the low-frequency Q value data and the attenuated seismic data into a Q value estimation model to obtain a Q value estimation result;

[0046] S103, inputting the attenuation seismic data and the Q value estimation result into the attenuation compensation model to obtain the attenuation compensation result.

[0047] In this embodiment, combined with Figure 6 , taking low-frequency Q-value data and attenuation seismic data as input, and using the trained deep learning network model, the Q-value estimation and attenuation compensation of seismic data were finally achieved simultaneously, obtaining highly reliable Q-value estimation results and attenuation compensation results.

[0048] Among them, the present application adopts deep learning methods for earthquake Q value estimation and attenuation compensation. It no longer needs to rely on manually set analysis time windows like traditional Q value estimation methods, avoiding the problems of low accuracy and poor stability under the assumptions of traditional physical models.

[0049] Among them, this application adopts a deep learning method for seismic Q value estimation and attenuation compensation. Compared with the traditional inverse Q filtering method, it solves the problem that the traditional inverse Q filtering method belongs to the category of inverse problem solving, and the high-frequency component energy compensation is extremely unstable, resulting in insufficient deep energy compensation.

[0050] Among them, the seismic Q value estimation model and attenuation compensation model constructed in the application can be directly used for seismic Q value estimation and attenuation compensation after training. Compared with the traditional model-driven method and the inverse Q filtering method based on wave field propagation theory, it has higher computational efficiency and greatly reduces the computational time cost.

[0051] See also Figure 2 In some embodiments, the training method of the Q-value estimation model and the attenuation compensation model includes:

[0052] S201, obtaining training data, the training data including accurate Q value training data and unattenuated seismic training data, and corresponding low-frequency Q value training data and attenuated seismic training data;

[0053] S202, inputting the low-frequency Q-value training data and the attenuated earthquake training data into an untrained Q-value estimation model to obtain a first output result;

[0054] S203, inputting the attenuation seismic training data and the first output result into an untrained attenuation compensation model to obtain a second output result;

[0055] S204, determining a coupling loss function based on the precise Q-value training data, the unattenuated seismic training data, the first output result, and the second output result;

[0056] S205, performing joint iterative training on the Q-value estimation model and the attenuation compensation model based on the coupling loss function to obtain trained Q-value estimation model and attenuation compensation model.

[0057] In this embodiment, based on the characteristics between the precise Q-value training data and the attenuated earthquake training data, and the characteristics between the unattenuated earthquake training data and the attenuated earthquake training data, a suitable "end-to-end" deep learning network is established, and the Q-value estimation and attenuation compensation coupling loss function is further constructed, and the Q-value estimation network is connected with the attenuation compensation network, so as to construct a deep learning model that jointly implements Q-value estimation and attenuation compensation, thereby ensuring the effectiveness and stability of the earthquake Q-value estimation results and the attenuation compensation results.

[0058] Among them, since the traditional two-step seismic attenuation compensation method depends on the accuracy of the Q value, there is a large cumulative error in the process of implementing the inverse Q filter based on the estimated Q value. However, this application constructs a coupling loss function for seismic Q value estimation and attenuation compensation to simultaneously implement seismic Q value estimation and attenuation compensation, avoiding the cumulative error caused by the two-step method. This solution can effectively improve the accuracy of seismic Q value estimation and attenuation compensation;

[0059] Specifically, the constructed training data is input into the deep learning network model, the maximum number of iterations is controlled by coupling the loss function value, the network parameters are updated using optimization algorithms such as stochastic gradient, and the network training process is repeated to minimize the loss function, so as to obtain the optimal deep learning network parameters for the Q-value estimation model and the attenuation compensation model.

[0060] Specifically, for the structures of the Q-value estimation model and the attenuation compensation model, both are deep learning network models with the same structure. The deep learning network model specifically includes two convolutional layers and one fully connected layer. The size of the convolution kernel in the convolutional layer is 3*1, and the number of convolution kernels is 32. Before the fully connected layer, all feature vectors are expanded into one-dimensional vectors, and the number of neurons in the fully connected layer is 256.

[0061] Among them, a Dropout layer is added after the second convolutional layer of the deep learning network model of the Q-value estimation model and the attenuation compensation model to prevent network overfitting, and a Max-pooling layer is added to improve the network training efficiency.

[0062] Among them, ReLu is used as the activation function in the deep learning network model of the Q-value estimation model and the attenuation compensation model.

[0063] Specifically, the Adam optimizer is selected to optimize the network. During the network training process, the batch size and the number of iterations are 128 and 1000 respectively, and the learning rate is 0.00005.

[0064] Preferably, the coupling loss function is specifically:

[0065] L=‖QF([d i ,Q i ],θ1)‖ 2 +‖d t -G([d i ,F([d i ,Q i ],θ2)])‖ 2

[0066] Among them, L represents the coupling loss function, Q represents the accurate Q value training data, F([d i ,Q i ],θ1) represents the first output result of the Q-value estimation network, dt represents the unattenuated earthquake training data, G([d i ,F([d i ,Q i ],θ2)]) represents the second output result of the attenuation compensation network;

[0067] Among them, d i represents the attenuation earthquake training data, Q i represents the low-frequency Q-value training data, θ1 represents the Q-value estimation network optimization parameters, and θ2 represents the attenuation compensation network optimization parameters.

[0068] Specifically, i and t are used to distinguish different meanings of the same letter.

[0069] Among them, the low-frequency Q-value training data and the attenuated earthquake training data are taken as input, and are input into the Q-value estimation network and the attenuation compensation network at the same time. The output result of the Q-value estimation network is further input into the attenuation compensation network. By establishing a coupling loss function between the two, the stability of the network training process is guaranteed, and earthquake Q-value estimation and attenuation compensation are realized synchronously.

[0070] Since seismic Q-value estimation and attenuation compensation belong to regression problems, the root mean square error (MSE) is specifically selected as the loss function.

[0071] In some embodiments, the method for obtaining training data mainly includes obtaining four parts: accurate Q value training data, unattenuated seismic training data, low frequency Q value training data and attenuated seismic training data:

[0072] See also Figure 3 ,Specifically, the method of obtaining accurate Q value training data includes:

[0073] S301, collect seismic structural interpretation data and well logging data in the study area, the seismic structural interpretation data includes geological structure information;

[0074] S302, based on geological structure information and well logging data, using a random simulation algorithm to randomly generate accurate Q value training data.

[0075] See also Figure 4 , the method of obtaining low-frequency Q value training data includes:

[0076] S401, interpolating and extrapolating the logging data according to the structural model to obtain a velocity model in the time domain of each measuring line in the logging data;

[0077] S402, bringing the velocity model into the relationship function between the seismic Q value and the velocity to obtain initial Q value data, and performing smoothing processing on the initial Q value data to obtain low-frequency Q value training data.

[0078] Among them, the low-frequency Q value is the background of the precise Q value. In practical applications, the low-frequency Q value can be obtained through the relationship function between the seismic Q value and the velocity, and then any number of low-frequency Q values ​​can be obtained through smoothing processing.

[0079] By using low-frequency Q values ​​as input parameters in the model training and application process, the ambiguity of the inversion results can be reduced and the accuracy of the model output results can be improved.

[0080] Among them, the geological background information is ignored in the conventional inverse Q filtering and Q value estimation process, which increases the instability in the data processing process; in the method based on deep learning, the stability of the network prediction results and the generalization ability of the network can be improved by adding geological background information. The establishment of a low-frequency Q value model mainly uses the spatial interpolation method. Its technical process is to first complete the picking of geological layers, use the layer information as the control point and combine the logging data for extrapolation interpolation, that is, to perform horizontal interpolation of the logging information, calculate the parameter value at each point underground, and complete the task of initial parameter modeling. Since the logging data usually does not directly contain Q value information, the present invention is based on the initial velocity modeling results, and uses the empirical formula for conversion between Q value and velocity to complete the task of initial Q value modeling.

[0081] Specifically, the relationship function between earthquake Q value and velocity is:

[0082] Q=a*V p b

[0083] Where Q represents the Q value, V p represents speed, and a and b represent constants.

[0084] Among them, a and b will vary due to the different geological structures in the study area.

[0085] Among them, in practical applications, the velocity is very easy to obtain through existing methods. When the velocity is known, the low-frequency Q value can be obtained by using the empirical formula of seismic Q value and velocity.

[0086] In addition, low-frequency Q-value training data can be obtained by smoothing the precise Q-value training data.

[0087] See also Figure 5 , the method of obtaining unattenuated earthquake training data includes:

[0088] S501, extracting the first seismic wavelet from the seismic structural interpretation data and the well logging data using a statistical method;

[0089] S502, calculating and extracting the frequency spectrum information of the first seismic wavelet based on the Fourier transform method, and analyzing the frequency spectrum characteristics of the first seismic wavelet to obtain the main frequency characteristics of the first seismic wavelet;

[0090] S503, constructing a second seismic wavelet with a different main frequency according to the main frequency characteristics;

[0091] Among them, any number of second seismic wavelets can be constructed according to the main frequency characteristics of the seismic wavelet extracted from the study area. The corresponding unattenuated seismic training data and attenuated seismic training data generated subsequently will contain the data distribution characteristics of the study area, thereby improving the applicability of the trained Q-value estimation model and attenuation compensation model.

[0092] S504, inputting the second seismic wavelet and the velocity model into the convolution model to obtain unattenuated seismic training data.

[0093] Specifically, the method for obtaining attenuation earthquake training data includes:

[0094] Substituting the unattenuated earthquake training data and the accurate Q value training data into the attenuation theory model to obtain the attenuated training data;

[0095] The attenuation model is as follows:

[0096]

[0097] Among them, x(f,τ) represents the attenuated earthquake training data, x(f,0) represents the unattenuated earthquake training data, f represents the frequency of the earthquake wavelet, τ represents the total propagation time of the earthquake wavelet, Q eff represents the equivalent Q value of the exact Q value training data, f r represents the reference frequency of the seismic wavelet;

[0098] Among them, the total time τ is obtained according to the following formula:

[0099]

[0100] Among them, τ i It represents the time required for the seismic wavelet to propagate through the i-th layer in the model, and n represents the total number of layers.

[0101] Among them, the equivalent Q value Q is obtained according to the following formula eff :

[0102]

[0103] Among them, Q i represents the exact Q value of the i-th layer in the exact Q value training data.

[0104] In this embodiment, accurate Q-value training data is first randomly generated through geological structure information and logging data; a velocity model is then generated through the logging data, and unattenuated seismic training data is obtained using the velocity model; subsequently, the unattenuated seismic training data and the accurate Q-value training data are substituted into the attenuation theory model to obtain attenuated seismic training data.

[0105] From the process of obtaining the training set, it can be seen that since the precise Q-value training data and velocity model are generated based on the collected seismic structural interpretation data and logging data, the subsequent unattenuated seismic training data is obtained through the velocity model. Therefore, for the precise Q-value training data and unattenuated training data, it is preset as precise data during the entire model training process. In this way, the low-frequency Q-value training data and attenuated seismic training data are obtained through the precise Q-value training data and unattenuated training data.

[0106] Due to the unclear physical mechanism of Q-value estimation, existing Q-value estimation methods have problems such as unstable Q-value solution and limited accuracy and resolution of Q-value calculation results. However, under the condition that the precise Q-value training data and unattenuated training data are accurately preset, the present application inputs the acquired low-frequency Q-value training data and attenuated seismic training data into the Q-value estimation model and the attenuation compensation model, and learns the mapping relationship between the low-frequency Q-value training data and the attenuated seismic training data and the precise Q-value training data and the unattenuated training data through the model. After the model learns the rules through training, accurate Q-value estimation results and attenuation compensation results can be obtained by inputting low-frequency Q-value training data and attenuated seismic data.

[0107] In addition, deep learning network model training requires sufficient training data as support. This application uses the collected seismic structure interpretation data and logging data to construct any number of training sample databases through random simulation methods, thereby using the training data to train high-precision Q value estimation models and attenuation compensation models.

[0108] Based on the above embodiments, the Q value estimation and earthquake attenuation compensation method based on the fusion network are further described in combination with relevant data:

[0109] 1. See Figure 7 , inputting the training data into the untrained Q-value estimation model and the attenuation compensation model to obtain the trained Q-value estimation model and the attenuation compensation model;

[0110] in, Figure 7 (a) represents low-frequency Q-value training data, (b) represents attenuated earthquake training data, (c) represents accurate Q-value training data, and (d) represents attenuated earthquake training data;

[0111] 2. See Figure 8, the noise-free low-frequency Q-value data and attenuated seismic data are input into the Q-value estimation model and the attenuation compensation model, and the output results are shown in Fig.10 ;

[0112] in, Figure 8 (a) shows low-frequency Q-value data, and (b) shows attenuated seismic data;

[0113] in, Fig.10 (a) shows the attenuation compensation result, and (b) shows the Q value estimation result;

[0114] 3. See Fig. 9 , low-frequency Q-value data containing noise (signal-to-noise ratio is 2) and attenuated seismic data are input into the Q-value estimation model and attenuation compensation model, and the output results are shown in Fig.11 ;

[0115] in, Fig. 9 (a) shows low-frequency Q-value data, and (b) shows attenuated seismic data;

[0116] in, Fig.11 (a) shows the attenuation compensation result, and (b) shows the Q value estimation result.

[0117] In summary, it can be seen from the data in the figure that high-precision Q value estimation results and attenuation compensation results can be obtained through the Q value estimation and seismic attenuation compensation method based on the fusion network of the present application.

[0118] In some embodiments, a fusion network-based seismic attenuation compensation and Q-value synchronous estimation device is also disclosed, including:

[0119] A data acquisition module is used to acquire seismic data and well logging data in the study area, and process the seismic data and well logging data to obtain low-frequency Q value data and attenuation seismic data;

[0120] A Q value estimation module, the Q value estimation module is used to input low-frequency Q value data and attenuated seismic data into a Q value estimation model to obtain a Q value estimation result;

[0121] The attenuation compensation module is used to input the attenuation seismic data and the Q value estimation result into the attenuation compensation model to obtain the attenuation compensation result.

[0122] In order to solve the above technical problems, the present application also discloses a computer device, which is characterized in that it includes a memory and a processor, and a computer program is stored in the memory. When the computer program is executed by the processor, the processor executes the steps of the above-mentioned Q value estimation and seismic attenuation compensation method based on the fusion network.

[0123] The computer device may be a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with the user through a keyboard, a mouse, a remote control, a touch pad, or a voice control device.

[0124] The memory includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or D interface display memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory can be an internal storage unit of the computer device, such as a hard disk or memory of the computer device. In other embodiments, the memory can also be an external storage device of the computer device, such as a plug-in hard disk equipped on the computer device, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. Of course, the memory can also include both the internal storage unit of the computer device and its external storage device. In this embodiment, the memory is often used to store the operating system and various application software installed on the computer device, such as the program code of the Q value estimation and seismic attenuation compensation method based on the fusion network. In addition, the memory can also be used to temporarily store various data that have been output or are to be output.

[0125] In some embodiments, the processor may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip. The processor is generally used to control the overall operation of the computer device. In this embodiment, the processor is used to run the program code stored in the memory or process data, such as running the program code of the Q value estimation and seismic attenuation compensation method based on the fusion network.

[0126] In order to solve the above technical problems, the present application also discloses a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the processor executes the steps of the above-mentioned Q value estimation and seismic attenuation compensation method based on the fusion network.

[0127] The computer-readable storage medium stores an interface display program, and the interface display program can be executed by at least one processor so that the at least one processor executes the steps of the Q value estimation and seismic attenuation compensation method based on the fusion network as described above.

[0128] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server or network device, etc.) to execute the methods described in each embodiment of the present application.

[0129] The above is the embodiment of the present application. The above embodiments and the specific parameters in the embodiments are only for the purpose of clearly describing the verification process of the application, and are not intended to limit the scope of patent protection of the present application. The scope of patent protection of the present application is still subject to the claims, and any equivalent structural changes made by using the description and drawings of the present application should be included in the scope of protection of the present application.

Claims

1. A Q-value estimation and seismic attenuation compensation method based on a fusion network, characterized in that: include: Acquire seismic data and well logging data of the study area, and process the seismic data and the well logging data to obtain low-frequency Q value data and attenuation seismic data; Inputting the low-frequency Q value data and the attenuated seismic data into a Q value estimation model to obtain a Q value estimation result; Inputting the attenuation seismic data and the Q value estimation result into an attenuation compensation model to obtain an attenuation compensation result; The training method of the Q value estimation model and the attenuation compensation model includes: Acquire training data, wherein the training data includes accurate Q-value training data and unattenuated seismic training data, and corresponding low-frequency Q-value training data and attenuated seismic training data; Inputting the low-frequency Q-value training data and the attenuated earthquake training data into an untrained Q-value estimation model to obtain a first output result; Inputting the attenuated seismic training data and the first output result into an untrained attenuation compensation model to obtain a second output result; Determine a coupling loss function based on the precise Q-value training data, the unattenuated seismic training data, the first output result, and the second output result; The coupling loss function is specifically: in, represents the coupling loss function, represents the exact Q-value training data, represents the first output result of the Q-value estimation network, represents the unattenuated earthquake training data, represents the second output result of the attenuation compensation network; in, represents the attenuated earthquake training data, represents the low-frequency Q-value training data, represents the Q-value estimation network optimization parameters, represents the optimization parameters of the attenuation compensation network; Performing joint iterative training on the Q-value estimation model and the attenuation compensation model based on the coupling loss function to obtain the trained Q-value estimation model and the attenuation compensation model; The method for obtaining the attenuation earthquake training data includes: Substituting the unattenuated earthquake training data and the accurate Q value training data into the attenuation theory model to obtain the attenuation training data; The attenuation theoretical model is specifically: in, represents the attenuated earthquake training data, represents the unattenuated earthquake training data, represents the frequency of the seismic wavelet, represents the total propagation time of the seismic wavelet, represents the equivalent Q-value of the exact Q-value training data, represents the reference frequency of the seismic wavelet; Among them, the total time is obtained according to the following formula : in, Indicates that the seismic wavelet propagates through the model The time required for the layer, Indicates the total number of layers; Among them, the equivalent Q value is obtained according to the following formula : in, Indicates the exact Q value in the training data The exact Q value of the layer.

2. The Q value estimation and seismic attenuation compensation method based on fusion network according to claim 1 is characterized in that: The method for obtaining the accurate Q value training data includes: Collecting seismic structural interpretation data and well logging data in the study area, wherein the seismic structural interpretation data includes geological structural information; Based on the geological structure information and the well logging data, a random simulation algorithm is used to randomly generate accurate Q value training data.

3. The Q value estimation and seismic attenuation compensation method based on fusion network according to claim 2 is characterized in that: The method for obtaining the low-frequency Q value training data includes: Interpolating and extrapolating the logging data according to the structural model to obtain a velocity model in the time domain of each measuring line in the logging data; The velocity model is introduced into the relationship function between the seismic Q value and the velocity to obtain the initial Q value data, and the initial Q value data is smoothed to obtain the low-frequency Q value training data.

4. The Q value estimation and seismic attenuation compensation method based on fusion network according to claim 3 is characterized in that: The method for obtaining the unattenuated seismic training data includes: extracting a first seismic wavelet from the seismic structure interpretation data and the well logging data using a statistical method; Calculating and extracting the frequency spectrum information of the first seismic wavelet based on the Fourier transform method, and analyzing the frequency spectrum characteristics of the first seismic wavelet to obtain the main frequency characteristics of the first seismic wavelet; constructing a second seismic wavelet with a different main frequency according to the main frequency characteristics; The second seismic wavelet and the velocity model are input into a convolution model to obtain the unattenuated seismic training data.

5. A seismic attenuation compensation and Q value synchronous estimation device based on a fusion network, characterized in that: include: A data acquisition module, the data acquisition module is used to acquire seismic data and well logging data in the study area, and process the seismic data and the well logging data to obtain low-frequency Q value data and attenuation seismic data; A Q-value estimation module, the Q-value estimation module is used to input the low-frequency Q-value data and the attenuated seismic data into a Q-value estimation model to obtain a Q-value estimation result; An attenuation compensation module, the attenuation compensation module is used to input the attenuation seismic data and the Q value estimation result into an attenuation compensation model to obtain an attenuation compensation result; The training method of the Q value estimation model and the attenuation compensation model includes: Acquire training data, wherein the training data includes accurate Q-value training data and unattenuated seismic training data, and corresponding low-frequency Q-value training data and attenuated seismic training data; Inputting the low-frequency Q-value training data and the attenuated earthquake training data into an untrained Q-value estimation model to obtain a first output result; Inputting the attenuated seismic training data and the first output result into an untrained attenuation compensation model to obtain a second output result; Determine a coupling loss function based on the precise Q-value training data, the unattenuated seismic training data, the first output result, and the second output result; The coupling loss function is specifically: in, represents the coupling loss function, represents the exact Q-value training data, represents the first output result of the Q-value estimation network, represents the unattenuated earthquake training data, represents the second output result of the attenuation compensation network; in, represents the attenuated earthquake training data, represents the low-frequency Q-value training data, represents the Q-value estimation network optimization parameters, represents the optimization parameters of the attenuation compensation network; Performing joint iterative training on the Q-value estimation model and the attenuation compensation model based on the coupling loss function to obtain the trained Q-value estimation model and the attenuation compensation model; The method for obtaining the attenuation earthquake training data includes: Substituting the unattenuated earthquake training data and the accurate Q value training data into the attenuation theory model to obtain the attenuation training data; The attenuation theoretical model is specifically: in, represents the attenuated earthquake training data, represents the unattenuated earthquake training data, represents the frequency of the seismic wavelet, represents the total propagation time of the seismic wavelet, represents the equivalent Q-value of the exact Q-value training data, represents the reference frequency of the seismic wavelet; Among them, the total time is obtained according to the following formula : in, Indicates that the seismic wavelet propagates through the model The time required for the layer, Indicates the total number of layers; Among them, the equivalent Q value is obtained according to the following formula : in, Indicates the exact Q value in the training data The exact Q value of the layer.

6. A computer device, characterized in that: It comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the Q value estimation and seismic attenuation compensation method based on the fusion network as claimed in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that: A computer program is stored, and when the computer program is executed by a processor, the processor executes the steps of the Q value estimation and seismic attenuation compensation method based on a fusion network as claimed in any one of claims 1 to 4.

Images