CN116559949A - Carbonate reservoir prediction method, system and equipment based on deep learning - Google Patents
Carbonate reservoir prediction method, system and equipment based on deep learning Download PDFInfo
- Publication number
- CN116559949A CN116559949A CN202310573765.3A CN202310573765A CN116559949A CN 116559949 A CN116559949 A CN 116559949A CN 202310573765 A CN202310573765 A CN 202310573765A CN 116559949 A CN116559949 A CN 116559949A
- Authority
- CN
- China
- Prior art keywords
- data
- seismic
- deep learning
- seismic attribute
- attribute data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 77
- BVKZGUZCCUSVTD-UHFFFAOYSA-L Carbonate Chemical compound [O-]C([O-])=O BVKZGUZCCUSVTD-UHFFFAOYSA-L 0.000 title claims abstract description 49
- 238000013135 deep learning Methods 0.000 title claims abstract description 35
- 238000013136 deep learning model Methods 0.000 claims abstract description 46
- 238000013528 artificial neural network Methods 0.000 claims abstract description 13
- 238000011176 pooling Methods 0.000 claims description 25
- 238000003860 storage Methods 0.000 claims description 25
- 238000005070 sampling Methods 0.000 claims description 14
- 238000010586 diagram Methods 0.000 claims description 12
- 238000013527 convolutional neural network Methods 0.000 claims description 11
- 238000005520 cutting process Methods 0.000 claims description 11
- 238000009826 distribution Methods 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 11
- 238000012549 training Methods 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 7
- 230000004913 activation Effects 0.000 claims description 6
- 238000012544 monitoring process Methods 0.000 claims description 6
- 239000013598 vector Substances 0.000 claims description 6
- 239000000284 extract Substances 0.000 claims description 4
- 230000011218 segmentation Effects 0.000 claims description 3
- 230000008859 change Effects 0.000 description 5
- 238000012545 processing Methods 0.000 description 4
- 238000010276 construction Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 238000010606 normalization Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000005553 drilling Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000000306 recurrent effect Effects 0.000 description 2
- 239000011435 rock Substances 0.000 description 2
- 239000013049 sediment Substances 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000009933 burial Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 230000001351 cycling effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000035699 permeability Effects 0.000 description 1
- 230000000704 physical effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
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
- G01V1/28—Processing seismic data, e.g. for interpretation or for event detection
- G01V1/30—Analysis
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A10/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE at coastal zones; at river basins
- Y02A10/40—Controlling or monitoring, e.g. of flood or hurricane; Forecasting, e.g. risk assessment or mapping
Landscapes
- Engineering & Computer Science (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (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 relates to a carbonate reservoir prediction method, a carbonate reservoir prediction system and carbonate reservoir prediction equipment based on deep learning, which relate to the technical field of geophysical exploration and specifically comprise the steps of acquiring seismic attribute data; inputting the seismic attribute data into a trained deep learning model to generate wave impedance data; the deep learning model is a residual neural network comprising an input layer, a hidden layer and an output layer. The carbonate reservoir prediction method provided by the invention can effectively learn the complex nonlinear relation between the seismic attribute data and the logging wave impedance data, accurately and effectively predict the reservoir of the three-dimensional area covered by the seismic data body, and has important application value.
Description
Technical Field
The invention relates to the technical field of geophysical exploration, in particular to a carbonate reservoir prediction method, a carbonate reservoir prediction system, carbonate reservoir prediction equipment and a computer readable storage medium based on deep learning.
Background
Carbonate reservoirs often have the characteristics of poor reservoir continuity, low matrix porosity, strong permeability heterogeneity, deep burial depth and the like, and the exploration and development difficulty is extremely high. Accurate distribution prediction of reservoirs has an important role in assessing geologic distribution. The effective and accurate reservoir prediction data can provide priori geological information guidance in the actual drilling process, and the drilling success rate is greatly improved. Therefore, the method can accurately predict the complex carbonate reservoir and has important practical application value for formulating a reasonable oilfield development scheme.
At present, logging technology is a common technical means for monitoring a carbonate reservoir, but because of the huge consumption of manpower and material resources, logging data has serious spatial sparsity, which brings great challenges to accurate reservoir prediction. On the other hand, the seismic data volume has the advantages of wide coverage area, easy acquisition and the like, however, interpretation of the seismic data often has multiple solutions, which brings a certain uncertainty to reservoir prediction. Based on the above problems, it is necessary to combine the logging and seismic data volumes to take advantage of both, and to design an effective complex carbonate reservoir prediction method for accurate reservoir prediction of carbonate.
Disclosure of Invention
The invention creatively provides a complex carbonate reservoir prediction method based on deep learning, which comprises the following steps of
Acquiring seismic attribute data;
inputting the seismic attribute data into a trained deep learning model to generate wave impedance data;
the deep learning model is a residual neural network comprising an input layer, a hidden layer and an output layer.
Further, the training method of the deep learning model comprises the following steps:
acquiring a seismic attribute dataset of a seismic trace;
Acquiring a logging wave impedance data set of a logging position;
deducing the seismic attribute data of the well logging position to generate a seismic attribute data set and a wave impedance data set corresponding to the well logging position;
and inputting the seismic attribute data set corresponding to the logging position into a deep learning model, generating predicted wave impedance data, generating a loss function based on a prediction result, and optimizing the deep learning model to obtain a trained deep learning model.
Further, the method may further comprise selecting a seismic attribute in the seismic attribute dataset, optionally comprising one or more of the following: time, seismic velocity, seismic amplitude, dominant frequency, instantaneous phase, relative amplitude, root mean square amplitude, 14Hz, 30Hz, and/or 50Hz seismic waves;
optionally, the seismic attribute is a seismic attribute weighted by a weight factor, and the weight factor is calculated by λ= (1+v), where λ is the weight factor and v is the seismic velocity, and the time, the seismic velocity, and other seismic attributes weighted by the weight factor are input into a deep learning model.
Further, a three-dimensional interpolation technology is adopted to deduce seismic attribute data of the logging position; optionally, the three-dimensional interpolation process includes: and selecting K adjacent seismic channel attribute values according to the relative position relation between the seismic channels and the well logging by taking the plane x and the plane y as coordinate axes, calculating the seismic attribute of the well logging position by using a spatial interpolation method, selecting L well logging wave impedance values adjacent to the seismic attribute data point by taking the time t as the coordinate axes, and acquiring the wave impedance data of the seismic attribute data point by using a one-dimensional cubic interpolation method, wherein K, L is a natural integer.
Further, the method further comprises the step of carrying out data cutting on the seismic attribute data set corresponding to the logging position to obtain a seismic attribute data set with fixed dimension; preferably, firstly, two parameters are set, wherein the first parameter is the length of data cutting, namely the height corresponding to the convolutional neural network; and the second parameter is the number of steps of backward movement of the data starting point after each data cutting, the seismic attribute data corresponding to the logging position is cut for the first time according to the first parameter, the data corresponding to the height of the convolutional neural network is formed, the pointer is moved backward according to the second parameter, and the second data segmentation is carried out.
Further, the hidden layer is composed of a first residual error network module, a channel attention module, a second residual error network module, a space attention module and a third residual error network module, and the seismic attribute data are sequentially processed by the hidden layer modules and then output predicted wave impedance data; optionally, the channel attention module extracts the feature vector with the length of R in the feature map processed by the first residual error network module by using maximum pooling and average pooling respectively, compresses the spatial dimension of the feature map by using the same shared network, and generates a channel attention map after summing element by element, wherein R is a natural number integer;
Optionally, the spatial attention module uses a power pooling method to integrate the feature distribution of each pooling window with the input feature map processed by the second residual error network module, and then obtains the spatial attention map through a one-dimensional convolution module with an output channel R and an activation function;
optionally, the second residual network module comprises 1-3 residual network modules connected in series; the third residual network module comprises 1-2 residual network modules connected in series.
Further, the prediction method further includes: reading the seismic attribute data according to an SEG-Y storage format, converting the read data into a decimal format, inputting a trained deep learning model, and generating wave impedance data;
optionally, the reading the seismic attribute data according to the SEG-Y storage format includes: firstly, for each seismic channel, reading the space coordinate information and the seismic attribute data length of a seismic monitoring point from the first N bytes; then, skipping the first N bytes, and reading the seismic attribute data of the current seismic channel according to the acquired data length; according to the format of the seismic attribute data, adaptively calculating the number of bytes occupied by the seismic attribute data in the seismic channel;
Optionally, the method further comprises: converting the read data into decimal format, normalizing the data, inputting the normalized data into a trained deep learning model, and generating wave impedance data;
optionally, the method further comprises reversely storing the generated wave impedance data;
preferably, the generated wave impedance data are respectively formatted and calculated according to the data format, the sampling interval and the sampling point number information of the acquired seismic attribute data, and then are reversely stored to generate a reservoir prediction data body of the SEG-Y readable by commercial software.
The invention aims to disclose a carbonate reservoir prediction system based on deep learning, which comprises a computer program, wherein the computer program realizes the carbonate reservoir prediction method based on deep learning when being executed by a processor.
The invention aims to disclose a carbonate reservoir prediction system based on deep learning, which comprises the following components:
an acquisition unit for acquiring seismic attribute data;
the generating unit is used for inputting the seismic attribute data into a trained deep learning model and generating wave impedance data; the deep learning model is a residual neural network comprising an input layer, a hidden layer and an output layer.
Further, the system also comprises a reading unit for reading the seismic attribute data according to an SEG-Y storage format and converting the read data into a decimal format.
Further, the system further comprises a seismic attribute selection unit for selecting a seismic attribute in the seismic attribute dataset, optionally, the seismic attribute comprising one or more of the following: time, seismic velocity, seismic amplitude, dominant frequency, instantaneous phase, relative amplitude, root mean square amplitude, 14Hz, 30Hz, and/or 50Hz seismic waves.
Further, the system also comprises a storage unit for reversely storing the generated wave impedance data; preferably, the generated wave impedance data are respectively formatted and calculated according to the data format, the sampling interval and the sampling point number information of the acquired seismic attribute data, and then are reversely stored to generate a reservoir prediction data body of the SEG-Y readable by commercial software.
The invention aims to disclose a carbonate reservoir prediction system based on deep learning, which comprises the following components:
an acquisition unit for acquiring seismic attribute data;
the reading unit is used for reading the seismic attribute data according to an SEG-Y storage format and converting the read data into a decimal format;
The seismic attribute selecting unit is used for selecting the seismic attributes in the seismic attribute data set to obtain selected seismic attribute data;
the generating unit is used for inputting the selected seismic attribute data into a trained deep learning model to generate wave impedance data; the deep learning model is a residual neural network comprising an input layer, a hidden layer and an output layer;
and the storage unit is used for reversely storing the generated wave impedance data.
The object of the present invention is to disclose a carbonate reservoir prediction apparatus based on deep learning, said apparatus comprising a memory and a processor,
the memory is used for storing program instructions; the processor is used for calling program instructions, and when the program instructions are executed, the carbonate reservoir prediction method based on deep learning is executed.
The object of the present invention is to disclose a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the above-mentioned method for deep learning based carbonate reservoir prediction.
The invention has the beneficial effects that:
(1) The invention provides a full-automatic accurate complex carbonate reservoir prediction method, which can effectively learn complex nonlinear relations between seismic attribute data (various) and logging wave impedance data, accurately and effectively predict a reservoir in a three-dimensional area covered by a seismic data body without human interaction and parameter setting;
(2) The complex carbonate reservoir prediction method provided by the invention can rapidly predict the reservoir of the target work area, provides accurate reservoir prediction data bodies for business analysis software, is convenient for subsequent detailed data reading, processing and deep analysis, and can effectively overcome the defects of small logging data quantity, sparse distribution and the like;
(3) In order to solve the dislocation problem of the space and time of the logging position and the seismic channel, the method interpolates the three-dimensional data formed by plane space and time to the seismic attribute data, so that the seismic attribute data and the wave impedance data form a one-to-one correspondence in the space dimension;
(4) In order to ensure the prediction precision of the model, the method screens the seismic attribute on one hand and introduces a weight factor on the other hand so as to solve the problem that the change of the seismic wave velocity along with the logging time can effectively reflect the change of the geological layer components;
(5) In order to ensure the prediction precision of the model, a channel attention module and a space attention module are respectively introduced at the front end and the rear end in the residual error network model, the channel attention module is constructed to give weight to the attributes of different channels, the weight value of the high-correlation seismic attribute in the whole prediction network is increased, and the space attention module is constructed to give weight to the characteristics of different time (depths) so as to effectively distinguish the components of different geological layers and improve the prediction of wave impedance.
Drawings
FIG. 1 is a flow chart of a carbonate reservoir prediction method based on deep learning provided by an embodiment of the invention;
FIG. 2 is a schematic diagram of a carbonate reservoir prediction apparatus based on deep learning according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a carbonate reservoir prediction system based on deep learning provided by an embodiment of the present invention;
FIG. 4 is a schematic view of spatial interpolation of seismic attributes of a logging neighborhood provided by an embodiment of the present invention;
FIG. 5 is a schematic illustration of time domain interpolation of log data to (interpolated) seismic attribute data provided by an embodiment of the invention;
FIG. 6 is a schematic diagram of data slicing of a recurrent convolutional neural network according to an embodiment of the present invention;
FIG. 7 is a schematic diagram of a recurrent convolutional neural network module provided by an embodiment of the present invention;
FIG. 8 is a schematic diagram of a channel attention module according to an embodiment of the present invention;
FIG. 9 is a schematic diagram of a multi-scale spatial attention module according to an embodiment of the present invention.
Detailed Description
In order to enable those skilled in the art to better understand the present invention, the following description will make clear and complete descriptions of the technical solutions according to the embodiments of the present invention with reference to the accompanying drawings.
In some of the flows described in the specification and claims of the present invention and in the foregoing figures, a plurality of operations occurring in a particular order are included, but it should be understood that the operations may be performed out of order or performed in parallel, with the order of operations such as 101, 102, etc., being merely used to distinguish between the various operations, the order of the operations themselves not representing any order of execution. In addition, the flows may include more or fewer operations, and the operations may be performed sequentially or in parallel. It should be noted that, the descriptions of "first" and "second" herein are used to distinguish different messages, devices, modules, etc., and do not represent a sequence, and are not limited to the "first" and the "second" being different types.
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments according to the invention without any creative effort, are within the protection scope of the invention.
Fig. 1 is a flowchart of a carbonate reservoir prediction method based on deep learning according to an embodiment of the present invention, specifically, the method includes the following steps:
101: acquiring seismic attribute data;
seismic attributes refer to parameters obtained from raw seismic amplitude data through a series of mathematical transformations that effectively reflect geometric, statistical, kinematic, or kinetic characteristics of the seismic wave.
In one embodiment, after acquiring the seismic attribute data, a seismic attribute in the seismic attribute data set is selected, optionally including one or more of the following: time, seismic velocity, seismic amplitude, dominant frequency, instantaneous phase, relative amplitude, root mean square amplitude, 14Hz, 30Hz, and/or 50Hz seismic waves. To ensure the prediction accuracy of the model, we first choose as input to the model a relatively varying seismic attribute comprising a series of attribute parameters of seismic wave of seismic amplitude, dominant frequency, instantaneous phase, relative amplitude, root mean square amplitude, 14Hz, 30Hz and 50 Hz. In consideration of the increase of monitoring time (depth), the geological layer composition changes, so that the seismic wave velocity caused by the change presents a regular distribution, and the obtained wave impedance of well logging and the absolute change of time, seismic wave velocity and the like have a certain close relation. Therefore, we also take as input of the model the time, seismic velocity, etc. seismic attributes.
In one embodiment, the seismic attribute is a seismic attribute weighted by a weighting factor calculated as λ= (1+v), where λ is the weighting factor and V is the seismic velocity, the time, seismic velocity, and other seismic attributes weighted by the weighting factor are input into a deep learning model. The seismic attribute weighted by the weight factors can be used as a model input to effectively improve the sensitivity to the change of the logging wave impedance.
In one embodiment, after normalization of all seismic attributes, the time, seismic velocity, and weighted seismic attributes (including: seismic amplitude, dominant frequency, instantaneous phase, relative amplitude, root mean square amplitude, 14Hz, 30Hz, and/or 50 Hz) are combined to form 11 channels of input data as a data set of the model.
In one embodiment, the prediction method further comprises: reading the obtained seismic attribute data according to an SEG-Y storage format, converting the read data into a decimal format, inputting a trained deep learning model, and generating wave impedance data; specifically, reading the obtained seismic attribute data according to the SEG-Y storage format comprises: firstly, for each seismic channel, reading the space coordinate information and the seismic attribute data length of a seismic monitoring point from the first N bytes; then, skipping the first N bytes, and reading the seismic attribute data of the current seismic channel according to the acquired data length; according to the format of the seismic attribute data, adaptively calculating the number of bytes occupied by the seismic attribute data in the seismic channel; optionally, the method further comprises: and converting the read data into a decimal format, carrying out normalization processing on the data, and inputting the normalized data into a trained deep learning model to generate wave impedance data.
In a specific embodiment, the reading is performed on the acquired seismic attribute data according to the SEG-Y storage format, and the read data is converted into a decimal format, and a specific reading mode is described in detail as follows: (1) And opening the seismic attribute data body in the SEG-Y format, reading the EBCDIC file header corresponding to the first 3200 bytes from the zero pointer, and obtaining the description information of the seismic attribute data body. (2) And reading the following 400 bytes, and acquiring information such as the format, the sampling point number, the sampling interval and the like of the seismic attribute data. (3) reading the seismic trace data. First, for each seismic trace, the spatial coordinate information and the seismic attribute data length of the seismic monitoring point are read from the first 240 bytes. Then, the first 240 bytes are skipped, and the seismic attribute data of the current seismic trace are read according to the acquired data length. At this time, the number of bytes occupied by the seismic attribute data in the seismic trace is adaptively calculated according to the format of the seismic attribute data. (4) And converting the read seismic attribute data (generally hexadecimal) into decimal format, and carrying out normalization processing on the data to finally generate the data which can be read by the deep regression neural network model.
102: inputting the seismic attribute data into a trained deep learning model to generate wave impedance data; the deep learning model is a residual neural network comprising an input layer, a hidden layer and an output layer.
In the residual neural network, a residual connection (shortcut) is added in each basic unit (block), so that the characteristics of the bottom layer are directly transmitted to the high layer, and the problem of reduced output accuracy caused by network degradation can be avoided or reduced by the residual neural network.
In one embodiment, the training method of the deep learning model includes: acquiring a seismic attribute dataset of a seismic trace; acquiring a logging wave impedance data set of a logging position; deducing the seismic attribute data of the well logging position to generate a seismic attribute data set and a wave impedance data set corresponding to the well logging position; and inputting the seismic attribute data set corresponding to the logging position into a deep learning model, generating predicted wave impedance data, generating a loss function based on a prediction result, and optimizing the deep learning model to obtain a trained deep learning model.
In one embodiment, three-dimensional interpolation techniques are used to derive seismic attribute data for the well log locations; optionally, the three-dimensional interpolation process includes: and selecting K adjacent seismic channel attribute values according to the relative position relation between the seismic channels and the well logging by taking the plane x and the plane y as coordinate axes, calculating the seismic attribute of the well logging position by using a spatial interpolation method, selecting L well logging wave impedance values adjacent to the seismic attribute data point by taking the time t as the coordinate axes, and acquiring the wave impedance data of the seismic attribute data point by using a one-dimensional cubic interpolation method, wherein K, L is a natural integer. In order to accurately predict a reservoir of carbonate rock, three-dimensional data consisting of plane space and time are needed to be interpolated on seismic attribute data, so that the problem of dislocation of the space and time of logging positions and seismic channels is solved, and a one-to-one correspondence relationship between the seismic attribute data and wave impedance data in time and space dimensions is realized.
In one embodiment, as shown in FIG. 4, for a particular spatial log location, the 10 seismic traces nearest to it are found, and the seismic attributes of the log location are calculated using spatial interpolation. In the interpolation process, the plane x and the plane y are taken as coordinate axes, 10 adjacent seismic channel attribute values are selected according to the relative position relation between the seismic channel and the well logging, and the seismic attribute data of the well logging position is obtained by adopting a bicubic interpolation method.
In a specific embodiment, as shown in fig. 5, taking time t as a coordinate axis, considering that the logging wave impedance data has higher resolution, selecting 3 logging wave impedance values close to the seismic attribute data point, and obtaining the wave impedance value of the seismic attribute data point by using a one-dimensional cubic interpolation method.
In one embodiment, the method further comprises performing data cutting on the seismic attribute data set corresponding to the logging position to obtain a seismic attribute data set with a fixed dimension, and inputting the seismic attribute data set into a model; preferably, firstly, two parameters are set, wherein the first parameter is the length of data cutting, namely the height corresponding to the convolutional neural network; and the second parameter is the number of steps of backward movement of the data starting point after each data cutting, the seismic attribute data corresponding to the logging position is cut for the first time according to the first parameter, the data corresponding to the height of the convolutional neural network is formed, the pointer is moved backward according to the second parameter, and the second data segmentation is carried out. In a specific embodiment, considering that the data length of each log is often different in the training data, meanwhile, since the input data of the convolutional neural network model must have a fixed dimension, that is, a fixed value of height, width and channel number (h×w×11), it is necessary to cut the training data to generate a training data set with uniform length, so that the length of each data in the data set matches the dimension of the residual neural network. As shown in fig. 6, the specific logic of the data cut is: firstly, setting two parameters, wherein the first parameter is the Length Data Length of Data cutting, namely the height Data length=h corresponding to the convolutional neural network; the second is the number of steps Step by which the data start point moves backward after each data cut, and is set to step=1 by default. For example, multi-seismic attribute data (multi-channel one-dimensional image) has dimensions 628×1×11, which is first cut to form first data having dimensions h× 1×11, the data including the first H points. And sequentially cycling, moving the pointer backwards according to the parameter Step, dividing the data for the second time, wherein the data length is H, and moving the starting point backwards by one bit. And so on until the last set of data of length H is obtained. At this time, for seismic attribute data having dimensions (628×1×11), when h=400, 628-400+1=229 sets of data having dimensions 400×1×11 are generated in total.
In one embodiment, the deep learning model is a residual neural network including an input layer, a hidden layer and an output layer, wherein the hidden layer is composed of a first residual network module, a channel attention module, a second residual network module, a spatial attention module and a third residual network module, and the seismic attribute data are sequentially processed by the modules of the hidden layer and then output predicted wave impedance data; optionally, the channel attention module firstly compresses the space dimension of the feature map of the input seismic attribute data, and the space information can be compressed to the channel descriptor through global average pooling and maximum pooling, so that network parameters are reduced, and the effect of preventing overfitting is achieved; sending the obtained channel descriptors to two fully connected networks to obtain an attention weight matrix, and multiplying the attention weight matrix with an original image to obtain an attention characteristic image after calibration; preferably, the channel attention module extracts the feature vector with the length of 1 in the feature map processed by the first residual error network module by using maximum pooling and average pooling respectively, compresses the space dimension of the feature map by using the same shared network, and generates a channel attention map after element-by-element summation; optionally, the spatial attention module applies average pooling and maximum pooling operations to the input feature map processed by the second residual network module along the channel axis, then connects them to generate an effective feature descriptor, sends the feature descriptor into a convolution network to convolve, and the obtained feature map obtains a final spatial attention feature map through an activation function; preferably, the spatial attention module integrates the characteristic distribution of each pooling window by adopting a power pooling method on the input characteristic graph processed by the second residual error network module, and then obtains the spatial attention diagram through a one-dimensional convolution module with an output channel of 1 and an activation function; optionally, the second residual network module comprises 1-3 residual network modules connected in series; the third residual network module comprises 1-2 residual network modules connected in series.
In one embodiment, the deep learning model is a residual neural network including an input layer, a hidden layer and an output layer, where the hidden layer is formed by a first residual network module, a channel attention module, a fourth residual network module, a fifth residual network module, a sixth residual network module, a spatial attention module, a seventh residual network module and an eighth residual network module, and the seismic attribute data is processed by the modules of the hidden layer in sequence, and then predicted wave impedance data is output. In a specific embodiment, as shown in fig. 7, the specific structure of the deep learning model network is as follows: (a) The input of the network is a one-dimensional image of the cut multichannel (i.e. the combined multi-attribute seismic data), the dimension is 400 multiplied by 1 multiplied by 11, and the hidden layer of the network is composed of six residual network modules. According to the data transmission direction, the output channel size of the residual error module sequentially changes from 11 to 32 to 64 to 32 to 16 to 1. The output characteristic length of the residual error module is changed in sequence from 400 to 200 to 100 to 200 to 400. (c) Near the input end, considering that the channels of the input data represent different seismic attributes, and the seismic attributes have different degrees of influence on wave impedance prediction, a channel attention module is constructed to assign weights to the attributes of the different channels, and the weight value of the high-correlation seismic attributes in the whole prediction network is increased, as shown in fig. 8. In the construction of the channel attention module, feature vectors with the length of 1 are extracted by using maximum pooling and average pooling respectively, and the spatial dimension of the input feature map is compressed by using the same shared network, and the channel attention map is generated after element-by-element summation. (d) Near the output end, the wave impedance and the logging time (depth) are closely related, and the space attention module is constructed to give weights to the characteristics of different times (depths) so as to effectively distinguish the components of different geological layers and improve the prediction of the wave impedance, as shown in fig. 9. In the construction of the spatial attention module, a power pooling method is adopted to accumulate the characteristic distribution of each pooling window. For more random superposition of different attribute combinations, a power pooling function with core sizes of 3, 5, 7 and 9 is adopted, the obtained output features are respectively connected with a one-dimensional convolution module with an output channel of 4, then all the output features are connected in series to form feature vectors of 16 channels, and the final space attention diagram is obtained through the one-dimensional convolution module with the output channel of 1 and a sigmoid activation function. After the input features pass through the channel and the spatial attention module, weight influence is only generated on the feature values, and the dimension of the features is not changed. (e) The final output of the model is a single-channel one-dimensional image with the same length as the input image, and the dimension of the output data is 400×1, namely the predicted seismic channel wave impedance data.
In one embodiment, the method further comprises storing the generated wave impedance data in reverse; preferably, the generated wave impedance data are respectively formatted and calculated according to the data format, the sampling interval and the sampling point number information of the acquired seismic attribute data, and then are reversely stored to generate the reservoir prediction data of the SEG-Y readable by commercial software.
In one particular embodiment, the SEG-Y format of the reservoir prediction data is stored in reverse: during the storage process, the data related information contained in the 3200 byte EBCDIC file header and 400 bytes of SEG-Y is read and copied from the input SEG-Y file; then for each seismic trace, 240 bytes of seismic trace information are first read and copied from the input SEG-Y file. And then, carrying out formatting calculation on wave impedance data obtained by predicting the regression convolutional neural network model according to the data format, the sampling interval and the sampling point number information of the input data, and further carrying out reverse storage. And finally, generating a reservoir prediction data body of the SEG-Y which can be read by commercial software, and effectively assisting further geological deep analysis.
In a specific embodiment, during the training process of the deep learning model, firstly, a seismic attribute data body is read, a plurality of seismic attributes are selected, and seismic channel attribute data of a logging position is extracted through a three-dimensional interpolation technology. And secondly, superposing and combining the extracted various seismic attribute data into a multi-channel one-dimensional image, and simultaneously, regarding the logging wave impedance data as a single-channel one-dimensional image, wherein the lengths of the two images are the same. And then training a deep learning model to learn the nonlinear relation between the two images. In the using stage, aiming at each seismic channel in a seismic data volume, combining various seismic attribute data on the seismic channel into a multi-channel one-dimensional image, and outputting the wave impedance distribution of the seismic channel by using the model as the input of a trained deep learning model. By applying the method, wave impedance prediction can be performed on all seismic channels in the target work area. Finally, the predicted wave impedance data is used for generating a data volume in a SEG-Y format which can be read by commercial software and is used for further effectively analyzing and evaluating the geological distribution of the carbonate rock.
In a specific embodiment, during training of the deep learning model, firstly, seismic attribute data are read, and a plurality of seismic attributes are selected, wherein the seismic attributes comprise one or more of the following: time, seismic velocity and other seismic attributes weighted by weighting factors (seismic amplitude, dominant frequency, instantaneous phase, relative amplitude, root mean square amplitude, 14Hz, 30Hz and/or 50Hz seismic waves), and extracting seismic trace attribute data of the logging location by a three-dimensional interpolation technique; secondly, data cutting is carried out on the data of the logging position, and a seismic attribute data set and a wave impedance data set corresponding to the logging position with fixed dimension are obtained; and training a deep learning model to generate wave impedance data, and obtaining a trained deep learning model based on the generated wave impedance data optimization model.
Fig. 2 is a schematic diagram of a deep learning based carbonate reservoir prediction apparatus of the present invention, the apparatus comprising a memory and a processor,
The memory is used for storing program instructions; the processor is used for calling program instructions, and when the program instructions are executed, the carbonate reservoir prediction method based on deep learning is executed.
The invention aims to disclose a carbonate reservoir prediction system based on deep learning, which comprises a computer program, wherein the computer program realizes the carbonate reservoir prediction method based on deep learning when being executed by a processor.
The invention aims to disclose a carbonate reservoir prediction system based on deep learning, which comprises the following components:
an acquisition unit for acquiring seismic attribute data;
the generating unit is used for inputting the seismic attribute data into a trained deep learning model and generating wave impedance data; the deep learning model is a residual neural network comprising an input layer, a hidden layer and an output layer.
Further, the system also comprises a reading unit for reading the seismic attribute data according to an SEG-Y storage format and converting the read data into a decimal format.
Further, the system further comprises a seismic attribute selection unit for selecting a seismic attribute in the seismic attribute dataset, optionally, the seismic attribute comprising one or more of the following: time, seismic velocity, seismic amplitude, dominant frequency, instantaneous phase, relative amplitude, root mean square amplitude, 14Hz, 30Hz, and/or 50Hz seismic waves.
Further, the system also comprises a storage unit for reversely storing the generated wave impedance data; preferably, the generated wave impedance data are respectively formatted and calculated according to the data format, the sampling interval and the sampling point number information of the acquired seismic attribute data, and then are reversely stored to generate a reservoir prediction data body of the SEG-Y readable by commercial software.
FIG. 3 is a schematic diagram of a carbonate reservoir prediction system based on deep learning according to the present invention, specifically comprising:
an acquisition unit 301 for acquiring seismic attribute data;
the reading unit 302 is configured to read the seismic attribute data according to an SEG-Y storage format, and convert the read data into a decimal format;
a seismic attribute selection unit 303, configured to select a seismic attribute in the seismic attribute data set, and obtain selected seismic attribute data; optionally, the seismic attribute includes one or more of the following: time, seismic velocity, seismic amplitude, dominant frequency, instantaneous phase, relative amplitude, root mean square amplitude, 14Hz, 30Hz, and/or 50Hz seismic waves;
a generating unit 304, configured to input the selected seismic attribute data into a trained deep learning model, and generate wave impedance data; the deep learning model is a residual neural network comprising an input layer, a hidden layer and an output layer; optionally, the hidden layer is formed by a first residual error network module, a channel attention module, a second residual error network module, a space attention module and a third residual error network module, and the seismic attribute data are sequentially processed by the hidden layer modules and then output predicted wave impedance data; preferably, the channel attention module extracts the feature vector with the length of 1 in the feature map processed by the first residual error network module by using maximum pooling and average pooling respectively, compresses the space dimension of the feature map by using the same shared network, and generates a channel attention map after element-by-element summation; preferably, the spatial attention module integrates the characteristic distribution of each pooling window by adopting a power pooling method on the input characteristic graph processed by the second residual error network module, and then obtains the spatial attention diagram through a one-dimensional convolution module with an output channel of 1 and an activation function;
A storage unit 305 for reversely storing the generated wave impedance data.
The object of the present invention is to disclose a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the above-mentioned method for deep learning based carbonate reservoir prediction.
Definition:
seismic attribute data are typically parameters related to the geometry, kinematics, dynamics, etc. of the seismic wave. By studying these parameters, the characteristics of the structure, lithology, fluid and the like of the subsurface medium of the exploration area can be obtained, and further the reservoir information of the oil and gas can be deduced. The seismic attributes can be divided into geometric attributes and physical attributes, wherein the geometric attributes mainly refer to reflection characteristics and the like and are used for construction explanation, layer sequence division and seismic phase research; physical properties refer primarily to amplitude, frequency, phase, etc. for lithology and reservoir characterization interpretation. Four types of post-stack seismic attributes can also be divided: time, frequency, amplitude, and decay properties. In addition, there is a functional classification by seismic attribute. Common are: such as root mean square Amplitude RMS Amplitude: identifying amplitude anomalies or describing layer sequences; seismic anomalies in the chase strata, such as amplitude anomalies caused by delta, river and gas-bearing sandstone, distinguish integrated sediments, hillock sediments, etc.
The results of the verification of the present verification embodiment show that assigning an inherent weight to an indication may moderately improve the performance of the present method relative to the default settings.
It will be clear to those skilled in the art that, for convenience and brevity of description, specific working procedures of the above-described systems, apparatuses and units may refer to corresponding procedures in the foregoing method embodiments, which are not repeated herein.
In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods may be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of the units is merely a logical function division, and there may be additional divisions when actually implemented, e.g., multiple units or components may be combined or integrated into another system, or some features may be omitted or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or units, which may be in electrical, mechanical or other form.
The units described as separate units may or may not be physically separate, and units shown 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 units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.
Those of ordinary skill in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be implemented by a program to instruct related hardware, the program may be stored in a computer readable storage medium, and the storage medium may include: read Only Memory (ROM), random access Memory (RAM, random Access Memory), magnetic or optical disk, and the like.
Those of ordinary skill in the art will appreciate that all or a portion of the steps in implementing the methods of the above embodiments may be implemented by a program to instruct related hardware, where the program may be stored in a computer readable storage medium, where the storage medium may be a read only memory, a magnetic disk or optical disk, etc.
While the foregoing describes a computer device provided by the present invention in detail, those skilled in the art will appreciate that the foregoing description is not meant to limit the invention thereto, as long as the scope of the invention is defined by the claims appended hereto.
Claims (10)
1. Carbonate reservoir prediction method based on deep learning, comprising
Acquiring seismic attribute data;
inputting the seismic attribute data into a trained deep learning model to generate wave impedance data;
the deep learning model is a residual neural network comprising an input layer, a hidden layer and an output layer.
2. The deep learning based carbonate reservoir prediction method of claim 1, wherein the training method of the deep learning model comprises:
acquiring a seismic attribute dataset of a seismic trace;
acquiring a logging wave impedance data set of a logging position;
deducing the seismic attribute data of the well logging position to generate a seismic attribute data set and a wave impedance data set corresponding to the well logging position;
and inputting the seismic attribute data set corresponding to the logging position into a deep learning model, generating predicted wave impedance data, generating a loss function based on a prediction result, and optimizing the deep learning model to obtain a trained deep learning model.
3. The deep learning based carbonate reservoir prediction method of claim 1 or 2, further comprising selecting a seismic attribute in a seismic attribute dataset, optionally comprising one or more of the following: time, seismic velocity, seismic amplitude, dominant frequency, instantaneous phase, relative amplitude, root mean square amplitude, 14Hz, 30Hz, and/or 50Hz seismic waves;
Optionally, the seismic attribute is a seismic attribute weighted by a weight factor, and the weight factor is calculated by λ= (1+v), where λ is the weight factor and v is the seismic velocity, and the time, the seismic velocity, and other seismic attributes weighted by the weight factor are input into a deep learning model.
4. The deep learning based carbonate reservoir prediction method of claim 2, wherein the seismic attribute data for the logging location is deduced using a three-dimensional interpolation technique; optionally, the three-dimensional interpolation process includes: and selecting K adjacent seismic channel attribute values according to the relative position relation between the seismic channels and the well logging by taking the plane x and the plane y as coordinate axes, calculating the seismic attribute of the well logging position by using a spatial interpolation method, selecting L well logging wave impedance values adjacent to the seismic attribute data point by taking the time t as the coordinate axes, and acquiring the wave impedance data of the seismic attribute data point by using a one-dimensional cubic interpolation method, wherein K, L is a natural integer.
5. The deep learning based carbonate reservoir prediction method of claim 2, further comprising data cutting the seismic attribute dataset corresponding to the logging location to obtain a fixed-dimension seismic attribute dataset; preferably, firstly, two parameters are set, wherein the first parameter is the length of data cutting, namely the height corresponding to the convolutional neural network; and the second parameter is the number of steps of backward movement of the data starting point after each data cutting, the seismic attribute data corresponding to the logging position is cut for the first time according to the first parameter, the data corresponding to the height of the convolutional neural network is formed, the pointer is moved backward according to the second parameter, and the second data segmentation is carried out.
6. The carbonate reservoir prediction method based on deep learning according to claim 1, wherein the hidden layer is composed of a first residual network module, a channel attention module, a second residual network module, a spatial attention module and a third residual network module, and the seismic attribute data are sequentially processed by the hidden layer modules and then output predicted wave impedance data; optionally, the channel attention module extracts the feature vector with the length of R in the feature map processed by the first residual error network module by using maximum pooling and average pooling respectively, compresses the spatial dimension of the feature map by using the same shared network, and generates a channel attention map after summing element by element;
optionally, the spatial attention module accumulates the characteristic distribution of each pooling window by adopting a power pooling method to the input characteristic graph processed by the second residual error network module, and then obtains a spatial attention diagram by using a one-dimensional convolution module and an activation function with an output channel of R, wherein R is a natural integer;
optionally, the second residual network module comprises 1-3 residual network modules connected in series; the third residual network module comprises 1-2 residual network modules connected in series.
7. The deep learning based carbonate reservoir prediction method of claim 1, wherein the prediction method further comprises: reading the seismic attribute data according to an SEG-Y storage format, converting the read data into a decimal format, inputting a trained deep learning model, and generating wave impedance data;
optionally, the reading the seismic attribute data according to the SEG-Y storage format includes: firstly, for each seismic channel, reading the space coordinate information and the seismic attribute data length of a seismic monitoring point from the first N bytes; then, skipping the first N bytes, and reading the seismic attribute data of the current seismic channel according to the acquired data length; according to the format of the seismic attribute data, adaptively calculating the number of bytes occupied by the seismic attribute data in the seismic channel;
optionally, the method further comprises: converting the read data into decimal format, normalizing the data, inputting the normalized data into a trained deep learning model, and generating wave impedance data;
optionally, the method further comprises reversely storing the generated wave impedance data;
preferably, the generated wave impedance data are respectively formatted and calculated according to the data format, the sampling interval and the sampling point number information of the acquired seismic attribute data, and then are reversely stored to generate a reservoir prediction data body of the SEG-Y readable by commercial software.
8. A carbonate reservoir prediction apparatus based on deep learning, the apparatus comprising a memory for storing program instructions and a processor; the processor is configured to invoke program instructions that, when executed, perform the deep learning based carbonate reservoir prediction method of any of claims 1-7.
9. A deep learning based carbonate reservoir prediction system comprising a computer program, wherein the computer program when executed by a processor implements the deep learning based carbonate reservoir prediction method of any of claims 1-7.
10. A computer readable storage medium having stored thereon a computer program, which when executed by a processor, implements the deep learning based carbonate reservoir prediction method of any of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310573765.3A CN116559949A (en) | 2023-05-19 | 2023-05-19 | Carbonate reservoir prediction method, system and equipment based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310573765.3A CN116559949A (en) | 2023-05-19 | 2023-05-19 | Carbonate reservoir prediction method, system and equipment based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116559949A true CN116559949A (en) | 2023-08-08 |
Family
ID=87499870
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310573765.3A Pending CN116559949A (en) | 2023-05-19 | 2023-05-19 | Carbonate reservoir prediction method, system and equipment based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116559949A (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210117737A1 (en) * | 2019-10-18 | 2021-04-22 | Korea University Research And Business Foundation | Earthquake event classification method using attention-based convolutional neural network, recording medium and device for performing the method |
| CN112883992A (en) * | 2020-12-11 | 2021-06-01 | 太原理工大学 | Breast cancer lump classification method based on attention ResNet model |
| CN113109874A (en) * | 2021-03-30 | 2021-07-13 | 中国石油大学(华东) | Wave impedance inversion method using neural network and neural network system |
| CN113343799A (en) * | 2021-05-25 | 2021-09-03 | 山东师范大学 | Method and system for realizing automatic classification of white blood cells based on mixed attention residual error network |
| CN114863283A (en) * | 2022-05-25 | 2022-08-05 | 南昌航空大学 | SAR image target identification method combining transfer learning and attention mechanism |
| CN115330807A (en) * | 2022-07-12 | 2022-11-11 | 国电南瑞科技股份有限公司 | Choroidal neovascularization image segmentation method based on hybrid convolutional network |
| CN115527107A (en) * | 2021-06-25 | 2022-12-27 | 中国石油天然气股份有限公司 | Carbonate reservoir plane prediction method, system, equipment and storage medium |
| CN115700403A (en) * | 2022-09-30 | 2023-02-07 | 中国石油化工股份有限公司 | Acoustic impedance inversion method and system based on well control semi-supervised deep learning |
| CN116009080A (en) * | 2022-12-31 | 2023-04-25 | 吉首大学 | Seismic wave impedance inversion method and system, electronic equipment and storage medium |
| CN116047583A (en) * | 2021-10-27 | 2023-05-02 | 中国石油化工股份有限公司 | Adaptive wave impedance inversion method and system based on depth convolution neural network |
-
2023
- 2023-05-19 CN CN202310573765.3A patent/CN116559949A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210117737A1 (en) * | 2019-10-18 | 2021-04-22 | Korea University Research And Business Foundation | Earthquake event classification method using attention-based convolutional neural network, recording medium and device for performing the method |
| CN112883992A (en) * | 2020-12-11 | 2021-06-01 | 太原理工大学 | Breast cancer lump classification method based on attention ResNet model |
| CN113109874A (en) * | 2021-03-30 | 2021-07-13 | 中国石油大学(华东) | Wave impedance inversion method using neural network and neural network system |
| CN113343799A (en) * | 2021-05-25 | 2021-09-03 | 山东师范大学 | Method and system for realizing automatic classification of white blood cells based on mixed attention residual error network |
| CN115527107A (en) * | 2021-06-25 | 2022-12-27 | 中国石油天然气股份有限公司 | Carbonate reservoir plane prediction method, system, equipment and storage medium |
| CN116047583A (en) * | 2021-10-27 | 2023-05-02 | 中国石油化工股份有限公司 | Adaptive wave impedance inversion method and system based on depth convolution neural network |
| CN114863283A (en) * | 2022-05-25 | 2022-08-05 | 南昌航空大学 | SAR image target identification method combining transfer learning and attention mechanism |
| CN115330807A (en) * | 2022-07-12 | 2022-11-11 | 国电南瑞科技股份有限公司 | Choroidal neovascularization image segmentation method based on hybrid convolutional network |
| CN115700403A (en) * | 2022-09-30 | 2023-02-07 | 中国石油化工股份有限公司 | Acoustic impedance inversion method and system based on well control semi-supervised deep learning |
| CN116009080A (en) * | 2022-12-31 | 2023-04-25 | 吉首大学 | Seismic wave impedance inversion method and system, electronic equipment and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 丁燕;杜启振;QAMAR YASIN;张强;刘力辉;: "基于深度学习的裂缝预测在S区潜山碳酸盐岩储层中的应用", 石油物探, no. 02, pages 113 - 121 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11668853B2 (en) | Petrophysical inversion with machine learning-based geologic priors | |
| CA3122684C (en) | Data augmentation for seismic interpretation systems and methods | |
| CN113759425B (en) | Method and system for evaluating filling characteristics of deep paleo-karst reservoir stratum by well-seismic combination | |
| CN111239802B (en) | Deep learning speed modeling method based on seismic reflection waveform and velocity spectrum | |
| EP1782328B1 (en) | Multiple-point statistic (mps) simulation with enhanced computational efficiency | |
| US5966672A (en) | Visualization technology method | |
| CN113759424B (en) | Karst reservoir filling analysis method and system based on spectral decomposition and machine learning | |
| EP2409176B1 (en) | Seismic imaging systems and methods employing a fast target-oriented illumination calculation | |
| CN104081226B (en) | Iteration inclination angle for seismic data process guides medium filtering system and method | |
| EP2497043B1 (en) | Seismic imaging systems and methods employing a 3d reverse time migration with tilted transverse isotropy | |
| US20200264327A1 (en) | Method for fast calculation of seismic attributes using artificial intelligence | |
| CN116047583A (en) | Adaptive wave impedance inversion method and system based on depth convolution neural network | |
| EP3978961B1 (en) | System and method for quantitative seismic integration modeling workflow | |
| US11397273B2 (en) | Full waveform inversion in the midpoint-offset domain | |
| CN116559949A (en) | Carbonate reservoir prediction method, system and equipment based on deep learning | |
| US20230140656A1 (en) | Method and system for determining seismic processing parameters using machine learning | |
| CN113806674A (en) | Method and device for quantifying longitudinal dimension of ancient river channel, electronic equipment and storage medium | |
| Menzel | Constrained indicator data resampling—A parameter constrained irregular resampling method for scattered point data | |
| CN113568040B (en) | Repeatability analysis method and system for time-lapse seismic acquisition data | |
| Xin et al. | Integrated Carbonate Lithofacies Modeling Based on the Deep Learning and Seismic Inversion and its Application | |
| CN110954956B (en) | Method for evaluating acquisition trace of observation system and computer-readable storage medium | |
| CA2928827A1 (en) | Geocellular modeling | |
| CN117724151A (en) | Construction method of river facies compact heterogeneous reservoir low-frequency model | |
| Manchuk et al. | Algorithms for Unstructured Grids |
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 |