CN113296150A - High-dimensional closed-loop network seismic inversion method under logging constraint - Google Patents
High-dimensional closed-loop network seismic inversion method under logging constraint Download PDFInfo
- Publication number
- CN113296150A CN113296150A CN202110752889.9A CN202110752889A CN113296150A CN 113296150 A CN113296150 A CN 113296150A CN 202110752889 A CN202110752889 A CN 202110752889A CN 113296150 A CN113296150 A CN 113296150A
- Authority
- CN
- China
- Prior art keywords
- network
- data
- inversion
- seismic data
- dimensional
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000012549 training Methods 0.000 claims abstract description 30
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 9
- 238000005516 engineering process Methods 0.000 claims abstract description 4
- 238000011156 evaluation Methods 0.000 claims abstract description 3
- 230000001427 coherent effect Effects 0.000 claims description 7
- 238000013135 deep learning Methods 0.000 abstract description 3
- 230000006870 function Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 4
- 238000005457 optimization Methods 0.000 description 3
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002146 bilateral effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000005553 drilling Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000003786 synthesis reaction 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
-
- 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 invention discloses a high-dimensional closed-loop network seismic inversion method under logging constraint, belonging to the technical field of geophysical technology. The method specifically comprises the following steps: step 1: building a convolutional neural network; comprises a forward network and an inverse network; step 2: preparing training data; the method comprises logging wave impedance data, interpolation wave impedance data and synthetic seismic data; and step 3: training the network and carrying out fine adjustment: training the normal network and the inversion network, then applying one-dimensional logging data to the two-dimensional model and the three-dimensional model, and carrying out fine adjustment; and 4, step 4: prediction and evaluation: firstly, inverting seismic data; then forward modeling is carried out on the inversion result to obtain reconstructed seismic data; and finally, evaluating the effectiveness of the inversion result by using the reconstructed seismic data. According to the method, additional collection of input and reference images is not needed, good transverse continuity can be guaranteed, and the accuracy is higher than that of the traditional and other deep learning inversion methods.
Description
Technical Field
The invention relates to the technical field of geophysical, in particular to a high-dimensional closed-loop network seismic inversion method under logging constraint.
Background
In geophysical exploration, seismic inversion is used for reversely deducing subsurface physical parameters and reconstructing an internal structure of the earth through observed seismic data, wherein the reversely deduced parameters comprise speed, density, wave impedance and the like, and the parameters play an important role in positioning and evaluating oil and gas deposits and determining a drilling position. The usual inversion method is as forward actingAnd (5) performing inverse calculation. Depending on the basis of the inversion, these methods can be distinguished into model-based inversion and data-based inversion, such as trace-integral inversion[1]Recursive inversion[2]Generalized linear inversion[3]Etc., these inversion methods require hypothetical, simplified models that are generally linear, while the actual geophysical process is nonlinear, thereby introducing large inversion errors.
The deep neural network has strong feature expression and nonlinear fitting capability due to the fact that dozens or even hundreds of nonlinear layers can be stacked, and a mapping model can be learned from training data through a complex nonlinear structure[4]. Das et al[5](2018) Firstly, the method proposes to use a network consisting of two layers of one-dimensional CNNs to learn the mapping from the seismic data to the wave impedance, and obtains a better result than the conventional method, but the method needs a large number of label samples and has poor generalization and robustness. Wang et al propose the use of Cycle-GAN[6]And a one-dimensional closed-loop network[7]And seismic inversion is carried out, and the demand of label samples is reduced by introducing label-free data in training. In a subsequent improvement, Wang et al[8]On the basis of a one-dimensional closed-loop network, bilateral filtering constraint is added, so that the inversion effect is further improved, but the transverse continuity is still poor.
The invention utilizes the high-dimensional closed-loop network to directly invert two-dimensional and three-dimensional seismic data, obtains good transverse continuity, has stronger generalization and robustness, and has more reasonable inversion result.
Reference documents:
[1]Ferguson,R.J.,&Margrave,G.F.(1996).A simple algorithm for band-limited impedance inversion.CREWES annual
[2]Lindseth,R.O.(1979).Synthetic sonic logs—A process for stratigraphic interpretation.Geophysics,44(1),3-26.
[3]Cooke D A,Schneider W A.Generalized linear inversion of reflection seismic data[J].Geophysics,1983,48(6):665-676.
[4]LeCun Y,Bengio Y,Hinton G.Deep learning[J].nature,2015,521(7553):436.
[5]Das,V.,et al.(2018).Convolutional neural network for seismic impedance inversion.SEG Technical Program Expanded Abstracts 2018,Society of Exploration Geophysicists:2071-2075.
[6].Wang,Q.Ge,W.Lu,and X.Yan,“Seismic impedance inversion basedon cycle-consistent generative adversarial network,”inSEG Technical Program Expanded Abstracts 2019.Society of Exploration Geophysi-cists,2019,pp.2498–2502
[7]Y.Wang,Q.Ge,W.Lu,and X.Yan.[2020]Well-logging constrained seismic inversion based on closed-loop convolutional neural network IEEE Transactions on Geoscience and Remote Sensing,2020.
[8]Y.Wang,Q.Wang,W.Lu and H.Li,"Physics-Constrained Seismic Impedance Inversion Based on Deep Learning,"in IEEE Geoscience and Remote Sensing Letters,doi:10.1109/LGRS.2021.3072132.
[9]H.Huang,L.Lin,R.Tong,H.Hu,Q.Zhang,Y.Iwamoto,X.Han,Y.-W.Chen,and J.Wu,[2020].Unet 3+:A full-scale connected unet for medical image segmentation.in ICASSP2020-2020IEEE International Conference on Acoustics,Speech and Signal Processing(ICASSP).IEEE,2020,pp.1055–1059.
disclosure of Invention
The invention aims to provide a high-dimensional closed-loop network seismic inversion method under logging constraint, which is characterized by comprising the following steps:
step 1: building a convolutional neural network; the network structure is a closed loop structure and comprises a forward network and an inversion network, wherein the forward network and the inversion network are both U-net + + + models, and input data and output data of the network are both two-dimensional data or three-dimensional data;
step 2: preparing training data; the training data comprises log wave impedance data, interpolated wave impedance data and synthetic seismic data; the interpolated wave impedance data is generated by the constraint of log wave impedance data and actual seismic data, the synthetic seismic data is input data of network training, and the interpolated wave impedance data is calculated by a Robinson convolution model;
and step 3: training the network and carrying out fine adjustment: training the forward network and the inversion network by using the data in the step 2 to obtain a forward model and an inversion model, then applying one-dimensional logging data to a two-dimensional model and a three-dimensional model by using a Mask technology, and finely adjusting the two-dimensional model and the three-dimensional model to obtain the trained forward network and inversion network;
and 4, step 4: prediction and evaluation: firstly, inverting the seismic data by using the inversion network trained in the step 3 to obtain a wave impedance prediction result; then forward modeling is carried out on the inversion result by using the forward modeling network trained in the step 3 to obtain reconstructed seismic data; and finally, evaluating the effectiveness of the inversion result by using the reconstructed seismic data, wherein if the error value of the reconstructed seismic data and the input seismic data is smaller, the inversion result is more effective, and otherwise, the applicability of the inversion network on the input seismic data is poor.
The step 3 specifically comprises the following substeps:
step 31: according toUsing the label seismic data to optimize forward network weight, WF is forward network weight, S is label seismic data, AI is label wave impedance data,calculating a result for the forward network; according toOptimizing the inversion network weights using the tag wave impedance data, WB being the inversion network weights,calculating a result for the inversion network; while according to a closed-loop uniform loss function Synchronously optimizing the forward network and the inversion network; whereinFor closed-loop coherent loss of unlabeled seismic data, S*Representing the non-tagged seismic data and,closed-loop coherent loss for tagged seismic dataClosed-loop uniform loss of tag wave impedance data;
step 32: before calculating the network loss, multiplying the logging by the corresponding output of the network, and then calculating the loss reflection propagation for updating the network parameters; the loss function is:
wherein AI is*And representing the logging wave impedance, distinguishing whether a pixel point is on the logging or not by using a Mask, and setting the corresponding value of the pixel point on the logging to be 1, otherwise, setting the corresponding value to be 0.
The invention has the beneficial effects that:
according to the method, training samples are obtained through actual data synthesis, and input and reference images do not need to be collected additionally; the two-dimensional and three-dimensional inversion methods can ensure good transverse continuity; the logging data is used for fine tuning of the model, and the precision is higher than that of the traditional and other deep learning inversion methods.
Drawings
FIG. 1 is an overall flow diagram of the present invention;
FIG. 2 is a diagram of the forward and inverse network architecture of the present invention;
FIG. 3(a) is a schematic diagram of a two-dimensional Mask technique; FIG. 3(b) is a schematic diagram of a three-dimensional Mask technique;
FIG. 4(a) is actual seismic data; FIG. 4(b) is the inversion result of the method of the present invention; FIG. 4(c) is the one-dimensional closed-loop CNN inversion result; FIG. 4(d) is a comparison of waveforms on a blind well;
FIG. 5 is an inverted two-dimensional profile of the present invention on actual seismic data; FIG. 5(a) is a two-dimensional cross-sectional view of actual seismic data; FIG. 5(b) is an inverted cross-sectional view of a blind well at the well-plugging location of the present invention.
Detailed Description
The invention provides a high-dimensional closed-loop network seismic inversion method under logging constraint, and the invention is further explained by combining the attached drawings and specific embodiments.
Fig. 1 is an overall flow chart of the present invention, and the specific process is as follows:
(1) building convolutional neural networks
The network structure mainly comprises two parts: a forward network model and an inverse network model. Both the forward and inverse networks use a modified U-net + + + model, as shown in fig. 2. Mainly comprising an encoder and a decoder. The convolution layers are connected by flying leads, and the convolution kernels of the convolution layers are set to be two-dimensional in a two-dimensional method and three-dimensional in a three-dimensional method. Both the input data and the tag data are normalized to between-1 and 1 by dividing by the maximum of the absolute value. In the two-dimensional method, network input is two-dimensional data, and in the three-dimensional method, input and output are three-dimensional data.
(2) Preparing training data
The data preparation process is used to provide training data for the network training process. The method comprises actual seismic data, logging wave impedance data, interpolation wave impedance data generated by the logging data and the actual seismic data, and the interpolation wave impedance data serving as a label for training. The synthetic seismic data used for training are calculated by a corresponding interpolated wave impedance through a Robinson convolution model. The seismic data and the wave impedance data are both two-dimensional data in a two-dimensional method, and are both three-dimensional data in a three-dimensional method.
(3) Training network, fine tuning
In order to ensure the stability of the training process, the invention distinguishes two stages for training. The first stage is a training stage and the second stage is a fine tuning stage.
The first stage is as follows: forward and inverse network models are trained. The training data used in this stage is tag synthetic seismic data, corresponding tag interpolated wave impedance data, and non-tag seismic data.
According to a loss functionOptimizing forward network weights using tagged seismic data, where WFRepresenting forward network weights, S tag seismic data, AI tag wave impedance data,representing the results of the forward network computations.
According toOptimizing inversion network weights using tag wave impedance data, where WBThe inverse network weights are represented as a function of,representing the inversion network computation. Meanwhile, according to a combined closed-loop consistent loss function, the loss function is:synchronizing the optimization of the evolving and inverting networks, whereinFor closed-loop coherent loss of tag seismic data,for closed-loop coherent loss of tag wave impedance data,is a closed-loop coherent loss of unlabeled seismic data. Loss of each closed loop is consistentThe term expression is:
wherein S*Representing unlabeled seismic data. The optimization algorithm in the training process uses the Adam algorithm.
And a second stage: and applying one-dimensional logging data to a two-dimensional model and a three-dimensional model by using a Mask technology, and finely adjusting the model. The Mask technique is shown in fig. 3(a) and 3 (b). Is a mask of the same size as the network output, two-dimensional in the two-dimensional approach, and three-dimensional in the three-dimensional approach. The value of the logging tool is l only at the corresponding position of the logging tool, and the value of the logging tool is 0 at the rest positions. Before calculating the network loss, the corresponding outputs of the logging and the network are multiplied, and then the calculated loss reflection propagation is used for updating the network parameters. The loss function is:
wherein, the Mask distinguishes whether the pixel point is on the well logging, S*Representing unlabeled seismic data, AI*Representing the log wave impedance. For the pixel point on the logging, the corresponding value is l, otherwise, the corresponding value is 0. The optimization algorithm in the training process also uses the Adam algorithm.
(4) And (6) predicting and evaluating. And obtaining the final forward network weight and the final inversion network weight through the training process. And when in prediction, the whole actual seismic data S is inverted, and the inversion result can be further evaluated by using a forward model obtained by training. If the inversion result is accurate, the reconstructed seismic data should be more consistent with the input seismic data. The smaller the error value is, the more effective the inversion result is, otherwise, the poor applicability of the inversion network on the data is.
In order to verify the effectiveness and superiority of the method, the method provided by the invention is applied to actual seismic data. The experimental platform is Intel (R) core (TM) i9-9820X CPU @3.30GHz, 64GB RAM, GeForce RTX 2080 Ti. Fig. 4 and 5 show inversion results of the invention, and it can be seen that, compared with the one-dimensional closed-loop CNN, the blind well correlation coefficient of the invention is 0.91, and the one-dimensional closed-loop CNN blind well correlation coefficient is 0.45, and meanwhile, the continuity of the invention is better.
The present invention is not limited to the above embodiments, and any changes or substitutions that can be easily made by those skilled in the art within the technical scope of the present invention are also within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (2)
1. A high-dimensional closed-loop network seismic inversion method under logging constraint is characterized by comprising the following steps:
step 1: building a convolutional neural network; the network structure is a closed loop structure and comprises a forward network and an inversion network, wherein the forward network and the inversion network are both U-net + + + models, and input data and output data of the network are both two-dimensional data or three-dimensional data;
step 2: preparing training data; the training data comprises log wave impedance data, interpolated wave impedance data and synthetic seismic data; the interpolated wave impedance data is generated by the constraint of log wave impedance data and actual seismic data, the synthetic seismic data is input data of network training, and the interpolated wave impedance data is calculated by a Robinson convolution model;
and step 3: training the network and carrying out fine adjustment: training the forward network and the inversion network by using the data in the step 2 to obtain a forward model and an inversion model, then applying one-dimensional logging data to a two-dimensional model and a three-dimensional model by using a Mask technology, and finely adjusting the two-dimensional model and the three-dimensional model to obtain the trained forward network and inversion network;
and 4, step 4: prediction and evaluation: firstly, inverting the seismic data by using the inversion network trained in the step 3 to obtain a wave impedance prediction result; then forward modeling is carried out on the inversion result by using the forward modeling network trained in the step 3 to obtain reconstructed seismic data; and finally, evaluating the effectiveness of the inversion result by using the reconstructed seismic data, wherein if the error value of the reconstructed seismic data and the input seismic data is smaller, the inversion result is more effective, and otherwise, the applicability of the inversion network on the input seismic data is poor.
2. The well logging constrained high-dimensional closed-loop network seismic inversion method according to claim 1, wherein the step 3 specifically comprises the following sub-steps:
step 31: according toOptimizing forward network weights, W, using tagged seismic dataFTo forward network weights, S is tag seismic data, AI is tag wave impedance data,calculating a result for the forward network; according toOptimizing inversion network weights, W, using tag wave impedance dataBIn order to invert the weights of the network,calculating a result for the inversion network; while according to a closed-loop uniform loss function Synchronously optimizing the forward network and the inversion network; whereinFor closed-loop coherent loss of unlabeled seismic data, S*Representing the non-tagged seismic data and,closed-loop coherent loss for tagged seismic dataClosed-loop uniform loss of tag wave impedance data;
step 32: before calculating the network loss, multiplying the logging by the corresponding output of the network, and then calculating the loss reflection propagation for updating the network parameters; the loss function is:
wherein AI is*And representing the logging wave impedance, distinguishing whether a pixel point is on the logging or not by using a Mask, and setting the corresponding value of the pixel point on the logging to be 1, otherwise, setting the corresponding value to be 0.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110752889.9A CN113296150B (en) | 2021-07-02 | 2021-07-02 | High-dimensional closed-loop network seismic inversion method under logging constraint |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110752889.9A CN113296150B (en) | 2021-07-02 | 2021-07-02 | High-dimensional closed-loop network seismic inversion method under logging constraint |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113296150A true CN113296150A (en) | 2021-08-24 |
CN113296150B CN113296150B (en) | 2022-06-10 |
Family
ID=77330330
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110752889.9A Active CN113296150B (en) | 2021-07-02 | 2021-07-02 | High-dimensional closed-loop network seismic inversion method under logging constraint |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113296150B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114065639A (en) * | 2021-11-19 | 2022-02-18 | 江苏科技大学 | Closed-loop real-time inversion method for construction parameters of dredger |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103454685A (en) * | 2013-08-09 | 2013-12-18 | 中国石油天然气股份有限公司 | Method and device for predicting sand body thickness by utilizing logging constrained wave impedance inversion |
WO2016008105A1 (en) * | 2014-07-15 | 2016-01-21 | 杨顺伟 | Post-stack wave impedance inversion method based on cauchy distribution |
CN111983681A (en) * | 2020-08-31 | 2020-11-24 | 电子科技大学 | Seismic wave impedance inversion method based on countermeasure learning |
CN112213771A (en) * | 2019-07-10 | 2021-01-12 | 中国石油天然气股份有限公司 | Seismic wave impedance inversion method and device |
CN112733449A (en) * | 2021-01-11 | 2021-04-30 | 中国海洋大学 | CNN well-seismic joint inversion method, CNN well-seismic joint inversion system, CNN well-seismic joint inversion storage medium, CNN well-seismic joint inversion equipment and CNN well-seismic joint inversion application |
CN112862710A (en) * | 2021-01-28 | 2021-05-28 | 西南石油大学 | Electrical imaging logging image well wall restoration method based on Criminisi algorithm optimization |
CN112882123A (en) * | 2021-01-11 | 2021-06-01 | 中国海洋大学 | CNN well-seismic joint inversion method, system and application based on two-step method |
-
2021
- 2021-07-02 CN CN202110752889.9A patent/CN113296150B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103454685A (en) * | 2013-08-09 | 2013-12-18 | 中国石油天然气股份有限公司 | Method and device for predicting sand body thickness by utilizing logging constrained wave impedance inversion |
WO2016008105A1 (en) * | 2014-07-15 | 2016-01-21 | 杨顺伟 | Post-stack wave impedance inversion method based on cauchy distribution |
CN112213771A (en) * | 2019-07-10 | 2021-01-12 | 中国石油天然气股份有限公司 | Seismic wave impedance inversion method and device |
CN111983681A (en) * | 2020-08-31 | 2020-11-24 | 电子科技大学 | Seismic wave impedance inversion method based on countermeasure learning |
CN112733449A (en) * | 2021-01-11 | 2021-04-30 | 中国海洋大学 | CNN well-seismic joint inversion method, CNN well-seismic joint inversion system, CNN well-seismic joint inversion storage medium, CNN well-seismic joint inversion equipment and CNN well-seismic joint inversion application |
CN112882123A (en) * | 2021-01-11 | 2021-06-01 | 中国海洋大学 | CNN well-seismic joint inversion method, system and application based on two-step method |
CN112862710A (en) * | 2021-01-28 | 2021-05-28 | 西南石油大学 | Electrical imaging logging image well wall restoration method based on Criminisi algorithm optimization |
Non-Patent Citations (1)
Title |
---|
符力耘 等: "Robinson地震道褶积模型的修正和推广", 《1995年中国地球物理学会第十一届学术年会论文集》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114065639A (en) * | 2021-11-19 | 2022-02-18 | 江苏科技大学 | Closed-loop real-time inversion method for construction parameters of dredger |
CN114065639B (en) * | 2021-11-19 | 2024-05-31 | 江苏科技大学 | Closed-loop real-time inversion method for construction parameters of dredger |
Also Published As
Publication number | Publication date |
---|---|
CN113296150B (en) | 2022-06-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Wang et al. | Well-logging constrained seismic inversion based on closed-loop convolutional neural network | |
Li et al. | Deep-learning inversion of seismic data | |
Wang et al. | Physics-constrained seismic impedance inversion based on deep learning | |
Azevedo et al. | Generative adversarial network as a stochastic subsurface model reconstruction | |
CA3122488A1 (en) | Subsurface models with uncertainty quantification | |
CN112946749B (en) | Method for suppressing seismic multiples based on data augmentation training deep neural network | |
US11181653B2 (en) | Reservoir characterization utilizing ReSampled seismic data | |
CN112213771B (en) | Seismic wave impedance inversion method and device | |
CN110895348B (en) | Method, system and storage medium for extracting low-frequency information of seismic elastic impedance | |
Wang et al. | Seismic velocity inversion transformer | |
CN111948712A (en) | Pre-stack linear inversion method based on depth domain seismic record | |
CN116047583A (en) | Adaptive wave impedance inversion method and system based on depth convolution neural network | |
CN113687433A (en) | Bi-LSTM-based magnetotelluric signal denoising method and system | |
Jin et al. | Efficient progressive transfer learning for full-waveform inversion with extrapolated low-frequency reflection seismic data | |
Kumar et al. | Ensemble-based assimilation of nonlinearly related dynamic data in reservoir models exhibiting non-Gaussian characteristics | |
CN113296150B (en) | High-dimensional closed-loop network seismic inversion method under logging constraint | |
Dhara et al. | Physics-guided deep autoencoder to overcome the need for a starting model in full-waveform inversion | |
CN114545494B (en) | Unsupervised seismic data reconstruction method and device based on sparse constraint | |
CN117349798A (en) | Unbalanced regression-based mechanical drilling rate prediction method and system | |
Wu et al. | Adaptive feedback convolutional‐neural‐network‐based high‐resolution reflection‐waveform inversion | |
Han et al. | A modified generative adversarial nets integrated with stochastic approach for realizing super-resolution reservoir simulation | |
Li et al. | TransInver: 3D data-driven seismic inversion based on self-attention | |
Yoo et al. | Impedance inversion based on domain adaptation technique with reconstruction | |
Ravasi et al. | Striking a balance: Seismic inversion with model-and data-driven priors | |
Yang et al. | Reflection coefficients inversion based on the bidirectional long short-term memory 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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |