CN115639605A - Automatic high-resolution fault identification method and device based on deep learning - Google Patents
Automatic high-resolution fault identification method and device based on deep learning Download PDFInfo
- Publication number
- CN115639605A CN115639605A CN202211331692.9A CN202211331692A CN115639605A CN 115639605 A CN115639605 A CN 115639605A CN 202211331692 A CN202211331692 A CN 202211331692A CN 115639605 A CN115639605 A CN 115639605A
- Authority
- CN
- China
- Prior art keywords
- fault
- seismic
- quality
- model
- seismic image
- 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 112
- 238000013135 deep learning Methods 0.000 title claims abstract description 38
- 238000012549 training Methods 0.000 claims abstract description 128
- 238000012795 verification Methods 0.000 claims abstract description 58
- 230000006872 improvement Effects 0.000 claims abstract description 53
- 238000004088 simulation Methods 0.000 claims abstract description 22
- 230000008569 process Effects 0.000 claims description 27
- 238000002310 reflectometry Methods 0.000 claims description 24
- 238000005070 sampling Methods 0.000 claims description 22
- 230000006870 function Effects 0.000 claims description 20
- 238000004590 computer program Methods 0.000 claims description 8
- 238000003860 storage Methods 0.000 claims description 8
- 238000010200 validation analysis Methods 0.000 claims description 6
- 238000007796 conventional method Methods 0.000 abstract description 3
- 238000013528 artificial neural network Methods 0.000 description 25
- 238000010586 diagram Methods 0.000 description 16
- 230000004913 activation Effects 0.000 description 12
- 230000001427 coherent effect Effects 0.000 description 11
- 238000009826 distribution Methods 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 7
- 230000011218 segmentation Effects 0.000 description 7
- 230000015572 biosynthetic process Effects 0.000 description 6
- 238000004891 communication Methods 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 5
- 208000010392 Bone Fractures Diseases 0.000 description 4
- 206010017076 Fracture Diseases 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 4
- 238000000605 extraction Methods 0.000 description 4
- 230000004927 fusion Effects 0.000 description 4
- 238000005457 optimization Methods 0.000 description 4
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 3
- 230000003042 antagnostic effect Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000010606 normalization Methods 0.000 description 3
- 238000010521 absorption reaction Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000003708 edge detection Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000011176 pooling Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000013526 transfer learning Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 230000037373 wrinkle formation Effects 0.000 description 2
- 208000013201 Stress fracture Diseases 0.000 description 1
- 230000008485 antagonism Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000013434 data augmentation Methods 0.000 description 1
- 230000008021 deposition Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 230000036039 immunity Effects 0.000 description 1
- 229910052500 inorganic mineral Inorganic materials 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 239000011707 mineral Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
Images
Landscapes
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a method and a device for automatically identifying a high-resolution fault based on deep learning, which comprises the following steps: acquiring a seismic image to be identified; inputting the seismic image to be identified into a quality improvement model, and outputting an improved seismic image; the quality improvement model is obtained by training a first training set and a first verification set constructed on the basis of a plurality of sample low-quality seismic images and corresponding high-quality seismic image labels, and the first training set and the first verification set are generated through random parameter simulation; inputting the improved seismic image into a fault recognition model, and outputting a fault probability map; the fault recognition model is obtained by training a second training set and a second verification set which are constructed on the basis of a plurality of sample seismic images and corresponding fault probability icon labels. Compared with the conventional method, the method can obtain a cleaner and clearer fault probability map, and the predicted fault position is more accurate.
Description
Technical Field
The invention relates to the technical field of geophysical exploration, in particular to a method and a device for automatically identifying a high-resolution fault based on deep learning.
Background
Seismic images are an important geophysical datum that can provide continuous, extensive subsurface geological information. Experienced geophysical workers can infer the depositional age of the formation from seismic images, analyze depositional environments and tectonic movements, and aid in oil and gas exploration. Subsurface faults are an important geological formation that controls basin deposition, affects the migration of subsurface fluids, and the distribution of mineral deposits. Identifying subsurface faults from seismic images is a common practice, but because of the large scale and complex data of seismic exploration, manually interpreting faults is a laborious and highly subjective task. Therefore, with the development of computer technology, computer aided automatic fault identification is becoming more popular.
Patent application CN202010188216.0 discloses a three-attribute fusion fault identification method based on seismic data dominant frequency, which comprises the steps of firstly carrying out frequency division processing on full-frequency-band seismic data, selecting dominant frequency body data to manufacture coherence, dip angle and azimuth angle attribute bodies, and finally applying HIS (layered acquisition system) superposition display on the three attributes to obtain a fusion body for fault interpretation. According to the method, the fault identification capability is improved through frequency division and image fusion technologies, but a large amount of energy is consumed to select the dominant frequency body, a large amount of calculation is needed for subsequent attribute body calculation, and the subsequent interpretation precision is influenced by parameter selection.
Patent application CN201910854606.4 discloses a coherent enhancement fault identification method based on seismic analysis under the layer position constraint, the traditional coherent algorithm is improved by introducing means such as Hilbert transform and histogram equalization, the noise immunity is high, but the method is in the way of interfering with the limitation of coherent bodies, so that multiple solutions are easily generated, and small faults cannot be clearly depicted.
Patent application cn201710186136.x discloses an earthquake coherent body image fault automatic identification method based on an AdaBoost algorithm, which realizes automatic identification of earthquake coherent body earthquake images, improves fault identification accuracy, but needs to adjust the size of a data block when characteristic information is mined, and increases the space occupancy rate of data.
Patent application CN202110946552.1 discloses a seismic fault identification method based on deep learning semantic segmentation, which realizes rapid and accurate fault identification by means of strong fitting capability of a deep convolutional neural network, but the identification precision of the fault identification method is influenced by the signal-to-noise ratio of seismic data.
In 2016, huang Cheng et al adopt analysis technologies such as body attributes, along-layer structure attributes and absorption attenuation attributes to optimize seismic attribute bodies, effectively predict a fault system of a research area in stages and layers, and reduce the multi-solution in the fault identification process by combining and applying multiple methods. Zhong Weijun in 2017 and the like comprehensively utilize technical means such as coherent bodies, variance bodies, structural curvature bodies, multi-window inclination angle scanning bodies, structural guiding filtering, coherent energy gradient bodies, ant body tracking, edge detection and the like, form a fault identification technical process represented by an inclination angle guiding technology, and identify faults developing at different levels and different stages. Zhang Rui and others propose a frequency division ant tracking technology in 2017 by means of generalized S transformation and wavelet transformation analysis technologies, and the method can better suppress noise interference, identify faults more clearly and accurately, and identify deep small faults which are difficult to identify by conventional full-band data. Li Quanhe Tong Li in 2017 identifies complex fracture through seismic attribute optimization combination, firstly controls fracture distribution trend according to guide-based coherent body attributes, and then preferentially and finely describes spatial distribution characteristics of three-level fracture, four-level fracture and micro fracture by fusing seismic attribute bodies such as dip angle bodies, azimuth angle bodies and curvature bodies which highlight local details. Similarly, in 2017, sun Zhenyu and the like construct a fault layer identification model by taking seismic attributes as input of a support vector machine, the accuracy rate reaches 98%, the influence of human subjective factors is reduced, and the interpretation period is shortened. In 2020, liu Mojin and salted sea dragon perform image fusion on the coherent body and the amplitude, and the coherent body and the maximum positive curvature, fully utilize respective structural information, enhance the capability of identifying small faults, and improve the reliability of fault plane combination, plane spread characteristics and mutual relations. In 2022, hou Juntao et al applied the automatic fault extraction technique to the guided filtering based frequency division coherent, and obtained a more precise and accurate fault characterization result by suppressing the noise. Ding Changwei and the like propose to reduce the seismic attributes by using information value in 2022, and optimize XGboost parameters by combining an improved Bayes optimization algorithm so as to further improve the accuracy of minor fault seismic interpretation. Liu Naihao et al introduced the edge detection technique in deep learning in 2022 and optimized the network structure according to seismic data and fault characteristics, and proposed an improved HED network using seismic fault interpretation, which has higher accuracy and better continuity for fault prediction. Similarly, in 2022, zhang Zheng and the like combine the deep residual error network with the transfer learning and apply the combination to fault recognition, and on the basis of a synthetic data training model, a small amount of actual fault samples are used for transfer learning, so that the generalization capability of the network is enhanced, and the performance of fault recognition is improved.
The methods contribute to automatic identification of the seismic fault to a certain extent, but the methods lack precision analysis of fault identification and analysis of identification effect when the quality of multi-seismic data is low, for example, seismic data contains a large amount of random noise and the resolution of the seismic data is low. However, the seismic data collected in the field inevitably have noise interference and low resolution due to the absorption of the high-frequency components of the seismic waves by the underground medium, especially in deep and ultra-deep exploration. These reasons result in the poor universality of the currently proposed fault identification method.
Therefore, how to avoid the low accuracy and reliability of automatic fault identification at low seismic data quality is still an urgent problem to be solved by those skilled in the art.
Disclosure of Invention
The invention provides a high-resolution fault automatic identification method based on deep learning, which is used for solving the problems of low fault automatic identification precision and low reliability in the prior art when the seismic data quality is low.
The invention provides a high-resolution fault automatic identification method based on deep learning, which comprises the following steps:
acquiring a seismic image to be identified;
inputting the seismic image to be identified into a quality improvement model, and outputting an improved seismic image;
the quality improvement model is obtained by training a first training set and a first verification set constructed on the basis of a plurality of sample low-quality seismic images and corresponding high-quality seismic image labels, and the first training set and the first verification set are generated through random parameter simulation;
inputting the lifted seismic image into a fault recognition model and outputting a fault probability map;
the fault recognition model is obtained by training a second training set and a second verification set which are constructed on the basis of a plurality of sample seismic images and corresponding fault probability icon labels.
The invention provides an automatic high-resolution fault identification method based on deep learning, which further comprises the following steps:
inputting the seismic image to be recognized into the fault recognition model, and outputting a seismic image direct recognition result;
and comparing the fault probability map with the direct identification result of the seismic image.
According to the automatic high-resolution fault identification method based on deep learning provided by the invention, the first training set and the first verification set are generated through random parameter simulation, and the method specifically comprises the following steps:
generating a stratum reflectivity model according to a preset rule;
generating different types of stratum reflectivity models through random parameter control;
convolving any stratum reflectivity model with low-frequency Rake wavelets and adding seismic data with random noise to perform double down-sampling to obtain a sample low-quality seismic image;
and convolving the any stratum reflectivity model with the high-frequency Rake wavelet to obtain a corresponding high-quality seismic image label.
According to the automatic high-resolution fault identification method based on deep learning, the network structure of the quality improvement model in the training process is generated into a countermeasure network, the countermeasure network comprises a generator and a discriminator, the generator comprises a jump layer connection and a plurality of residual error connections, and the discriminator comprises a convolution layer and seven convolution blocks.
According to the automatic high-resolution fault identification method based on deep learning, provided by the invention, in the training process of the quality improvement model, the network parameter corresponding to the highest peak signal-to-noise ratio of the quality improvement model on the verification set is the optimal model parameter, wherein the peak signal-to-noise ratio is the similarity between the high-quality seismic image reconstructed by the generated countermeasure network and the real high-quality seismic image.
According to the automatic identification method of the high-resolution fault based on the deep learning, the model network structure used in the fault identification model training process is a U-net network, the U-net network comprises a coding branch and a decoding branch, the coding branch consists of four convolution operations and down sampling, and the decoding branch consists of four up sampling and convolution operations.
According to the automatic recognition method of the high-resolution fault based on the deep learning, the weight assigned to the pixel representing the fault in the loss function of the fault recognition model training process is larger than the weight assigned to the pixel representing the non-fault.
The invention also provides an automatic high-resolution fault recognition device based on deep learning, which comprises the following components:
the acquisition unit is used for acquiring a seismic image to be identified;
the quality improving unit is used for inputting the seismic image to be identified into a quality improving model and outputting an improved seismic image;
the quality improvement model is obtained by training a first training set and a first verification set constructed on the basis of a plurality of sample low-quality seismic images and corresponding high-quality seismic image labels, and the first training set and the first verification set are generated through random parameter simulation;
the fault recognition unit is used for inputting the lifted seismic image into a fault recognition model and outputting a fault probability map;
the fault recognition model is obtained by training a second training set and a second verification set which are constructed on the basis of a plurality of sample seismic images and corresponding fault probability icon labels.
The invention also provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the steps of the method for automatic identification of high resolution faults based on deep learning as described in any one of the above when executing the program.
The present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the method for automatic identification of high resolution faults based on deep learning as described in any one of the above.
According to the method and the device for automatically identifying the high-resolution fault based on the deep learning, the seismic image to be identified is obtained; inputting the seismic image to be recognized into a quality improvement model, and outputting an improved seismic image; the quality improvement model is obtained by training a first training set and a first verification set constructed on the basis of a plurality of sample low-quality seismic images and corresponding high-quality seismic image labels, and the first training set and the first verification set are generated through random parameter simulation; inputting the improved seismic image into a fault recognition model, and outputting a fault probability map; the fault recognition model is obtained by training a second training set and a second verification set which are constructed on the basis of a plurality of sample seismic images and corresponding fault probability icon labels. The method and the device realize the obtaining of a cleaner and clearer fault probability map, and the predicted fault position is more accurate.
Drawings
In order to more clearly illustrate the technical solutions of the present invention or the prior art, the drawings needed for the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.
FIG. 1 is a schematic flow chart of a method for automatically identifying a high-resolution fault based on deep learning according to the present invention;
FIG. 2 is a schematic structural diagram of an automatic high-resolution fault identification device based on deep learning according to the present invention;
FIG. 3 is a diagram of a neural network architecture for seismic data quality enhancement provided by the present invention;
FIG. 4 is a schematic diagram of a training process of a seismic data quality improvement neural network provided by the present invention;
FIG. 5 is a diagram of a seismic fault recognition neural network architecture provided by the present invention;
FIG. 6 is a schematic diagram of a training process of a fault recognition neural network provided by the present invention;
FIG. 7 is a schematic diagram of a workflow for intelligently identifying end-to-end high resolution faults provided by the present invention;
FIG. 8 is a schematic diagram illustrating the application of the present invention to real seismic data in the field and comparing with the conventional implementation flow;
fig. 9 is a schematic structural diagram of an electronic device provided by the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The problems of low fault automatic identification precision and low reliability in the case of low seismic data quality generally exist in the prior art. The automatic identification method of the high resolution fault based on deep learning of the present invention is described below with reference to fig. 1 to 9. Fig. 1 is a schematic flow chart of an automatic high resolution fault identification method based on deep learning according to the present invention, as shown in fig. 1, the method includes:
and step 110, acquiring a seismic image to be identified.
Specifically, the seismic image to be identified is obtained, and the image quality of the commonly obtained seismic data image is not high due to the limitation of the field acquisition environment, where the image quality refers to resolution and image signal to noise ratio, that is, the resolution of the seismic data image acquired in the field is low, the image signal to noise ratio is also low, and the noise ratio is high.
the quality improvement model is obtained by training a first training set and a first verification set constructed on the basis of a plurality of sample low-quality seismic images and corresponding high-quality seismic image labels, and the first training set and the first verification set are generated through random parameter simulation.
Specifically, a low-quality seismic image to be identified is firstly input into a quality improvement model for quality improvement, the output improved seismic image has higher resolution and larger signal-to-noise ratio relative to an image before quality improvement, the quality improvement model is obtained by training based on a large number of sample low-quality seismic images and corresponding high-quality seismic image labels, training data are divided into a training set and a verification set in the training process, after the training set is trained, the verification set is used for verification to select a model with the optimal network parameters, the low-quality seismic image and the high-quality seismic image need to be described, the low-quality seismic image and the high-quality seismic image refer to an image with the image resolution lower than a first threshold and the image resolution higher than a second threshold, and an image with the image signal-to-noise ratio lower than a third threshold and the image signal-to-noise ratio higher than a fourth threshold, and the first threshold, the second threshold, the third threshold and the fourth threshold are set based on an application scenario. In addition, the first training set and the first verification set in the embodiment of the invention are generated through random parameter simulation, manual marking is not needed, and the labor cost is saved.
the fault recognition model is obtained by training a second training set and a second verification set which are constructed on the basis of a plurality of sample seismic images and corresponding fault probability icon labels.
Specifically, after the quality of the seismic image to be recognized is improved, the seismic image to be recognized is input into a fault recognition model, the model outputs a corresponding fault probability chart, the fault recognition model is obtained by training based on a large number of sample seismic images and corresponding fault probability icon labels, training data are also divided into a training set and a verification set in the training process as well as the quality improvement model, and after the training set is trained, the verification set is used for verifying to select the model with the optimal network parameters.
The method provided by the invention comprises the steps of acquiring a seismic image to be identified; inputting the seismic image to be identified into a quality improvement model, and outputting an improved seismic image; the quality improvement model is obtained by training a first training set and a first verification set constructed on the basis of a plurality of sample low-quality seismic images and corresponding high-quality seismic image labels, and the first training set and the first verification set are generated through random parameter simulation; inputting the lifted seismic image into a fault recognition model and outputting a fault probability map; the fault recognition model is obtained by training a second training set and a second verification set which are constructed on the basis of a plurality of sample seismic images and corresponding fault probability icon labels. The method and the device realize the obtaining of a cleaner and clearer fault probability map, and the predicted fault position is more accurate.
Based on the above embodiment, the method further includes:
inputting the seismic image to be recognized into the fault recognition model, and outputting a seismic image direct recognition result;
and comparing the fault probability map with the direct identification result of the seismic image.
Specifically, after the seismic image to be identified enters the two layers of input and output of the quality improvement model and the fault identification model, the obtained fault probability map is compared with the seismic image direct identification result output by directly inputting the seismic image to be identified into the fault identification model, and the experimental result shows that the fault probability map obtained through the quality improvement model is clearer at the fault display position and has better fault display effect.
Based on the above embodiment, in the method, the generating of the first training set and the first verification set by random parameter simulation specifically includes:
generating a stratum reflectivity model according to a preset rule;
generating different types of stratum reflectivity models through random parameter control;
convolving any stratum reflectivity model with low-frequency Rake wavelets and adding seismic data with random noise to perform double down-sampling to obtain a sample low-quality seismic image;
and convolving the any stratum reflectivity model with the high-frequency Rake wavelet to obtain a corresponding high-quality seismic image label.
Specifically, training data is generated by stochastic parametric simulation. First, a reflectivity model with horizontal layers is randomly generated. Random vertical perturbations are then added to the horizontal layer model to model the formation dip. Next, gaussian perturbations are randomly added to the model to simulate the wrinkle formation. Next, the fault distribution in reality is simulated by adding volume vector field perturbations to the model. Different types of formation reflectivity models may be generated by random parametric control. And finally, convolving the reflectivity model with the low-frequency Rake wavelet and performing double down-sampling on the seismic data added with random noise to obtain a low-quality sample seismic image, and convolving the reflectivity model with the high-frequency Rake wavelet to obtain a corresponding high-quality seismic image label.
Based on the above embodiment, in the method, the network structure of the quality improvement model in the training process is a generation countermeasure network, and the generation countermeasure network includes a generator and a discriminator, where the generator includes a layer jump connection and a plurality of residual error connections, and the discriminator includes a convolutional layer and seven convolutional blocks.
Specifically, the construction of the neural network builds a network for improving the seismic data quality based on the current popular generation countermeasure network. The network structure comprises a generator and a discriminator. The network structure of the generator comprises one layer jump connection and a plurality of residual connections. First, shallow features of low quality seismic images are extracted by a convolutional layer and parameterized ReLU (PReLU) activation function. Five convolution blocks with residual concatenation are then used to extract the deep features. And then splicing and fusing the shallow and deep features through skip layer connection. The sub-pixel convolution layer is used to upsample the feature map by a factor of two to obtain an output image of the same size as the high quality seismic image. Finally, a high quality seismic image is predicted using a layer of convolutional layers and the Tanh activation function. The network structure of the discriminator is simpler than the generator, as it only needs to determine the probability that the input image is a true high quality seismic image. The discriminator network comprises one convolutional layer and seven convolutional blocks. Each volume block consists of one convolution layer, one batch normalization layer (BN) and one leakage ReLU activation function. Finally, two fully connected layers (Dense) and a Sigmoid activation function are performed to predict the probability that the input image is a true high quality seismic image. Antagonistic learning by the generator and discrimination helps the generator to generate more realistic high quality seismic images.
Based on the above embodiment, in the method, in the training process of the quality improvement model, the network parameter corresponding to the highest peak signal-to-noise ratio of the quality improvement model on the verification set is the optimal model parameter, wherein the peak signal-to-noise ratio is the similarity between the high-quality seismic image reconstructed by the generated countermeasure network and the real high-quality seismic image.
Specifically, by alternately optimizing the generator and the discriminator, the generator generates a high quality seismic image closer to the true one in order to fool the discriminator, and the discriminator enhances the ability to discriminate between the image generated by the generator and the true high quality image through optimization. The generator and arbiter network parameters are updated using Adam optimizers. The peak signal-to-noise ratio is used for representing the similarity between a reconstructed image and an original real high-quality seismic image, an optimal model is determined by checking the change of the peak signal-to-noise ratio of the model on a training set and a verification set along with the increase of iteration times, and network parameters corresponding to the highest peak signal-to-noise ratio of the model on the verification set are stored.
Based on the above embodiment, in this method, the model network structure used in the training process of the fault recognition model is a U-net network, the U-net network includes an encoding branch and a decoding branch, the encoding branch is composed of four convolution operations and a down-sampling operation, and the decoding branch is composed of four up-sampling and convolution operations.
Specifically, the fault recognition is regarded as an image semantic segmentation task, and each sampling point in the image to be recognized can be divided into two types, namely a fault type and a non-fault type. The method is characterized in that U-net commonly used in semantic segmentation is used as a fault recognition network, the U-net is divided into a coding branch and a decoding branch, the coding branch is composed of four times of convolution operation and down sampling, and the decoding branch is composed of four times of up sampling and convolution operation. And the layer jump connection fuses the feature graph after the up-sampling of the decoding branch and the feature graph after the convolution of the coding branch. Finally, a convolution layer with a convolution kernel size of 1 × 1 and a Sigmoid activation function are used to output a predicted fault probability map, which is the same size as the input image to be recognized.
Based on the above embodiment, in the method, the weight assigned to the pixel representing the fault in the loss function of the fault recognition model training process is greater than the weight assigned to the pixel representing the non-fault.
In particular, there is a strong sample imbalance in the tomographic semantic segmentation since the positive samples (tomographic pixels) are much less than the negative samples (non-tomographic pixels). To solve this problem, the pixel representing the slice is given a higher weight using the focal penalty, as follows:
where N denotes the number of pixels in the fault probability map, y i Binary labels representing true (0 for non-fault, 1 for fault), p i Representing the predicted probability of the fault identification network. And the weight of the loss value of the positive and negative samples is adjusted, and the weight of the learning difficulty of the samples is controlled. According to our experimental experience, in this example, letLet α =0.9, γ =2.
The automatic identification device for high-resolution fault based on deep learning provided by the invention is described below, and the automatic identification device for high-resolution fault based on deep learning described below and the automatic identification method for high-resolution fault based on deep learning described above can be correspondingly referred to each other.
Fig. 2 is a schematic structural diagram of the automatic high-resolution fault identification device based on deep learning according to the present invention, as shown in fig. 2, the device includes an acquisition unit 210, a quality improvement unit 220, and a fault identification unit 230, wherein,
the acquiring unit 210 is configured to acquire a seismic image to be identified;
the quality improving unit 220 is configured to input the seismic image to be identified into a quality improving model, and output an improved seismic image;
the quality improvement model is obtained by training a first training set and a first verification set constructed on the basis of a plurality of sample low-quality seismic images and corresponding high-quality seismic image labels, and the first training set and the first verification set are generated through random parameter simulation;
the fault recognition unit 230 is configured to input the lifted seismic image into a fault recognition model, and output a fault probability map;
the fault recognition model is obtained by training a second training set and a second verification set which are constructed on the basis of a plurality of sample seismic images and corresponding fault probability icon labels.
According to the device provided by the invention, the seismic image to be identified is obtained; inputting the seismic image to be recognized into a quality improvement model, and outputting an improved seismic image; the quality improvement model is obtained by training a first training set and a first verification set constructed on the basis of a plurality of sample low-quality seismic images and corresponding high-quality seismic image labels, and the first training set and the first verification set are generated through random parameter simulation; inputting the lifted seismic image into a fault recognition model and outputting a fault probability map; the fault recognition model is obtained by training a second training set and a second verification set which are constructed on the basis of a plurality of sample seismic images and corresponding fault probability icon labels. The method and the device realize the obtaining of a cleaner and clearer fault probability map, and the predicted fault position is more accurate.
Based on the above embodiment, the apparatus further includes a comparison unit, configured to:
inputting the seismic image to be recognized into the fault recognition model, and outputting a seismic image direct recognition result;
and comparing the fault probability map with the direct identification result of the seismic image.
Based on the above embodiment, in the apparatus, the generating of the first training set and the first verification set by random parameter simulation specifically includes:
generating a stratum reflectivity model according to a preset rule;
generating different types of stratum reflectivity models through random parameter control;
convolving any stratum reflectivity model with low-frequency Rake wavelets and performing double down-sampling on the seismic data added with random noise to obtain a sample low-quality seismic image;
and convolving any stratum reflectivity model with the high-frequency Rake wavelet to obtain a corresponding high-quality seismic image label.
Based on the above embodiment, in the apparatus, the network structure of the quality improvement model in the training process is a generation countermeasure network, and the generation countermeasure network includes a generator and a discriminator, where the generator includes a layer jump connection and a plurality of residual error connections, and the discriminator includes a convolutional layer and seven convolutional blocks.
Based on the above embodiment, in the device, in the training process of the quality improvement model, the network parameter corresponding to the highest peak signal-to-noise ratio of the quality improvement model on the verification set is the optimal model parameter, wherein the peak signal-to-noise ratio is the similarity between the high-quality seismic image reconstructed by the generated countermeasure network and the real high-quality seismic image.
Based on the above embodiment, in the apparatus, the model network structure used in the training process of the fault recognition model is a U-net network, the U-net network includes a coding branch and a decoding branch, the coding branch is composed of four convolution operations and a down-sampling operation, and the decoding branch is composed of four up-sampling and convolution operations.
Based on the above embodiment, in the apparatus, the weight assigned to the pixel representing the fault in the loss function of the fault recognition model training process is greater than the weight assigned to the pixel representing the non-fault.
Based on the above embodiment, the present invention provides a method for improving seismic image resolution and noise reduction by using a synthetic seismic data generation countermeasure network, comprising the following steps:
s1, constructing a training data set, constructing a seismic data quality improvement neural network, and training and verifying the neural network;
s2, constructing a training data set, building a fault recognition neural network, and training and verifying the neural network;
s3, establishing a high-resolution fault automatic identification end-to-end workflow based on the seismic data quality improvement neural network and the fault identification neural network;
s4, directly sending the field real seismic data into a fault recognition network for fault recognition, comparing the results of the fault recognition by the method provided by the text, and verifying the effectiveness of the method provided by the text.
Specifically, in S1, three stages can be divided. The first stage is the establishment of a training data set, a supervised learning deep learning method is used, and in order to establish the mapping relationship from low-quality seismic data to high-quality seismic data, a large amount of low-quality seismic data and their corresponding labels (high-quality seismic data) are required for training, however, it is very difficult to collect such data in reality. Thus, training data is generated by stochastic parametric simulation. First, a reflectivity model with horizontal layers is randomly generated. Random vertical perturbations are then added to the horizontal layer model to model the formation dip. Next, gaussian perturbations are randomly added to the model to simulate the wrinkle formation. Next, the fault distribution in reality is simulated by adding volume vector field perturbations to the model. Different types of formation reflectivity models may be generated by random parametric control. And finally, convolving the reflectivity model with the low-frequency Rake wavelet and performing double down-sampling on the seismic data added with the random noise to obtain low-quality seismic data, and convolving the reflectivity model with the high-frequency Rake wavelet to obtain corresponding high-quality seismic data. By combining different random parameters, a large and diverse number of pairs of seismic data can be generated. The low-quality seismic data are used as the input of a seismic data quality improving neural network, and the high-quality seismic data are used as labels. The training set and validation set are partitioned in proportion to 8:2, which completes the preparation of the training data. The second stage is the construction of a neural network, fig. 3 is a neural network structure diagram for improving the seismic data quality provided by the invention, and a network for improving the seismic data quality is constructed based on a currently popular generation countermeasure network (as shown in fig. 3). The network structure comprises a generator and a discriminator. As shown in fig. 3 (a), the network structure of the generator comprises one layer jump connection and a plurality of residual connections. First, shallow features of low quality seismic images are extracted by a convolutional layer and parameterized ReLU (PReLU) activation function. Then, five convolution blocks with residual concatenation (as shown in fig. 3 (c)) are used to extract the deep features. And then splicing and fusing the shallow and deep features through skip layer connection. The sub-pixel convolution layer is used to upsample the feature map by a factor of two to obtain an output image of the same size as a high quality seismic image. Finally, a high quality seismic image is predicted using a layer of convolutional layers and the Tanh activation function. As shown in fig. 3 (b), the network structure of the discriminator is simpler than the generator, because it only needs to determine the probability that the input image is a true high-quality seismic image. The discriminator network contains one convolutional layer and seven convolutional blocks (see fig. 3 (d)). Each volume block consists of one convolution layer, one batch normalization layer (BN) and one leakage ReLU activation function. Finally, two fully connected layers (Dense) and a Sigmoid activation function are performed to predict the probability that the input image is a true high quality seismic image. Antagonistic learning by the generator and discrimination helps the generator to generate more realistic high quality seismic images. The third stage is training and verification of the neural network, and the training of generating the antagonistic network is divided into training of a generator and training of a discriminator. The loss function defining the generator is the sum of the mean square error loss plus the perceptual loss and the countering loss.
The losses to the generator are defined as follows:
l G ＝l MSE +0.6×l VGG +0.1×l adv
where lMSE represents MSE loss at the pixel level, lVGG represents VGG loss, and ladv represents antagonism loss.
The lMSE calculates the mean square error of the generated image and the corresponding real image pixel by pixel, and the calculation formula is as follows:
in the formula, W and H represent the dimensions of the low-resolution noise seismic image ILR, and t represents the resolution improvement factor from ILR to IHR.
VGG networks have proven to have powerful feature extraction capabilities. And the quality of the produced image is improved by calculating the difference between the generated image and the real image in the feature extraction space of the VGG network. A VGG loss function is defined based on a pre-trained 16-layer VGG network (VGG 16). And calculating the Euclidean distance between the high-resolution noise-free seismic image generated by the generator and the characteristic graph obtained by the j-th convolution (after an activation function) of the corresponding real image IHR before the ith layer of the maximum pooling layer in the VGG16 network. The Euclidean distance is used as a VGG loss to measure the similarity of a reconstructed image and a real image in a VGG network characteristic representation space, and the calculation formula is as follows:
in the formula, phi i,j Represents the operation of extracting the feature map of the jth convolutional layer (after activation function) before the ith max pooling layer, W i,j And H i,j And the size of the feature map of each feature extraction layer of the VGG network is represented.
To fool the discriminator, the generator learns as much as possible the distribution of the true high resolution noise-free seismic image. Thus, ladv is defined according to the probability that the discriminator will recognize the reconstructed seismic image as a true high resolution, noise-free image:
where N represents the number of low resolution noisy samples, representing the probability that the seismic image reconstructed by the generator will be considered by the discriminator as a true high resolution noise-free seismic image.
The loss of the discriminator is only the penalty loss. By alternately optimizing the generator and the discriminator, the generator generates high quality seismic data closer to the true one in order to fool the discriminator, and the discriminator enhances the ability to discriminate between the image generated by the generator and the true high quality image through optimization. The generator and arbiter network parameters are updated using Adam optimizers. The peak signal-to-noise ratio is used for representing the similarity between a reconstructed image and an original image, fig. 4 is a schematic diagram of a training process of the seismic data quality improvement neural network provided by the invention, as shown in fig. 4, an optimal model is determined by checking the change of the peak signal-to-noise ratio of the model on a training set and a verification set along with the increase of iteration times, and network parameters corresponding to the highest peak signal-to-noise ratio of the model on the verification set are stored and used for subsequent prediction tasks of the model.
It should be noted that, in the present invention, the peak frequency range of the randomly given high seismic quality seismic data is 30 to 55HZ, and the peak frequency range of the low quality seismic data is 10 to 30HZ. While the signal-to-noise ratio for random noise in the given low quality seismic data ranges from 3 to 13. For the division ratio of the training data, the division is performed according to the criterion that 80% is used for training and 20% is used for verification. In other embodiments, other parameters may be used for the simulation of the data and the partitioning of the training data set.
In particular toIn S2, three stages may be provided. The first stage is the creation of a training dataset, prepared in proportion to 8:2 using the published dataset (Wu et al, 2020), and a validation dataset. And then, building a second-stage fault recognition neural network, wherein fault recognition is regarded as an image semantic segmentation task, and each sampling point in the seismic image can be divided into a fault type and a non-fault type. Fig. 5 is a diagram of a seismic fault recognition neural network structure provided by the present invention, which uses U-net commonly used in semantic segmentation as a fault recognition network, and the network structure is shown in fig. 5, where U-net is divided into a coding branch and a decoding branch, the coding branch is composed of four convolution operations and a down-sampling operation, and the decoding branch is composed of four up-sampling and convolution operations. And the layer jump connection fuses the feature graph after the up-sampling of the decoding branch and the feature graph after the convolution of the coding branch. Finally, a convolution layer with a convolution kernel size of 1 × 1 and Sigmoid activation function are used to output a predicted fault probability map, which is the same size as the input seismic image. The third stage is the training and validation of the fault recognition network, and there is a strong sample imbalance in fault semantic segmentation because positive samples (fault pixels) are much less than negative samples (non-fault pixels). To solve this problem, a loss function L of the fault identification network is defined using the focal loss, giving higher weight to the pixels representing the fault fault As follows:
where N denotes the number of pixels in the fault probability map, y i Binary labels representing true (0 for non-fault, 1 for fault), p i Representing the predicted probability of the fault identification network. And the weight of the loss value used for adjusting the positive and negative samples is used for controlling the weight of the learning difficulty of the samples. According to experimental experience, in the present invention, α =0.9 and γ =2 are set.
During training, input seismic data are subjected to normalization processing to eliminate the influence of data distribution difference, and data augmentation operations of image turning and rotation are performed to enhance the robustness of the model. The Adam optimizer is used to update the parameters of the network. Fig. 6 is a schematic diagram of a training process of the fault recognition neural network provided by the present invention, and the training process of the fault recognition neural network is as shown in fig. 6, and network parameters corresponding to the minimum loss of the model on the validation set are stored for the subsequent fault prediction task of the model. In other embodiments, other optimizers, training parameters, loss functions, and data enhancement approaches may be used.
Specifically, in S3, an end-to-end workflow of high-resolution fault identification of field seismic data is constructed. Fig. 7 is a schematic diagram of a workflow from end to end for intelligent high-resolution fault identification, the workflow is formed by connecting trained neural networks in S1 and S2 in series, an original seismic image (as shown in fig. 7 (a)) is sent to a data quality improving network trained in S1 to improve the resolution and suppress random noise, so as to obtain high-quality seismic data (as shown in fig. 7 (b)), and then the data with improved data quality is sent to a fault identification network in S2 to perform fault automatic identification, so as to obtain a high-resolution fault identification result (as shown in fig. 7 (c)). The workflow effectively integrates two tasks of seismic data quality improvement and fault identification, and the high-precision fault identification result can be directly obtained from the original low-frequency seismic image full of random noise by the provided method.
Specifically, in S4, compared with the conventional method, the original seismic image is directly sent to the fault identification network for fault identification, and the result of firstly improving the quality of the seismic image and then performing fault identification is compared with the proposed result of firstly performing fault identification to verify the effectiveness and the advancement of the proposed new method. Firstly, directly sending the collected field seismic data into a fault recognition network trained in S2 for fault recognition. And then, adopting the workflow provided in the S3 to firstly improve the data quality of the field seismic data and then carry out fault identification. And finally, comparing the fault identification results of the two modes. In the present invention, fig. 8 is a schematic diagram illustrating the application of the present invention to the field real seismic data and the comparison with the conventional implementation process, and the quality of the original seismic image (as shown in fig. 8 (a)) is first improved by using the seismic data quality improvement network trained in S1. Fig. 8 (b) is a seismic image after quality improvement, random noise is significantly suppressed compared to the original seismic image, and some high-frequency geological features, such as faults, thin layers and the like, are easier to observe visually. Then, the seismic image with improved quality is fed to the trained fault recognition network, and a high-resolution fault recognition result is obtained (as shown in fig. 8 (d)). Comparing the result with the result (as shown in fig. 8 (b)) of the conventional fault identification process (the original seismic image (as shown in fig. 8 (a)) is directly fed to the fault identification network), the fault identified by the proposed high-resolution fault identification method is finer and cleaner, and the identified noise is less in interference. In addition, when some adjacent faults are identified directly through the original image, the identification results are mixed together, and the faults are separated and well characterized through the identification results of the proposed method. More importantly, because the original seismic image has low resolution and contains a lot of random noise, a plurality of small-scale faults are not identified, and after the quality of seismic data is improved, the small-scale faults are well detected by a fault identification neural network. Comparing fig. 8 (d) and fig. 8 (b) further, the method identifies a fault with a sharper fault line compared with the conventional method, which provides more accurate fault location information and is of great help for subsequent accurate geological modeling work. These all demonstrate the superiority of the proposed new high-resolution fault identification method compared with the existing fault identification method, especially for low-quality seismic data under complex geological conditions.
The invention provides an end-to-end high-resolution fault automatic identification workflow based on a deep learning technology. The workflow comprises two neural networks, one for seismic image quality improvement and the other for seismic fault identification. Firstly, inputting an original seismic image into a trained seismic data quality improvement network to obtain a high-quality seismic image. And then inputting the data into a fault recognition network with training elements to obtain a high-resolution fault recognition result. To overcome this problem, a large amount of training data was obtained by stochastic parametric simulation techniques, due to the lack of a large amount of labeled seismic data for neural network training and validation in practice. Then, the method is applied to the field real seismic data, and the result shows that the method can obtain a high-precision fault identification result from the original seismic image without any manual marking data. Compared with the conventional fault identification method, the fault identified by the provided high-resolution fault identification method is finer and cleaner, and the identified noise interference is less. In addition, when some adjacent faults are identified directly through the original image, the identification results are mixed together, and the faults are separated and well characterized through the identification results of the proposed method. More importantly, because the original seismic image has low resolution and contains a lot of random noise, a plurality of small-scale faults are not identified, and after the quality of seismic data is improved, the small-scale faults are well detected by a fault identification neural network. The method can provide more accurate fault position information, and is of great help for subsequent accurate geological modeling work. These all demonstrate the superiority of the proposed new high-resolution fault identification method compared with the existing fault identification method, especially for low-quality seismic data under complex geological conditions.
Fig. 9 illustrates a physical structure diagram of an electronic device, and as shown in fig. 9, the electronic device may include: a processor (processor) 910, a communication Interface (Communications Interface) 920, a memory (memory) 930, and a communication bus 940, wherein the processor 910, the communication Interface 920, and the memory 930 are coupled for communication via the communication bus 940. Processor 910 may invoke logic instructions in memory 930 to perform a method for automatic identification of high resolution faults based on deep learning, the method comprising: acquiring a seismic image to be identified; inputting the seismic image to be identified into a quality improvement model, and outputting an improved seismic image; the quality improvement model is obtained by training a first training set and a first verification set constructed on the basis of a plurality of sample low-quality seismic images and corresponding high-quality seismic image labels, and the first training set and the first verification set are generated through random parameter simulation; inputting the lifted seismic image into a fault recognition model and outputting a fault probability map; the fault recognition model is obtained by training a second training set and a second verification set which are constructed on the basis of a plurality of sample seismic images and corresponding fault probability icon labels.
Furthermore, the logic instructions in the memory 930 may be implemented in the form of software functional units and stored in a computer readable storage medium when the software functional units are sold or used as independent products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product comprising a computer program stored on a non-transitory computer readable storage medium, the computer program comprising program instructions, which when executed by a computer, enable the computer to perform the method for automatic deep learning based high resolution fault identification provided by the above methods, the method comprising: acquiring a seismic image to be identified; inputting the seismic image to be identified into a quality improvement model, and outputting an improved seismic image; the quality improvement model is obtained by training a first training set and a first verification set constructed on the basis of a plurality of sample low-quality seismic images and corresponding high-quality seismic image labels, and the first training set and the first verification set are generated through random parameter simulation; inputting the lifted seismic image into a fault recognition model and outputting a fault probability map; the fault recognition model is obtained by training a second training set and a second verification set which are constructed on the basis of a plurality of sample seismic images and corresponding fault probability icon labels.
In yet another aspect, the present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which when executed by a processor is implemented to perform the method for automatic deep learning based high resolution fault identification provided above, the method comprising: acquiring a seismic image to be identified; inputting the seismic image to be recognized into a quality improvement model, and outputting an improved seismic image; the quality improvement model is obtained by training a first training set and a first verification set constructed on the basis of a plurality of sample low-quality seismic images and corresponding high-quality seismic image labels, and the first training set and the first verification set are generated through random parameter simulation; inputting the lifted seismic image into a fault recognition model and outputting a fault probability map; the fault recognition model is obtained by training a second training set and a second verification set which are constructed on the basis of a plurality of sample seismic images and corresponding fault probability icon labels.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A method for automatically identifying a high-resolution fault based on deep learning is characterized by comprising the following steps:
acquiring a seismic image to be identified;
inputting the seismic image to be identified into a quality improvement model, and outputting an improved seismic image;
the quality improvement model is obtained by training a first training set and a first verification set constructed on the basis of a plurality of sample low-quality seismic images and corresponding high-quality seismic image labels, and the first training set and the first verification set are generated through random parameter simulation;
inputting the lifted seismic image into a fault recognition model and outputting a fault probability map;
the fault recognition model is obtained by training a second training set and a second verification set which are constructed on the basis of a plurality of sample seismic images and corresponding fault probability icon labels.
2. The method of claim 1, further comprising:
inputting the seismic image to be recognized into the fault recognition model, and outputting a seismic image direct recognition result;
and comparing the fault probability map with the direct identification result of the seismic image.
3. The method for automatically identifying a high-resolution fault based on deep learning of claim 1, wherein the first training set and the first validation set are generated by stochastic parametric simulation, and specifically comprises:
generating a stratum reflectivity model according to a preset rule;
generating different types of stratum reflectivity models through random parameter control;
convolving any stratum reflectivity model with low-frequency Rake wavelets and adding seismic data with random noise to perform double down-sampling to obtain a sample low-quality seismic image;
and convolving the any stratum reflectivity model with the high-frequency Rake wavelet to obtain a corresponding high-quality seismic image label.
4. The method of claim 3, wherein the network structure of the quality improvement model in the training process is a generation countermeasure network, the generation countermeasure network comprises a generator and a discriminator, the generator comprises a layer jump connection and a plurality of residual connection, and the discriminator comprises a convolutional layer and seven convolutional blocks.
5. The method for automatically identifying the high-resolution fault based on the deep learning of claim 4, wherein in the training process of the quality improvement model, a network parameter corresponding to the highest peak signal-to-noise ratio of the quality improvement model on the verification set is an optimal model parameter, wherein the peak signal-to-noise ratio is a similarity between a generated confrontation network reconstruction high-quality seismic image and a real high-quality seismic image.
6. The method for automatically identifying high-resolution fault based on deep learning of claim 1, wherein a model network structure used in the training process of the fault identification model is a U-net network, the U-net network comprises a coding branch and a decoding branch, the coding branch is composed of four convolution operations and a down-sampling operation, and the decoding branch is composed of four up-sampling and convolution operations.
7. The method of claim 6, wherein the weight assigned to the pixels representing faults in the loss function of the fault recognition model training process is greater than the weight assigned to the pixels representing non-faults.
8. An apparatus for automatically identifying a high resolution fault based on deep learning, comprising:
the acquisition unit is used for acquiring a seismic image to be identified;
the quality improving unit is used for inputting the seismic image to be identified into a quality improving model and outputting an improved seismic image;
the quality improvement model is obtained by training a first training set and a first verification set constructed on the basis of a plurality of sample low-quality seismic images and corresponding high-quality seismic image labels, and the first training set and the first verification set are generated through random parameter simulation;
the fault recognition unit is used for inputting the lifted seismic image into a fault recognition model and outputting a fault probability map;
the fault recognition model is obtained by training a second training set and a second verification set which are constructed on the basis of a plurality of sample seismic images and corresponding fault probability icon labels.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor when executing the program implements the steps of the method for automatic recognition of high resolution faults based on deep learning according to any one of claims 1 to 7.
10. A non-transitory computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method for automatic identification of high resolution faults based on deep learning according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211331692.9A CN115639605B (en) | 2022-10-28 | 2022-10-28 | Automatic identification method and device for high-resolution fault based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211331692.9A CN115639605B (en) | 2022-10-28 | 2022-10-28 | Automatic identification method and device for high-resolution fault based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115639605A true CN115639605A (en) | 2023-01-24 |
| CN115639605B CN115639605B (en) | 2024-05-28 |
Family
ID=84946907
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211331692.9A Active CN115639605B (en) | 2022-10-28 | 2022-10-28 | Automatic identification method and device for high-resolution fault based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115639605B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115880505A (en) * | 2023-03-08 | 2023-03-31 | 山东科技大学 | Low-order fault intelligent identification method for target edge detection neural network |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110516740A (en) * | 2019-08-28 | 2019-11-29 | 电子科技大学 | A kind of fault recognizing method based on Unet++ convolutional neural networks |
| CN112102322A (en) * | 2020-09-16 | 2020-12-18 | 电子科技大学 | Fault identification method based on multi-mode U-Net |
| CN112241741A (en) * | 2020-08-25 | 2021-01-19 | 华中农业大学 | Self-adaptive image attribute editing model and method based on classified countermeasure network |
| CN112396558A (en) * | 2019-08-15 | 2021-02-23 | 株式会社理光 | Image processing method, image processing apparatus, and computer-readable storage medium |
| CN112435309A (en) * | 2020-12-07 | 2021-03-02 | 苏州深透智能科技有限公司 | Method for enhancing quality and resolution of CT image based on deep learning |
| CN112686822A (en) * | 2020-12-30 | 2021-04-20 | 成都信息工程大学 | Image completion method based on stack generation countermeasure network |
| CN113160575A (en) * | 2021-03-15 | 2021-07-23 | 超级视线科技有限公司 | Traffic violation detection method and system for non-motor vehicles and drivers |
| CN113538324A (en) * | 2020-09-18 | 2021-10-22 | 腾讯科技（深圳）有限公司 | Evaluation method, model training method, device, medium, and electronic apparatus |
| CN113902769A (en) * | 2021-08-18 | 2022-01-07 | 南方海洋科学与工程广东省实验室(广州) | Seismic fault identification method based on deep learning semantic segmentation |
| CN114707534A (en) * | 2022-03-07 | 2022-07-05 | 华北电力大学（保定） | Rolling bearing fault diagnosis method under small sample data |
| CN114898160A (en) * | 2022-06-02 | 2022-08-12 | 电子科技大学 | Intelligent fault identification method based on multiple tasks |
| CN115201902A (en) * | 2022-07-04 | 2022-10-18 | 中海石油(中国)有限公司 | Fault intelligent identification method and system based on deep learning |
-
2022
- 2022-10-28 CN CN202211331692.9A patent/CN115639605B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112396558A (en) * | 2019-08-15 | 2021-02-23 | 株式会社理光 | Image processing method, image processing apparatus, and computer-readable storage medium |
| CN110516740A (en) * | 2019-08-28 | 2019-11-29 | 电子科技大学 | A kind of fault recognizing method based on Unet++ convolutional neural networks |
| CN112241741A (en) * | 2020-08-25 | 2021-01-19 | 华中农业大学 | Self-adaptive image attribute editing model and method based on classified countermeasure network |
| CN112102322A (en) * | 2020-09-16 | 2020-12-18 | 电子科技大学 | Fault identification method based on multi-mode U-Net |
| CN113538324A (en) * | 2020-09-18 | 2021-10-22 | 腾讯科技（深圳）有限公司 | Evaluation method, model training method, device, medium, and electronic apparatus |
| CN112435309A (en) * | 2020-12-07 | 2021-03-02 | 苏州深透智能科技有限公司 | Method for enhancing quality and resolution of CT image based on deep learning |
| CN112686822A (en) * | 2020-12-30 | 2021-04-20 | 成都信息工程大学 | Image completion method based on stack generation countermeasure network |
| CN113160575A (en) * | 2021-03-15 | 2021-07-23 | 超级视线科技有限公司 | Traffic violation detection method and system for non-motor vehicles and drivers |
| CN113902769A (en) * | 2021-08-18 | 2022-01-07 | 南方海洋科学与工程广东省实验室(广州) | Seismic fault identification method based on deep learning semantic segmentation |
| CN114707534A (en) * | 2022-03-07 | 2022-07-05 | 华北电力大学（保定） | Rolling bearing fault diagnosis method under small sample data |
| CN114898160A (en) * | 2022-06-02 | 2022-08-12 | 电子科技大学 | Intelligent fault identification method based on multiple tasks |
| CN115201902A (en) * | 2022-07-04 | 2022-10-18 | 中海石油(中国)有限公司 | Fault intelligent identification method and system based on deep learning |
Non-Patent Citations (1)
| Title |
|---|
| 汤嘉立;柳益君;杜卓明;: "基于GEP多标记学习的图像超分辨率复原算法", 计算机应用研究, no. 06, 14 June 2017 (2017-06-14) * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115880505A (en) * | 2023-03-08 | 2023-03-31 | 山东科技大学 | Low-order fault intelligent identification method for target edge detection neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115639605B (en) | 2024-05-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Shi et al. | Development of subsurface geological cross-section from limited site-specific boreholes and prior geological knowledge using iterative convolution XGBoost | |
| CN109709603B (en) | Seismic horizon identification and tracking method and system | |
| CN107688201B (en) | RBM-based seismic prestack signal clustering method | |
| EP3894906A1 (en) | Training machine learning systems for seismic interpretation | |
| US10359523B2 (en) | Exploration and extraction method and system for hydrocarbons | |
| Azevedo et al. | Generative adversarial network as a stochastic subsurface model reconstruction | |
| Zhang et al. | U-net generative adversarial network for subsurface facies modeling | |
| CN112731522B (en) | Intelligent recognition method, device and equipment for seismic stratum and storage medium | |
| Xiao et al. | Conditioning of deep-learning surrogate models to image data with application to reservoir characterization | |
| CN111915618A (en) | Example segmentation algorithm and computing device based on peak response enhancement | |
| CN115639605B (en) | Automatic identification method and device for high-resolution fault based on deep learning | |
| Yang et al. | A multi-task learning method for relative geologic time, horizons, and faults with prior information and transformer | |
| US20240045089A1 (en) | Generating realistic synthetic seismic data items | |
| CN115690772A (en) | Gravity fault automatic identification method based on deep learning | |
| Dhara et al. | Seismic image registration using multiscale convolutional neural networks | |
| Udegbe et al. | Big Data Analytics for Seismic Fracture Identification, Using Amplitude-Based Statistics | |
| CN116068644B (en) | Method for improving resolution and noise reduction of seismic data by using generation countermeasure network | |
| Wang et al. | Semisupervised semantic segmentation for seismic interpretation | |
| Han et al. | Algorithm for Intelligent Recognition Low-Grade Seismic Faults Using Codec Target Edges | |
| Denisenko et al. | Automated geosteering while drilling using machine learning. case studies | |
| Roberts et al. | Joint stochastic constraint of a large data set from a salt dome | |
| Fan et al. | Automatic reconstruction of geological reservoir models based on conditioning data constraints and BicycleGAN | |
| CN114022632A (en) | Geological model rapid generation method for generating countermeasure network based on deep convolution | |
| CN115880455A (en) | Three-dimensional intelligent interpolation method based on deep learning | |
| Sun et al. | Enabling uncertainty quantification in a standard full-waveform inversion method using normalizing flows |
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 |