CN115079257A - Q value estimation and seismic attenuation compensation method based on fusion network - Google Patents
Q value estimation and seismic attenuation compensation method based on fusion network Download PDFInfo
- Publication number
- CN115079257A CN115079257A CN202210756266.3A CN202210756266A CN115079257A CN 115079257 A CN115079257 A CN 115079257A CN 202210756266 A CN202210756266 A CN 202210756266A CN 115079257 A CN115079257 A CN 115079257A
- Authority
- CN
- China
- Prior art keywords
- seismic
- value
- data
- training data
- attenuation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 104
- 230000004927 fusion Effects 0.000 title claims abstract description 30
- 238000012545 processing Methods 0.000 claims abstract description 17
- 238000011160 research Methods 0.000 claims abstract description 13
- 238000012549 training Methods 0.000 claims description 126
- 230000002238 attenuated effect Effects 0.000 claims description 35
- 230000006870 function Effects 0.000 claims description 21
- 230000008878 coupling Effects 0.000 claims description 13
- 238000010168 coupling process Methods 0.000 claims description 13
- 238000005859 coupling reaction Methods 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 8
- 238000009499 grossing Methods 0.000 claims description 5
- 238000005457 optimization Methods 0.000 claims description 5
- 238000001228 spectrum Methods 0.000 claims description 5
- 238000004422 calculation algorithm Methods 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 3
- 238000005562 fading Methods 0.000 claims description 3
- 238000007619 statistical method Methods 0.000 claims description 2
- 230000001360 synchronised effect Effects 0.000 claims description 2
- 238000012512 characterization method Methods 0.000 abstract description 4
- 230000008569 process Effects 0.000 description 17
- 238000001914 filtration Methods 0.000 description 14
- 238000013135 deep learning Methods 0.000 description 13
- 238000004364 calculation method Methods 0.000 description 8
- 239000004215 Carbon black (E152) Substances 0.000 description 3
- 229930195733 hydrocarbon Natural products 0.000 description 3
- 150000002430 hydrocarbons Chemical class 0.000 description 3
- 230000036961 partial effect Effects 0.000 description 3
- 238000010521 absorption reaction Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000002829 reductive effect Effects 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000000712 assembly Effects 0.000 description 1
- 238000000429 assembly Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000013213 extrapolation Methods 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000000670 limiting effect Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 239000011435 rock Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
- G01V1/28—Processing seismic data, e.g. for interpretation or for event detection
- G01V1/282—Application of seismic models, synthetic seismograms
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
- G01V1/40—Seismology; Seismic or acoustic prospecting or detecting specially adapted for well-logging
Landscapes
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Remote Sensing (AREA)
- Acoustics & Sound (AREA)
- Environmental & Geological Engineering (AREA)
- Geology (AREA)
- General Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Geophysics (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The application relates to and discloses a Q value estimation and seismic attenuation compensation method based on a fusion network, which comprises the following steps: acquiring seismic data and logging data of a research area, and processing the seismic data and the logging data to obtain low-frequency Q value data and attenuation seismic data; inputting the low-frequency Q value data and the attenuation seismic data into a Q value estimation model to obtain a Q value estimation result; and inputting the attenuation seismic data and the Q value estimation result into an attenuation compensation model to obtain an attenuation compensation result. The method and the device effectively avoid the accumulated error of the traditional two-step method, provide reliable Q value information while realizing high-resolution processing of seismic data, and meet the requirements of high-precision seismic exploration and fine reservoir characterization.
Description
Technical Field
The application relates to the technical field of seismic exploration, in particular to a Q value estimation and seismic attenuation compensation method based on a fusion network.
Background
Along with the depth of the exploration and development degree, the geological conditions of the exploration and development are more and more complex, the depth of an exploration target is deeper and deeper, the requirement on the exploration precision is higher and higher, and the high-resolution seismic data is the premise of realizing the high-precision reservoir parameter prediction. However, due to the incomplete elastic property of the underground medium, the energy of the seismic waves can be irreversibly converted into heat energy in the propagation process, so that the amplitude of the seismic waves is attenuated; in addition, due to the difference of the energy absorption attenuation rate of different frequency components of the seismic wave, the phase of the seismic wavelet is also distorted along with the increase of the propagation distance of the seismic wave. Where attenuation due to imperfect elasticity of the subsurface medium is an inherent property of the medium, which can be quantitatively described in terms of quality factor Q. In addition, the Q value is closely related to lithology, saturation, porosity and other parameters, and can be used for reservoir identification and hydrocarbon detection.
Based on the reasons, the resolution of the seismic record is reduced, and the subsequent seismic data interpretation and reservoir characterization evaluation risks are increased. Therefore, in order to improve the resolution of seismic recordings, it is necessary to perform absorption compensation processing on the attenuation of seismic waves.
The seismic inverse Q filtering is a data processing technology for improving seismic resolution, and realizes seismic wave attenuation compensation and phase correction by using the inverse process of wave propagation. If the Q value is known, the seismic data is subjected to inverse Q filtering, so that the high-frequency energy of the seismic signals can be enhanced, and the longitudinal resolution of the seismic data is improved. However, the inverse Q filtering method based on the wave propagation principle needs to be done at the cost of huge time cost. Moreover, the inverse Q filtering method belongs to the inverse problem solving category, and the high-frequency component energy compensation has great instability and often has the problem of insufficient deep energy compensation. In addition, when inverse Q filtering is performed, the existing inverse Q filtering algorithm needs a known quality factor Q as a premise, and the accuracy of the Q value directly affects the inverse Q filtering effect. The quality factor Q is also an important parameter of the lithologic reservoir and can be used for hydrocarbon source rock indication and hydrocarbon reservoir distribution description. However, due to the reasons of unclear physical mechanism, seismic noise and the like, the Q value solving process is unstable, and the accuracy and the resolution of the Q value calculation result are limited. Q value estimation is both a hotspot and a difficulty of research.
Currently, Q estimation and seismic attenuation compensation are generally two independent processes. Firstly, extracting a Q value by using a Q value estimation method (such as a frequency spectrum ratio method, a rise time method, a center frequency offset and the like); then, the seismic attenuation compensation is realized by using an inverse Q filtering method. Because the traditional Q value estimation method mostly depends on the manual setting of an analysis time window, the Q value estimation result has larger uncertainty, the calculation precision of the Q value estimation result cannot meet the requirement of the fine characterization of an actual oil reservoir, and errors in the inverse Q filtering method are accumulated, so that the precision and the accuracy of the final earthquake attenuation compensation result cannot meet the requirements of earthquake structure interpretation and oil reservoir evaluation.
Disclosure of Invention
Based on the technical problems, the method for Q value estimation and seismic attenuation compensation based on the fusion network effectively avoids the accumulated error of the traditional two-step method, provides reliable Q value information while realizing high-resolution processing of seismic data, and meets the requirements of high-precision seismic exploration and fine reservoir characterization.
In order to solve the technical problems, the technical scheme adopted by the application is as follows:
a Q value estimation and seismic attenuation compensation method based on a fusion network comprises the following steps:
acquiring seismic data and logging data of a research area, and processing the seismic data and the logging data to obtain low-frequency Q value data and attenuation seismic data;
inputting the low-frequency Q value data and the attenuation seismic data into a Q value estimation model to obtain a Q value estimation result;
and inputting the attenuation seismic data and the Q value estimation result into an attenuation compensation model to obtain an attenuation compensation result.
A seismic attenuation compensation and Q value synchronous estimation device based on a fusion network comprises:
the data acquisition module is used for acquiring seismic data and logging data of a research area and processing the seismic data and the logging data to obtain low-frequency Q value data and attenuation seismic data;
the Q value estimation module is used for inputting the low-frequency Q value data and the attenuation seismic data into the Q value estimation model to obtain a Q value estimation result;
and the attenuation compensation module is used for inputting the attenuation seismic data and the Q value estimation result into the attenuation compensation model to obtain an attenuation compensation result.
A computer device comprising a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the above method for fusion network based Q estimation and seismic attenuation compensation.
A computer readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of the above-described fusion network based Q-value estimation and seismic attenuation compensation method.
Compared with the prior art, the beneficial effects of this application are:
method, apparatus, computer device and storage medium for the same
1. The method adopts a deep learning method to carry out seismic Q value estimation and attenuation compensation, and avoids the problems of low precision and poor stability under the assumption of the traditional physical model;
2. the method and the device construct a training data set by using a random simulation means, and guarantee the robustness of deep learning network training;
3. the seismic Q value estimation and attenuation compensation coupling loss function constructed by the method synchronously realizes seismic Q value estimation and attenuation compensation, avoids accumulated errors caused by two steps, and can effectively improve the accuracy of the seismic Q value estimation and attenuation compensation;
4. according to the method, the low-frequency Q value data constructed based on the geological background information is introduced in the network training and prediction stage, the stability of seismic attenuation compensation and Q value estimation can be well improved, and meanwhile the generalization performance of the seismic attenuation compensation and Q value estimation network is obviously improved.
5. The seismic Q value estimation model and the attenuation compensation model which are constructed by the method can be directly used for seismic Q value estimation and attenuation compensation after being trained, and have higher calculation efficiency compared with a traditional model driving method.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the application and together with the description serve to explain the application and not to limit the application. Wherein:
fig. 1 is a schematic flow chart of a Q value estimation and seismic attenuation compensation method based on a fusion network.
Fig. 2 is a schematic flow chart of a training method of the Q value estimation model and the attenuation compensation model.
Fig. 3 is a schematic flow chart of a method for obtaining accurate Q-value training data.
Fig. 4 is a flowchart illustrating a method for obtaining low-frequency Q-value training data.
FIG. 5 is a schematic flow chart of a method of acquiring unattenuated seismic training data.
FIG. 6 is a schematic diagram of a framework of a method for Q value estimation and seismic attenuation compensation based on a fusion network.
FIG. 7 is a partial example graph of training data.
FIG. 8 is a partial exemplary plot of model-input noiseless low frequency Q-value data and attenuated seismic data.
FIG. 9 is a partial exemplary plot of noisy (signal-to-noise ratio of 2) low frequency Q-value data and attenuated seismic data for a model input.
Fig. 10 shows the output results of the method for estimating the Q value and compensating for seismic attenuation based on the fusion network under the noise-free condition.
Fig. 11 shows the output result of the method for estimating the Q value and compensating for the seismic attenuation based on the fusion network in the case of noise (signal-to-noise ratio of 2).
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present disclosure more apparent, the technical solutions of the embodiments of the present disclosure will be described clearly and completely with reference to the drawings of the embodiments of the present disclosure. It is to be understood that the described embodiments are only a few embodiments of the present disclosure, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the described embodiments of the disclosure without any inventive step, are within the scope of protection of the disclosure.
It should be understood that "system", "device", "unit" and/or "module" as used in this specification is a method for distinguishing different components, elements, parts or assemblies at different levels. However, other words may be substituted by other expressions if they accomplish the same purpose.
As used in this specification and the appended claims, the terms "a," "an," "the," and/or "the" are not intended to be inclusive in the singular, but rather are intended to be inclusive in the plural, unless the context clearly dictates otherwise. In general, the terms "comprises" and "comprising" merely indicate that steps and elements are included which are explicitly identified, that the steps and elements do not form an exclusive list, and that a method or apparatus may include other steps or elements.
Flow charts are used in this description to illustrate operations performed by a system according to embodiments of the present description. It should be understood that the preceding or following operations are not necessarily performed in the exact order in which they are performed. Rather, the various steps may be processed in reverse order or simultaneously. Meanwhile, other operations may be added to the processes, or a certain step or several steps of operations may be removed from the processes.
Referring to fig. 1, in some embodiments, a method for Q value estimation and seismic attenuation compensation based on a fusion network includes:
s101, acquiring seismic data and logging data of a research area, and processing the seismic data and the logging data to obtain low-frequency Q value data and attenuation seismic data;
specifically, for the acquisition of low-frequency Q value data, a velocity model of each measurement line time domain in the logging data can be obtained by interpolating and extrapolating the logging data according to a construction mode; and then, substituting the velocity model into a relation function of the seismic Q value and the velocity to obtain a calculation result, and finally, performing smoothing processing on the calculation result to obtain low-frequency Q value data.
Specifically, for obtaining the attenuated seismic data, the attenuated seismic data can be directly obtained in a seismic data acquisition stage.
S102, inputting the low-frequency Q value data and the attenuation seismic data into a Q value estimation model to obtain a Q value estimation result;
and S103, inputting the attenuation seismic data and the Q value estimation result into an attenuation compensation model to obtain an attenuation compensation result.
In this embodiment, with reference to fig. 6, low-frequency Q value data and attenuation seismic data are used as inputs, and a trained deep learning network model is used to finally and synchronously realize Q value estimation and attenuation compensation of seismic data, so as to obtain a high-reliability Q value estimation result and attenuation compensation result.
The method adopts a deep learning method to carry out seismic Q value estimation and attenuation compensation, does not need to rely on artificial setting of an analysis time window like a traditional Q value estimation method, and avoids the problems of low precision and poor stability under the assumption of a traditional physical model.
Compared with the traditional inverse Q filtering method, the method solves the problem that the traditional inverse Q filtering method belongs to the inverse problem solving category, and the deep energy compensation is insufficient due to the fact that the high-frequency component energy compensation is extremely unstable.
The constructed seismic Q value estimation model and the attenuation compensation model can be directly used for seismic Q value estimation and attenuation compensation after being trained, and compared with a traditional model driving method and an inverse Q filtering method based on a wave field propagation theory, the method has higher calculation efficiency and greatly reduces calculation time cost.
Referring to fig. 2, in some embodiments, the method for training the Q estimation model and the fading compensation model includes:
s201, acquiring training data, wherein the training data comprises accurate Q value training data, unattenuated earthquake training data, corresponding low-frequency Q value training data and attenuated earthquake training data;
s202, inputting low-frequency Q-value training data and attenuation seismic training data into an untrained Q-value estimation model to obtain a first output result;
s203, inputting the attenuation seismic training data and the first output result into an untrained attenuation compensation model to obtain a second output result;
s204, determining a coupling loss function based on the accurate Q value training data, the unattenuated seismic training data, the first output result and the second output result;
and S205, 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 trained attenuation compensation model.
In the embodiment, based on the characteristics between the accurate Q value training data and the attenuated seismic training data and the characteristics between the unattenuated seismic training data and the attenuated seismic training data, a proper end-to-end deep learning network is established, a Q value estimation and attenuation compensation coupling loss function is further established, and the Q value estimation network is linked with the attenuation compensation network, so that a deep learning model jointly realized by the Q value estimation and the attenuation compensation is established, and the effectiveness and the stability of the seismic Q value estimation result and the attenuation compensation result are ensured.
Because the traditional two-step seismic attenuation compensation depends on the accuracy degree of a Q value, a large accumulated error exists in the process of realizing inverse Q filtering based on the estimated Q value. According to the method, the seismic Q value estimation and the attenuation compensation are synchronously realized through the constructed seismic Q value estimation and attenuation compensation coupling loss function, the accumulated error caused by a two-step method is avoided, and the accuracy of the seismic Q value estimation and the attenuation compensation can be effectively improved;
specifically, the constructed training data is input into a deep learning network model, the maximum iteration times are controlled through a coupling loss function value, network parameters are updated by adopting optimization algorithms such as random gradient and the like, and the network training process is repeated to minimize the loss function so as to obtain the optimal deep learning network parameters of a Q value estimation model and an attenuation compensation model.
Specifically, as for the structures of the Q value estimation model and the attenuation compensation model, both are deep learning network models having the same structure. The deep learning network model specifically comprises two convolutional layers and a full-link layer, the size of a convolutional core in each convolutional layer is 3 x 1, the number of the convolutional cores is 32, all feature vectors before the full-link layer are expanded into one-dimensional vectors, and the number of neurons in the full-link layer is 256.
And a Dropout layer is added after the second convolution layer of the deep learning network model of the Q value estimation model and the attenuation compensation model to prevent the network from being over-fitted, and a Max-posing layer is added to improve the network training efficiency.
Wherein ReLu is used as an activation function in a deep learning network model of a Q value estimation model and a decay compensation model.
Specifically, an Adam optimizer is selected for optimizing the network. The batch size and the number of iterations in the network training process are 128 and 1000 respectively, and the learning rate is 0.00005.
Preferably, the coupling loss function is specifically:
L＝‖Q-F([d i ,Q i ],θ 1 )‖ 2 +‖d t -G([d i ,F([d i ,Q i ],θ 2 )])‖ 2
where L represents the coupling loss function, Q represents the exact Q training data, F ([ d ] i ,Q i ],θ 1 ) Representing a first output result of the Q-value estimation network, d t Represents unattenuated seismic training data, G ([ d ] i ,F([d i ,Q i ],θ 2 )]) A second output result representing the fading compensation network;
wherein d is i Representing attenuated seismic training data, Q i Representing low frequency Q-value training data, θ 1 Expressing Q-value estimated network optimization parameter, theta 2 Representing the attenuation compensation network optimization parameters.
In particular, i, t are used to distinguish different meanings of the same letter.
The method comprises the steps of taking low-frequency Q value training data and attenuation seismic training data as input, simultaneously inputting a Q value estimation network and an attenuation compensation network, further inputting an output result of the Q value estimation network into the attenuation compensation network, ensuring the stability of a network training process by establishing a coupling loss function of the Q value estimation network and the attenuation compensation network, and synchronously realizing seismic Q value estimation and attenuation compensation.
Since seismic Q-value estimation and attenuation compensation belong to the regression problem, the root Mean Square Error (MSE) is specifically chosen as the loss function.
In some embodiments, the method for obtaining training data generally includes obtaining accurate Q-value training data, unattenuated seismic training data, low frequency Q-value training data, and attenuated seismic training data:
referring to fig. 3, in particular, the method for obtaining the accurate Q-value training data includes:
s301, collecting seismic structure interpretation data and well logging data of a research area, wherein the seismic structure interpretation data comprise geological structure information;
and S302, randomly generating accurate Q value training data by using a random simulation algorithm based on the geological structure information and the logging information.
Referring to fig. 4, the method of acquiring low frequency Q-value training data includes:
s401, interpolating and extrapolating the logging information according to a construction mode to obtain a speed model of each measuring line time domain in the logging information;
s402, the velocity model is brought into a relation function of the seismic Q value and the velocity to obtain initial Q value data, and the initial Q value data is subjected to smoothing processing to obtain low-frequency Q value training data.
The low-frequency Q value is the background of the accurate Q value, and in practical application, the low-frequency Q value can be obtained through a relation function of the seismic Q value and the velocity, and any number of low-frequency Q values can be obtained through smoothing.
By using the low-frequency Q value as an input parameter in the model training and application process, the multi-solution of the inversion result can be reduced, and the accuracy of the output result of the model can be improved.
Geological background information is ignored in the conventional inverse Q filtering and Q value estimation processes, and instability in the data processing process is increased; in the deep learning-based method, the stability of the network prediction result and the network generalization capability can be improved by adding geological background information. The method comprises the following steps of firstly picking up a geological layer, carrying out extrapolation interpolation by taking layer information as a control point and combining logging data, namely carrying out transverse interpolation on the logging information, calculating to obtain a parameter value of each underground point, and completing the task of initial parameter modeling. Because the logging data does not directly contain Q value information, the invention completes the task of initial Q value modeling by utilizing an empirical formula for conversion between the Q value and the speed based on the initial speed modeling result.
Specifically, the relation function of the seismic Q value and the velocity is specifically as follows:
Q＝a*V p b
wherein Q represents a Q value, V p Representing the velocity, a, b represent constants.
Wherein a and b vary according to different geological structures of the research area.
In practical application, the velocity is easily obtained by the existing method, and under the condition of known velocity, the low-frequency Q value can be obtained by using an empirical formula of the seismic Q value and the velocity.
In addition, low-frequency Q-value training data may also be obtained by smoothing the accurate Q-value training data.
Referring to FIG. 5, a method of acquiring unattenuated seismic training data includes:
s501, extracting a first seismic wavelet from seismic structure interpretation data and logging data by adopting a statistical method;
s502, calculating and extracting frequency spectrum information of the first seismic wavelet based on a Fourier transform method, and analyzing frequency spectrum characteristics of the first seismic wavelet to obtain dominant frequency characteristics of the first seismic wavelet;
s503, constructing second seismic wavelets with different dominant frequencies according to the dominant frequency characteristics;
any number of second seismic wavelets can be constructed according to the dominant frequency characteristics of the seismic wavelets extracted from the research area, and the non-attenuated seismic training data and the attenuated seismic training data which are generated correspondingly contain the data distribution characteristics of the research area, so that the applicability of a Q value estimation model and an attenuation compensation model obtained by training is improved.
And S504, inputting the second seismic wavelet and the velocity model into a convolution model to obtain non-attenuated seismic training data.
Specifically, the method for acquiring the attenuated seismic training data comprises the following steps:
substituting the unattenuated seismic training data and the accurate Q value training data into an attenuation theoretical model to obtain attenuation training data;
wherein, the attenuation model specifically is:
where x (f, τ) represents attenuated seismic training data, x (f,0) represents unattenuated seismic training data, f represents the frequency of the seismic wavelet, τ represents the total time of propagation of the seismic wavelet, and Q eff Equivalent Q value, f, representing accurate Q training data r Representing a reference frequency of a seismic wavelet;
wherein the total time τ is obtained according to the following formula:
wherein, tau i Representing the time required for the seismic wavelet to propagate through the ith layer in the model, and n represents the total number of layers.
Wherein the equivalent Q value Q is obtained according to the following formula eff ：
Wherein Q is i The exact Q value of the ith layer in the exact Q value training data is represented.
In the embodiment, accurate Q value training data is randomly generated through geological structure information and logging data; generating a velocity model through the logging data, and obtaining non-attenuated seismic training data by using the velocity model; and subsequently substituting the un-attenuated seismic training data and the accurate Q value training data into the attenuation theoretical model so as to obtain attenuated seismic training data.
From the acquisition process of the training set, it can be seen that since the accurate Q training data and velocity model are generated based on the acquired seismic structure interpretation data and well log data, the subsequent unattenuated seismic training data are obtained via the velocity model. Therefore, for the accurate Q training data and the un-attenuated training data, it is preset that they are accurate data during the whole model training process. Therefore, low-frequency Q-value training data and attenuated seismic training data are obtained through the accurate Q-value training data and the unattenuated training data.
Because the physical mechanism of Q value estimation is not clear, the existing Q value estimation method has the problems of unstable Q value solving and limited Q value calculation result precision and resolution. According to the method, under the condition that accurate Q-value training data and unattenuated training data are preset accurately, the obtained low-frequency Q-value training data and attenuated seismic training data are input into a Q-value estimation model and an attenuated compensation model, the mapping relation between the low-frequency Q-value training data and the attenuated seismic training data and the mapping relation between the accurate Q-value training data and the unattenuated training data are learned through the models, and after the models learn the rules through training, the low-frequency Q-value training data and the attenuated seismic data are input, so that accurate Q-value estimation results and attenuated compensation results can be obtained.
In addition, the deep learning network model training requires enough training data as support, and any number of training sample databases can be constructed by utilizing the acquired seismic structure interpretation data and logging data through a random simulation method, so that a high-precision Q value estimation model and an attenuation compensation model are trained by utilizing the training data.
By integrating the above embodiments, the Q value estimation and seismic attenuation compensation method based on the fusion network is further explained by combining the relevant data:
1. referring to fig. 7, training data is input into the untrained Q value estimation model and the attenuation compensation model to obtain a trained Q value estimation model and an attenuation compensation model;
wherein (a) represents low frequency Q-value training data, (b) represents attenuated seismic training data, (c) represents accurate Q-value training data, and (d) represents attenuated seismic training data in fig. 7;
2. referring to fig. 8, the noiseless low-frequency Q value data and the attenuation seismic data are input into the Q value estimation model and the attenuation compensation model, and the obtained output result is referred to fig. 10;
wherein (a) represents low frequency Q-value data and (b) represents attenuated seismic data in fig. 8;
in fig. 10, (a) shows the result of attenuation compensation, and (b) shows the result of Q value estimation;
3. referring to fig. 9, the low frequency Q value data and the attenuation seismic data containing noise (signal to noise ratio of 2) are input into the Q value estimation model and the attenuation compensation model, and the obtained output result refers to fig. 11;
wherein (a) represents low frequency Q-value data and (b) represents attenuated seismic data in fig. 9;
in fig. 11, (a) shows the attenuation compensation result, and (b) shows the Q value estimation result.
In conclusion, it can be seen from the data in the figure that the high-precision Q value estimation result and the high-precision attenuation compensation result can be obtained by the fusion network-based Q value estimation and seismic attenuation compensation method.
In some embodiments, a seismic attenuation compensation and Q value synchronization estimation apparatus based on a fusion network is further disclosed, including:
the data acquisition module is used for acquiring seismic data and logging data of a research area and processing the seismic data and the logging data to obtain low-frequency Q value data and attenuation seismic data;
the Q value estimation module is used for inputting the low-frequency Q value data and the attenuation seismic data into the Q value estimation model to obtain a Q value estimation result;
and the attenuation compensation module is used for inputting the attenuation seismic data and the Q value estimation result into the attenuation compensation model to obtain an attenuation compensation result.
In order to solve the technical problem, the present application further discloses a computer device, which is characterized by comprising 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 method for estimating the Q value and compensating the seismic attenuation based on the fusion network.
The computer device may be a desktop computer, a notebook, a palm computer, a cloud server, or other computing devices. The computer equipment can carry out man-machine interaction with a user through a keyboard, a mouse, a remote controller, a touch panel or voice control equipment and the like.
The memory includes at least one type of readable storage medium including flash memory, hard disks, multimedia cards, card-type memory (e.g., 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, magnetic disks, optical disks, and the like. In some embodiments, the storage may be an internal storage unit of the computer device, such as a hard disk or a memory of the computer device. In other embodiments, the memory may also be an external storage device of the computer device, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), or the like provided on the computer device. Of course, the memory may also include both internal and external storage devices of the computer device. In this embodiment, the memory is commonly used for storing an operating system and various application software installed in the computer device, such as program codes of the method for estimating the Q value and compensating the seismic attenuation based on the converged network. Further, the memory may be used to temporarily store various types of data that have been output or are to be output.
The processor may be a Central Processing Unit (CPU), controller, microcontroller, microprocessor, or other data Processing chip in some embodiments. The processor is typically used to control the overall operation of the computer device. In this embodiment, the processor is configured to execute the program code stored in the memory or process data, for example, execute the program code of the method for estimating the Q value and compensating for seismic attenuation based on the fusion network.
In order to solve the above technical problem, the present application further discloses a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the method for estimating the Q value and compensating for the seismic attenuation based on the fusion network.
Wherein the computer readable storage medium stores an interface display program executable by at least one processor to cause the at least one processor to perform the steps of the fusion network based Q-value estimation and seismic attenuation compensation method as described above.
Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solutions of the present application may be embodied in the form of a software product, which is stored in a storage medium (e.g., ROM/RAM, magnetic disk, optical disk) and includes instructions for enabling a terminal device (e.g., a mobile phone, a computer, a server, or a network device) to execute the method according to the embodiments of the present application.
The above is an embodiment of the present application. The embodiments and specific parameters in the embodiments are only used for clearly illustrating the verification process of the application and are not used for limiting the patent protection scope of the application, which is defined by the claims, and all the equivalent structural changes made by using the contents of the specification and the drawings of the application should be included in the protection scope of the application.
Claims (10)
1. The Q value estimation and seismic attenuation compensation method based on the fusion network is characterized by comprising the following steps:
acquiring seismic data and logging data of a research area, and processing the seismic data and the logging data to obtain low-frequency Q value data and attenuation seismic data;
inputting the low-frequency Q value data and the attenuation seismic data into a Q value estimation model to obtain a Q value estimation result;
and inputting the attenuation seismic data and the Q value estimation result into an attenuation compensation model to obtain an attenuation compensation result.
2. The method for Q-value estimation and seismic attenuation compensation based on a fusion network of claim 1, wherein the training method of the Q-value estimation model and the attenuation compensation model comprises the following steps:
acquiring training data, wherein the training data comprises accurate Q value training data, 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 seismic 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;
determining a coupling loss function based on the refined Q-value training data, the unattenuated seismic training data, the first output result, and the second output result;
and performing joint iterative training on the Q value estimation model and the attenuation compensation model based on the coupling loss function to obtain the Q value estimation model and the attenuation compensation model which are trained.
3. The method for Q-value estimation and seismic attenuation compensation based on a converged network of claim 2, wherein the coupling loss function is specifically:
L＝‖Q-F([d i ,Q i ],θ 1 )‖ 2 +‖d t -G([d i ,F([d i ,Q i ],θ 2 )])‖ 2
where L represents the coupling loss function, Q represents the exact Q training data, F ([ d ] i ,Q i ],θ 1 ) Representing a first output result of the Q-value estimation network, d t Represents unattenuated seismic training data, G ([ d ] i ,F([d i ,Q i ],θ 2 )]) A second output result representing the fading compensation network;
wherein d is i Representing attenuated seismic training data, Q i Representing low frequency Q-value training data, θ 1 Expressing Q-value estimated network optimization parameter, theta 2 Representing the attenuation compensation network optimization parameters.
4. The fusion network based Q-value estimation and seismic attenuation compensation method of claim 2, wherein the method of obtaining the accurate Q-value training data comprises:
acquiring seismic structure interpretation data and well logging data of a research area, wherein the seismic structure interpretation data comprise geological structure information;
and randomly generating accurate Q value training data by using a random simulation algorithm based on the geological structure information and the logging information.
5. The fusion network based Q-value estimation and seismic attenuation compensation method of claim 4, wherein the method of obtaining the low frequency Q-value training data comprises:
interpolating and extrapolating the logging information according to a construction mode to obtain a speed model of each measuring line time domain in the logging information;
and substituting the velocity model into a relation function of 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 the low-frequency Q value training data.
6. The fusion network based Q-value estimation and seismic attenuation compensation method of claim 5, wherein the method of obtaining the unattenuated seismic training data comprises:
extracting a first seismic wavelet from the seismic structure interpretation data and the logging data by adopting a statistical method;
calculating and extracting frequency spectrum information of the first seismic wavelet based on a Fourier transform method, and analyzing frequency spectrum characteristics of the first seismic wavelet to obtain dominant frequency characteristics of the first seismic wavelet;
constructing second seismic wavelets with different dominant frequencies according to the dominant frequency characteristics;
and inputting the second seismic wavelet and the velocity model into a convolution model to obtain the unattenuated seismic training data.
7. The fusion network based Q-value estimation and seismic attenuation compensation method of claim 6, wherein the method of obtaining the attenuated seismic training data comprises:
substituting the unattenuated seismic training data and the accurate Q value training data into an attenuation theoretical model to obtain attenuation training data;
wherein, the attenuation model specifically comprises:
where x (f, τ) represents attenuated seismic training data, x (f,0) represents unattenuated seismic training data, f represents the frequency of the seismic wavelet, τ represents the total time of propagation of the seismic wavelet, and Q eff Equivalent Q value, f, representing accurate Q training data r Representing a reference frequency of a seismic wavelet;
wherein the total time τ is obtained according to the following formula:
wherein, tau i Representing the time required for the seismic wavelet to propagate through the ith layer in the model, and n represents the total number of layers.
Wherein the equivalent Q value Q is obtained according to the following formula eff ：
Wherein Q is i The exact Q value of the ith layer in the exact Q value training data is represented.
8. Earthquake attenuation compensation and Q value synchronous estimation device based on fusion network, characterized by comprising:
the data acquisition module is used for acquiring seismic data and logging data of a research area and processing the seismic data and the logging data to obtain low-frequency Q value data and attenuation seismic data;
the Q value estimation module is used for inputting the low-frequency Q value data and the attenuation seismic data into a Q value estimation model to obtain a Q value estimation result;
and the attenuation compensation module is used for inputting the attenuation seismic data and the Q value estimation result into an attenuation compensation model to obtain an attenuation compensation result.
9. A computer device, characterized by: comprising a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the fusion network based Q-value estimation and seismic attenuation compensation method according to any of claims 1 to 7.
10. A computer-readable storage medium characterized by: a computer program is stored which, when executed by a processor, causes the processor to carry out the steps of the method of fusion network based Q-value estimation and seismic attenuation compensation according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210756266.3A CN115079257A (en) | 2022-06-30 | 2022-06-30 | Q value estimation and seismic attenuation compensation method based on fusion network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210756266.3A CN115079257A (en) | 2022-06-30 | 2022-06-30 | Q value estimation and seismic attenuation compensation method based on fusion network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115079257A true CN115079257A (en) | 2022-09-20 |
Family
ID=83256582
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210756266.3A Pending CN115079257A (en) | 2022-06-30 | 2022-06-30 | Q value estimation and seismic attenuation compensation method based on fusion network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115079257A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117471529A (en) * | 2023-10-30 | 2024-01-30 | 西南交通大学 | Unsteady seismic wavelet self-adaptive extraction method |
-
2022
- 2022-06-30 CN CN202210756266.3A patent/CN115079257A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117471529A (en) * | 2023-10-30 | 2024-01-30 | 西南交通大学 | Unsteady seismic wavelet self-adaptive extraction method |
| CN117471529B (en) * | 2023-10-30 | 2024-05-07 | 西南交通大学 | Unsteady seismic wavelet self-adaptive extraction method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111596366B (en) | Wave impedance inversion method based on seismic signal optimization processing | |
| CN112444841B (en) | Thin-layer-containing lithology earthquake prediction method based on scale-division multi-input convolution network | |
| CN113919219A (en) | Stratum evaluation method and system based on logging big data | |
| US20240045093A1 (en) | Denoising methods and systems for preserving amplitude variation with offset (avo) features of pre-stack seismic data | |
| CN113687433B (en) | Bi-LSTM-based magnetotelluric signal denoising method and system | |
| CN108897042A (en) | Content of organic matter earthquake prediction method and device | |
| CN110879412A (en) | Underground transverse wave velocity inversion method, device, computing equipment and storage medium | |
| CN114966861B (en) | Seismic denoising method based on Lp pseudo-norm and gamma-norm sparse low-rank constraint | |
| CN110687597B (en) | Wave impedance inversion method based on joint dictionary | |
| CN113109874A (en) | Wave impedance inversion method using neural network and neural network system | |
| CN115079257A (en) | Q value estimation and seismic attenuation compensation method based on fusion network | |
| CN111582114A (en) | Seismic fault identification method, device, equipment and storage medium | |
| Wang et al. | Denoising with weak signal preservation by group-sparsity transform learning | |
| CN115877464B (en) | Lithology recognition method and device, computer equipment and storage medium | |
| CN112444850A (en) | Seismic data velocity modeling method, storage medium, and computing device | |
| CN112578458B (en) | Pre-stack elastic impedance random inversion method and device, storage medium and processor | |
| Zhang et al. | First break of the seismic signals in oil exploration based on information theory | |
| CN111897004B (en) | Logging prediction method based on big data analysis technology | |
| CN113722893A (en) | Seismic record inversion method, device, equipment and storage medium | |
| CN109633744B (en) | Method, device and equipment for extracting seismic wavelets and storage medium | |
| CN113341463A (en) | Pre-stack seismic data non-stationary blind deconvolution method and related components | |
| CN112346126A (en) | Low-order fault identification method, device, equipment and readable storage medium | |
| CN110646846A (en) | VIT medium anisotropy parameter determination method, device and equipment | |
| Xu et al. | Simultaneous inversion for reflectivity and q using nonstationary seismic data with deep-learning-based decoupling | |
| CN117849875B (en) | Earthquake signal analysis method, system, device and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |