CN113495291A - Method for intelligently and quantitatively evaluating amplitude preservation of forward gather based on deep learning - Google Patents
Method for intelligently and quantitatively evaluating amplitude preservation of forward gather based on deep learning Download PDFInfo
- Publication number
- CN113495291A CN113495291A CN202110848280.1A CN202110848280A CN113495291A CN 113495291 A CN113495291 A CN 113495291A CN 202110848280 A CN202110848280 A CN 202110848280A CN 113495291 A CN113495291 A CN 113495291A
- Authority
- CN
- China
- Prior art keywords
- quantitative evaluation
- data
- gather
- deep learning
- amplitude
- 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 50
- 238000004321 preservation Methods 0.000 title claims abstract description 42
- 238000013135 deep learning Methods 0.000 title claims abstract description 23
- 238000011158 quantitative evaluation Methods 0.000 claims abstract description 56
- 238000012545 processing Methods 0.000 claims abstract description 34
- 238000012549 training Methods 0.000 claims abstract description 17
- 230000000903 blocking effect Effects 0.000 claims description 11
- 238000000605 extraction Methods 0.000 claims description 8
- 238000013528 artificial neural network Methods 0.000 claims description 6
- 238000011176 pooling Methods 0.000 claims description 6
- 239000006185 dispersion Substances 0.000 claims description 4
- 238000013527 convolutional neural network Methods 0.000 claims description 3
- 238000005192 partition Methods 0.000 claims description 3
- 238000000638 solvent extraction Methods 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims description 2
- 238000013139 quantization Methods 0.000 abstract description 4
- 238000004458 analytical method Methods 0.000 description 16
- 238000011156 evaluation Methods 0.000 description 9
- 230000000694 effects Effects 0.000 description 7
- 238000003672 processing method Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 230000001419 dependent effect Effects 0.000 description 3
- 238000012417 linear regression Methods 0.000 description 3
- 230000014759 maintenance of location Effects 0.000 description 3
- 239000004215 Carbon black (E152) Substances 0.000 description 2
- 230000015572 biosynthetic process Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 229930195733 hydrocarbon Natural products 0.000 description 2
- 150000002430 hydrocarbons Chemical class 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000010521 absorption reaction Methods 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001427 coherent effect Effects 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000011157 data evaluation Methods 0.000 description 1
- 230000008021 deposition Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009792 diffusion process Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000000704 physical effect Effects 0.000 description 1
- 239000011435 rock Substances 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 238000011410 subtraction method Methods 0.000 description 1
- 230000002194 synthesizing effect Effects 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
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V13/00—Manufacturing, calibrating, cleaning, or repairing instruments or devices covered by groups G01V1/00 – G01V11/00
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Remote Sensing (AREA)
- General Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Geophysics (AREA)
- Acoustics & Sound (AREA)
- Environmental & Geological Engineering (AREA)
- Geology (AREA)
- Manufacturing & Machinery (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The invention discloses a method for intelligently and quantitatively evaluating the amplitude preservation of a forward gather based on deep learning, which comprises the following steps of: step one, constructing a quantitative evaluation training data set; training a quantitative evaluation deep learning network; and thirdly, quantitative evaluation prediction based on deep learning. The invention can intelligently predict the amplitude-preserving quantization parameter in a data-driven mode without depending on any prior information. The relative amplitude preservation of the seismic processing result data can be improved, and the prediction precision of the lithologic oil and gas reservoir is further improved.
Description
Technical Field
The invention relates to a method for intelligently and quantitatively evaluating the amplitude preservation of a forward gather based on deep learning.
Background
With the continuous development of oil and gas exploration technology and the demand for deeper oil and gas resources, the exploration of concealed oil and gas reservoirs gradually becomes an important direction of research work in the field of oil and gas exploration. Lithological reservoirs are an important category of covert reservoirs, but exploration prediction efforts for lithological reservoirs face a number of challenges.
Research work on lithologic hydrocarbon reservoirs from qualitative to quantitative, reservoir lithology to physical property and hydrocarbon property discrimination and the like faces a plurality of challenges. The lithologic trap identification method requires higher-quality seismic data, and simultaneously requires that the processed seismic data have higher amplitude retention. If the relative amplitude retention of the processed seismic data is low, the application of the fine reservoir prediction technique generally does not achieve a good effect. Therefore, how to enable the seismic data to still keep higher relative amplitude retention after being processed so as to improve the accuracy of reservoir prediction of lithologic oil and gas reservoirs is a problem which needs to be solved urgently at present. However, the problem of judging whether the amplitude preservation standard of the seismic data processing technology reaches the standard still lacks a set of corresponding amplitude preservation analysis and evaluation system.
There are many methods for discriminating and evaluating the amplitude preservation processing of seismic data, which evaluate the amplitude preservation of a plurality of seismic data processing techniques at different angles, and by using the seismic wave amplitude characteristics, frequency characteristics, seismic attribute slices, consistency of wave impedance and logging results, etc. as evaluation bases, the conventional methods for evaluating the amplitude preservation of seismic data can be broadly divided into 9 types, and each evaluation method is briefly introduced below.
1) Residual analysis method: the residual analysis method comprises a subtraction method and a residual method, and is mainly used for amplitude preservation evaluation of the denoising method. And directly subtracting the seismic records before and after the denoising treatment, and observing whether the residual data contains effective signals or not to judge the amplitude preservation of the denoising treatment method. In general, the result after the filter process with the characteristics of zero phase and amplitude all-pass can be considered to be relatively amplitude-preserved.
2) A time-frequency analysis method: the time-frequency analysis method can obtain the distribution condition of frequency components or frequencies at a certain time in the seismic signals at different moments. And performing time-frequency analysis on the seismic signals before and after processing, and judging that the processing technology is suitable for amplitude-preserving processing if the change of the time-frequency characteristics (instantaneous amplitude, instantaneous frequency and instantaneous phase) of the wavelets is not abnormal.
3) Amplitude curve comparison method: in the process of seismic wave propagation, due to the diffusion effect, construction factors, noise and the influence of stratum absorption, the amplitude of the reflected wave of the middle-deep layer and the amplitude between adjacent records are lost. These losses have an effect on the seismic data, and the damaged amplitude is usually compensated for by amplitude correction. The amplitude preservation of the processing method of the amplitude compensation class can be judged by using the amplitude curves before and after the compensation processing.
4) Amplitude ratio calculation method: the method is used for evaluating the amplitude preservation performance of a resolution improving processing method such as deconvolution. After the seismic data are processed, the absolute value of the amplitude is changed, but the relative relation between the amplitudes is kept unchanged, and the amplitude preservation of the resolution processing method such as deconvolution can be judged by comparing the consistency of the amplitude ratios of corresponding positions before and after processing.
5) Wavelet consistency analysis: the method is used for improving the amplitude preservation evaluation of the resolution processing method. The existing resolution enhancement processing means directly or indirectly widen and widen the seismic record amplitude spectrum, which destroys the relative relation between reflection coefficients. The amplitude preservation evaluation of the resolution enhancement processing can investigate the damage degree of the relative relation between the before-processing and the after-processing reflection coefficients, so as to judge the amplitude preservation of the used processing technology.
6) AVO attribute analysis: the method is used for evaluating the amplitude preservation of seismic gather processing. The logging data provides rock physical parameters in a shaft, an AVA curve of each reflecting layer can be calculated according to the parameters, and wave impedance can be used for synthesizing a zero offset seismic trace. The reliability of AVA relation at the well point after amplitude preservation processing and amplitude preservation of the zero offset seismic trace can be judged by taking the two results as standards. Comparing the variation relation of the amplitudes of the reflected waves in the synthetic trace set with the variation relation of the amplitudes of the reflected waves in the processed seismic trace set can verify the amplitude preservation of the processing method on the pre-stack trace set.
7) Seismic attribute analysis along the horizon: the method is mainly used for amplitude preservation analysis of the post-stack and pre-stack seismic data along the layer attributes (amplitude, frequency and the like). In a certain exploration area, a stable deposition environment stage exists near a target layer, a continuous reflecting layer with all-area stability is formed, and seismic attributes extracted along the continuous reflecting layer have stable consistency by taking the continuous reflecting layer as a reference. Amplitude preservation analysis can be performed by comparing the analysis maps of the amplitude of the seismic data along the horizon before and after the earth surface consistency amplitude compensation processing.
8) Slice analysis method: the method can be applied to coherent body slice analysis and other attribute data body slice analysis. The reasonableness of the construction of different imaging data volumes can be explained and judged by using time slicing. The slice analysis method can also be used for evaluating the amplitude preservation performance of a resolution-improving processing method based on post-stack data, and the judgment method is used for comparing the seismic data before and after resolution-improving processing.
9) Synthetic recording method: the method is mainly used for evaluating the amplitude preservation of the processing result data. And converting the formation speed according to the logging data, calculating the normal reflection coefficient of each formation interface, and obtaining a synthetic record beside the well by using a convolution model. The amplitude variation relationship of the reflected waves at different times in the synthetic record is compared with the amplitude variation relationship of the reflected waves at different times in the processed seismic channel beside the well, so that the amplitude preservation performance of the processing method in the longitudinal direction can be verified.
The methods have respective emphasis, evaluate the amplitude preservation of various seismic data processing technologies, and have strong practicability. However, the methods have a common defect that the methods depend on the hypothesis or the prior knowledge of the model, so that the application of amplitude preservation evaluation has limitation. When these assumptions and priors are violated, the performance of amplitude preservation evaluation is greatly reduced.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a method capable of intelligently predicting amplitude-preserving quantization parameters in a data-driven manner without depending on any prior information. The method can improve the relative amplitude preservation of the seismic processing result data, and further improve the accuracy of lithologic oil and gas reservoir prediction.
The purpose of the invention is realized by the following technical scheme: the method for intelligently and quantitatively evaluating the amplitude preservation of the forward gather based on deep learning comprises the following steps:
step one, constructing a quantitative evaluation training data set;
training a quantitative evaluation deep learning network;
and thirdly, quantitative evaluation prediction based on deep learning.
Further, the first specific implementation method of the step is as follows: bias previous gather dataWherein n isxAnd nyRespectively representing the data number of xline and inline in a front-channel set;
dividing a data set into m along an xline direction and an inline directionxAnd myBlock, dxAnd dyRespectively representing the size of each partition:
each picture region having d after being partitionedx×dyOne pixel, the data after being blocked is recorded as Si,j,Wherein Si,jThe blocks representing ith row and j column are output as the quantitative evaluation value f of the corresponding blocki,jSolving through the step two;
to the front gatherCarrying out noise reduction treatment to obtain a new front-biased gatherPerforming the same blocking processing on the Q as the P to obtain blocking data
To Si,jAnd Ti,jCalculating SNR of the signal to noise ratioi,jAnd according to the signal-to-noise ratio SNRi,jThe magnitude of (d) discretizes the signal-to-noise ratio into five indices: the indexes of the dispersion are taken as quantitative evaluation labels of the data and recorded as the indexes of the dispersionWherein N represents the number of complete pictures for partitioning;
the partial front gather after noise reduction processingAnd quantitative evaluation labelAs training data.
Further, the second specific implementation method of the step is as follows: obtaining nonlinear relation fitting of a regression model by adopting a convolutional neural network model based on residual learning, so as to obtain quantitative evaluation values of corresponding blocks from the data of the partial front gather after the blocking processing; the network consists of two major parts: a forward trace set feature extraction part and a seismic image quality quantitative evaluation output part;
the partial front gather feature extraction part consists of a VGG-19 network and a residual error layer, wherein the VGG-19 network consists of 5 convolution modules CONV 1-CONV 5 and three full connection layers FC1-FC3, wherein CONV1 and CONV2 respectively comprise two convolution layers and one pooling layer, and CONV 3-CONV 5 respectively comprise three convolution layers and one pooling layer; extracting the seismic image characteristics of the first convolution module by the residual layer, and sending the seismic image characteristics to the last convolution layer to be summed with the seismic image characteristics of the convolution layer; and then, obtaining quantitative evaluation values from the summed result through three full-connection layers FC1-FC3, and inputting the extracted quantitative evaluation values into a seismic image quality quantitative evaluation module so as to output a seismic image quality quantitative evaluation result.
Further, the third specific implementation method of the step is as follows: and (4) blocking the new two-dimensional forward gather data, and inputting the data into a residual neural network to obtain a quantitative evaluation value corresponding to the block.
The invention has the beneficial effects that: the invention can intelligently predict the amplitude-preserving quantization parameter in a data-driven mode without depending on any prior information. The relative amplitude preservation of the seismic processing result data can be improved, and the prediction precision of the lithologic oil and gas reservoir is further improved.
Drawings
FIG. 1 is a schematic diagram of regression model-based quantitative evaluation of the amplitude preservation of an off-the-front gather;
FIG. 2 is a diagram illustrating the prediction effect of a deep learning based nonlinear regression model and a model-based regression method;
FIG. 3 is a schematic diagram of quantitative evaluation deep learning network training;
FIG. 4 is a schematic structural diagram of a quantitative evaluation deep learning network;
fig. 5 is a schematic diagram of depth learning based quantitative assessment prediction.
Detailed Description
The invention provides a novel prediction method for quantitative evaluation indexes of the amplitude preservation of an anterior gather. The concrete content is as follows: firstly, converting the problem of carrying out amplitude preservation quantitative evaluation on a forward gather into a nonlinear regression scientific problem, and outputting the denoised seismic image data subjected to blocking processing as input data to be quantitative evaluation values corresponding to seismic image blocks; secondly, constructing a training sample set, firstly carrying out block processing on the selected seismic data set and the image in the seismic data set after noise reduction processing, calculating the signal-to-noise ratio (signal noise ratio) of each image area, and discretizing the image area into five indexes according to the SNR: very good, general, poor, very poor, which is used as a quantitative assessment label for the dataset; secondly, training a deep neural network aiming at a training data set constructed by the method, namely a seismic data set subjected to noise reduction and a quantitative evaluation label discretized according to an SNR value; finally, amplitude preservation quantitative evaluation can be carried out on the seismic image of the front gather by using the trained model. The technical scheme of the invention is further explained by combining the attached drawings.
Modeling can be performed using a multiple regression model, with the processed data and quantitative evaluation labels known. Multiple regression models are commonly used to study the relationship between the dependent variables and the independent variables, and the relationship between the dependent variables and the independent variables can be described in a linear form or in a nonlinear form, as shown in fig. 1. Suppose the block data of the partial gather is x and the quantization evaluation value is y. Then the simplest multiple linear regression model can be used to describe the random linear relationship between variables y and x, namely:
y=β0+β1x1+...βkxk+ξ (1)
in the formula, x1,...,xkA non-random variable; y is a random dependent variable; beta is a0,...,βkIs a regression coefficient; ξ is the random error term. If multiple observations of y and x are made, n sets of observations are obtained that satisfy the following relationships:
yi=β0+β1x1i+...+βkxki+ξi(2) expressed in a matrix:
at this point, the model can be written as:
y=Xβ+ξ (4)
the linear regression model can be solved by using a least square method to obtain a regression coefficient beta0,...,βkHowever, it is obvious that the linear regression model cannot fit well to the complex nonlinear relationship between the actual forward gather and the quantitative evaluation value, so we need to model this complex nonlinear relationship as a nonlinear regression model:
y=f(X,β)+ξ (5)
where f () is a complex nonlinear function. There are many methods for solving the nonlinear regression model, such as SVM, neural network, bayesian filter, etc. Considering that a complex relationship is presented between the processed data and the quantitative evaluation label, if a model-based regression method is simply adopted, a large error exists, and a data-driven method can achieve a fitting effect with a smaller error based on a deep learning regression method, so that a better prediction effect is achieved, as shown in fig. 2.
Based on the above modeling thought and the application characteristics of the deep learning network, the invention provides a method for intelligently and quantitatively evaluating the amplitude preservation of a forward gather based on deep learning, which comprises the following steps:
step one, constructing a quantitative evaluation training data set; the specific implementation method comprises the following steps: bias previous gather dataWherein n isxAnd nyRespectively representing the data number of xline and inline in a front-channel set;
dividing a data set into m along an xline direction and an inline directionxAnd myBlock, dxAnd dyRespectively representing the size of each partition:
each picture region having d after being partitionedx×dyOne pixel, the data after being blocked is recorded as Si,j,Wherein Si,jThe blocks representing ith row and j column are output as the quantitative evaluation value f of the corresponding blocki,j, solving through the step two;
to the front gatherCarrying out noise reduction treatment to obtain a new front-biased gatherPerforming the same blocking processing on the Q as the P to obtain blocking data
To Si,jAnd Ti,jCalculating SNR of the signal to noise ratioi,jAnd according to the signal-to-noise ratio SNRi,jThe magnitude of (d) discretizes the signal-to-noise ratio into five indices: the specific indexes can be set by users, and the scattered indexes are taken as quantitative evaluation labels of the data and recorded asWherein N represents the number of complete pictures for partitioning;
the partial front gather after noise reduction processingAnd quantitative evaluation labelAs training data.
Training a quantitative evaluation deep learning network; the specific implementation method comprises the following steps: after obtaining the training data setAnd corresponding calculated quantitative evaluation labelThen, the present invention proposes
Obtaining nonlinear relation fitting of a regression model by adopting a convolutional neural network model based on residual learning, so as to obtain quantitative evaluation values of corresponding blocks from the data of the partial front gather after the blocking processing, as shown in fig. 3; the network consists of two major parts: a forward trace set feature extraction part and a seismic image quality quantitative evaluation output part;
in the neural network, the more the number of convolution layers is, the deeper the network structure is, the more abstract and higher the seismic image feature map is, however, the number of layers is increased, and the loss of some bottom-layer detail information and position information is also caused, and the loss of the information has an adverse effect on the final seismic image quality quantitative evaluation result. In order to solve the problem that the quantitative evaluation result of the seismic image quality is inaccurate due to incomplete seismic feature extraction caused by loss of seismic image detail information in the convolution process, the partial front gather feature extraction part introduces the idea of residual learning on the basis of a VGG-19 model and adds a residual layer, as shown in FIG. 4. The partial front gather feature extraction part consists of a VGG-19 network and a residual error layer, wherein the VGG-19 network consists of 5 convolution modules CONV 1-CONV 5 and three full connection layers FC1-FC3, wherein CONV1 and CONV2 respectively comprise two convolution layers and one pooling layer, and CONV 3-CONV 5 respectively comprise three convolution layers and one pooling layer; the residual layer extracts the seismic image features of the first convolution module and feeds the seismic image features into the last convolution layer to be summed with the seismic image features of the convolution layer, the structure diagram of the part is shown in FIG. 4, and the calculation formula is shown in (8):
yR=F(x,{Wi})+x1 (8)
wherein the function of each convolution layer is set as F (x), and the feature vector from the first convolution layer is x1,WiIs the weight of the i-th layer convolution, yRAnd the feature vectors are output after residual learning.
In this way, the seismic features with more detailed information are combined with the seismic features with more deep dimension information in the last layer, so that the output seismic features have more geological detailed information. And then, obtaining quantitative evaluation values from the summed result through three full-connection layers FC1-FC3, and inputting the extracted quantitative evaluation values into a seismic image quality quantitative evaluation module so as to output a seismic image quality quantitative evaluation result.
Thirdly, quantitative evaluation prediction based on deep learning; after the training is completed, the new two-dimensional anterior gather data is blocked and then input into the residual neural network, so as to obtain a quantitative evaluation value corresponding to the block, as shown in fig. 5. The final output result is also the prediction result that we need.
It will be appreciated by those of ordinary skill in the art that the embodiments described herein are intended to assist the reader in understanding the principles of the invention and are to be construed as being without limitation to such specifically recited embodiments and examples. Those skilled in the art can make various other specific changes and combinations based on the teachings of the present invention without departing from the spirit of the invention, and these changes and combinations are within the scope of the invention.
Claims (4)
1. The method for intelligently and quantitatively evaluating the amplitude preservation of the forward gather based on deep learning is characterized by comprising the following steps of:
step one, constructing a quantitative evaluation training data set;
training a quantitative evaluation deep learning network;
and thirdly, quantitative evaluation prediction based on deep learning.
2. The method for intelligently and quantitatively evaluating the amplitude preservation of the forward gather based on the deep learning according to claim 1, wherein the step one specific implementation method is as follows: bias previous gather dataWherein n isxAnd nyRespectively representing the data number of xline and inline in a front-channel set;
dividing a data set into m along an xline direction and an inline directionxAnd myBlock, dxAnd dyRespectively representing the size of each partition:
each picture region having d after being partitionedx×dyOne pixel, the data after being blocked is recorded as Si,j,Wherein Si,jThe blocks representing ith row and j column are output as the quantitative evaluation value f of the corresponding blocki,jSolving through the step two;
to the front gatherCarrying out noise reduction treatment to obtain a new front-biased gatherPerforming the same blocking processing on the Q as the P to obtain blocking data
To Si,jAnd Ti,jCalculating SNR of the signal to noise ratioi,jAnd according to the signal-to-noise ratio SNRi,jThe magnitude of (d) discretizes the signal-to-noise ratio into five indices: the indexes of the dispersion are taken as quantitative evaluation labels of the data and recorded as the indexes of the dispersionWherein N represents the number of complete pictures for partitioning;
3. The method for intelligently and quantitatively evaluating the amplitude preservation of the forward gather based on the deep learning according to claim 1, wherein the second step is realized by the following specific method: obtaining nonlinear relation fitting of a regression model by adopting a convolutional neural network model based on residual learning, so as to obtain quantitative evaluation values of corresponding blocks from the data of the partial front gather after the blocking processing; the network consists of two major parts: a forward trace set feature extraction part and a seismic image quality quantitative evaluation output part;
the partial front gather feature extraction part consists of a VGG-19 network and a residual layer, wherein the VGG-19 network consists of 5 convolution modules CONV 1-CONV 5 and three full connection layers FC1-FC3, wherein CONV1 and CONV2 respectively comprise two convolution layers and one pooling layer, and CONV 3-CONV 5 respectively comprise three convolution layers and one pooling layer; the residual layer extracts the seismic image features of the first convolution module, sends the seismic image features to the last convolution layer to be summed with the seismic image features of the convolution layer, obtains quantitative evaluation values through three full-connection layers FC1-FC3 according to the summed result, and inputs the extracted quantitative evaluation values into the seismic image quality quantitative evaluation module, so that the seismic image quality quantitative evaluation result is output.
4. The method for intelligently and quantitatively evaluating the amplitude preservation of the forward gather based on the deep learning according to claim 1, wherein the third specific implementation method is as follows: and (4) dividing the new two-dimensional anterior gather data into blocks, and inputting the blocks into a residual neural network to obtain a quantitative evaluation value corresponding to the blocks.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110848280.1A CN113495291A (en) | 2021-07-27 | 2021-07-27 | Method for intelligently and quantitatively evaluating amplitude preservation of forward gather based on deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110848280.1A CN113495291A (en) | 2021-07-27 | 2021-07-27 | Method for intelligently and quantitatively evaluating amplitude preservation of forward gather based on deep learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113495291A true CN113495291A (en) | 2021-10-12 |
Family
ID=77996381
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110848280.1A Pending CN113495291A (en) | 2021-07-27 | 2021-07-27 | Method for intelligently and quantitatively evaluating amplitude preservation of forward gather based on deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113495291A (en) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106873033A (en) * | 2017-03-13 | 2017-06-20 | 中国海洋石油总公司 | A kind of method of many attribute earthquake information fusion FRACTURE PREDICTIONs based on neutral net |
CN110568483A (en) * | 2019-07-22 | 2019-12-13 | 中国石油化工股份有限公司 | Automatic evaluation method for seismic linear noise suppression effect based on convolutional neural network |
CN110580682A (en) * | 2019-09-16 | 2019-12-17 | 电子科技大学 | Countermeasure network seismic data super-resolution reconstruction method based on optimization generation |
CN110806602A (en) * | 2018-08-06 | 2020-02-18 | 中国石油化工股份有限公司 | Intelligent seismic data random noise suppression method based on deep learning |
CN111754403A (en) * | 2020-06-15 | 2020-10-09 | 南京邮电大学 | Image super-resolution reconstruction method based on residual learning |
CN112017289A (en) * | 2020-08-31 | 2020-12-01 | 电子科技大学 | Well-seismic combined initial lithology model construction method based on deep learning |
-
2021
- 2021-07-27 CN CN202110848280.1A patent/CN113495291A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106873033A (en) * | 2017-03-13 | 2017-06-20 | 中国海洋石油总公司 | A kind of method of many attribute earthquake information fusion FRACTURE PREDICTIONs based on neutral net |
CN110806602A (en) * | 2018-08-06 | 2020-02-18 | 中国石油化工股份有限公司 | Intelligent seismic data random noise suppression method based on deep learning |
CN110568483A (en) * | 2019-07-22 | 2019-12-13 | 中国石油化工股份有限公司 | Automatic evaluation method for seismic linear noise suppression effect based on convolutional neural network |
CN110580682A (en) * | 2019-09-16 | 2019-12-17 | 电子科技大学 | Countermeasure network seismic data super-resolution reconstruction method based on optimization generation |
CN111754403A (en) * | 2020-06-15 | 2020-10-09 | 南京邮电大学 | Image super-resolution reconstruction method based on residual learning |
CN112017289A (en) * | 2020-08-31 | 2020-12-01 | 电子科技大学 | Well-seismic combined initial lithology model construction method based on deep learning |
Non-Patent Citations (3)
Title |
---|
FENG QIAN 等: "Unsupervised seismic facies analysis via deep convolutional autoencoders:", 《GEOPHYSICS》, vol. 83, no. 3, pages 1 - 19 * |
吕小伟;钮学民;: "几项关键处理技术的保幅性分析及应用", 中国石油勘探, no. 02, pages 44 - 49 * |
李海山 等: "叠前随机噪声深度残差网络压制方法", 《石油地球物理勘探》, vol. 55, no. 3, pages 493 - 503 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO1995008781A1 (en) | Method for analising seismic traces using a statistical calibration technique for deduction of geological characteristics | |
CN107065013B (en) | A kind of interval velocity under earthquake scale determines method and device | |
CN111522063B (en) | Pre-stack high-resolution fluid factor inversion method based on frequency division joint inversion | |
US11668848B2 (en) | Method and system for seismic imaging using S-wave velocity models and machine learning | |
US20220237891A1 (en) | Method and system for image-based reservoir property estimation using machine learning | |
CN114966860A (en) | Seismic data denoising method based on convolutional neural network | |
WO2022159698A1 (en) | Method and system for image-based reservoir property estimation using machine learning | |
Gao et al. | Deep learning vertical resolution enhancement considering features of seismic data | |
CN112578475B (en) | Data mining-based dense reservoir dual dessert identification method | |
CN110568490B (en) | Identification method for high-speed stratum top thin reservoir | |
US12013508B2 (en) | Method and system for determining seismic processing parameters using machine learning | |
RU2126984C1 (en) | Method determining depth and speed parameters of medium and construction of its image by seismic data- prime system | |
CN113495291A (en) | Method for intelligently and quantitatively evaluating amplitude preservation of forward gather based on deep learning | |
US11768303B2 (en) | Automatic data enhancement for full waveform inversion in the midpoint-offset domain | |
CN114740528A (en) | Pre-stack multi-wave joint inversion method based on ultramicro Laplace block constraint | |
Lu | Deep learning realm for geophysics: Seismic acquisition, processing, interpretation, and inversion | |
CN110988991B (en) | Elastic parameter inversion method, device and system | |
CN117434592B (en) | Seismic data processing method and device and electronic equipment | |
Santos et al. | Automatic zero-phase wavelet estimation from seismic trace using a multilayer perceptron neural network: An application in a seismic well-tie | |
Li et al. | An efficient deep learning method for VSP wavefield separation: A DAS-VSP case | |
CN113589386B (en) | Block acoustic wave impedance inversion method, device and equipment based on contrast function | |
US11852768B2 (en) | Multimodal approach to target stratigraphic plays through seismic sequence stratigraphy, rock physics, seismic inversion and machine learning | |
Yeleskina et al. | Application of Deep Learning Methods in Seismic Image Enhancement | |
CN117805907A (en) | Earthquake phase driven hydrocarbon source rock prediction method, device and equipment | |
CN116047588A (en) | Non-stationary seismic data attenuation compensation method and system based on space-time neural network |
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 |