CN115619950B - Three-dimensional geological modeling method based on deep learning - Google Patents
Three-dimensional geological modeling method based on deep learning Download PDFInfo
- Publication number
- CN115619950B CN115619950B CN202211258278.XA CN202211258278A CN115619950B CN 115619950 B CN115619950 B CN 115619950B CN 202211258278 A CN202211258278 A CN 202211258278A CN 115619950 B CN115619950 B CN 115619950B
- Authority
- CN
- China
- Prior art keywords
- dimensional
- model
- coding
- layer
- stratum
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 53
- 238000013135 deep learning Methods 0.000 title claims abstract description 29
- 238000012549 training Methods 0.000 claims abstract description 72
- 238000013507 mapping Methods 0.000 claims abstract description 54
- 238000005553 drilling Methods 0.000 claims abstract description 40
- 238000012545 processing Methods 0.000 claims abstract description 20
- 238000007781 pre-processing Methods 0.000 claims abstract description 14
- 239000013598 vector Substances 0.000 claims description 37
- 230000015654 memory Effects 0.000 claims description 21
- 230000006870 function Effects 0.000 claims description 18
- 230000015572 biosynthetic process Effects 0.000 claims description 16
- 238000003860 storage Methods 0.000 claims description 13
- 238000012360 testing method Methods 0.000 claims description 6
- 230000000903 blocking effect Effects 0.000 claims description 5
- 238000004590 computer program Methods 0.000 claims description 5
- 238000011478 gradient descent method Methods 0.000 claims description 5
- 238000006243 chemical reaction Methods 0.000 claims description 4
- 238000004519 manufacturing process Methods 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 claims description 3
- 238000009826 distribution Methods 0.000 claims description 2
- 230000000873 masking effect Effects 0.000 claims description 2
- 238000005192 partition Methods 0.000 claims 2
- 238000000638 solvent extraction Methods 0.000 claims 2
- 238000005516 engineering process Methods 0.000 description 6
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 230000018109 developmental process Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 239000003208 petroleum Substances 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 239000008280 blood Substances 0.000 description 1
- 210000004369 blood Anatomy 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 229910052500 inorganic mineral Inorganic materials 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 239000011707 mineral Substances 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
- G06T17/05—Geographic models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
- G06T17/20—Finite element generation, e.g. wire-frame surface description, tesselation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20021—Dividing image into blocks, subimages or windows
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
Abstract
The invention discloses a three-dimensional geological modeling method based on deep learning, which comprises the following steps: acquiring a three-dimensional geological model sample set, and preprocessing the three-dimensional geological model sample set; performing mask processing on the preprocessed three-dimensional geological model sample set to extract drilling data, and then constructing a training data set; training a three-dimensional geological modeling model to be trained based on the training data set to obtain a trained three-dimensional geological modeling model, wherein the three-dimensional geological modeling model at least comprises a coding layer, a linear mapping layer, a transducer decoding layer and an inverse mapping layer which are connected in sequence, and the coding layer is used for carrying out stratum coding and position coding on training data; acquiring acquired drilling data, and inputting the drilling data into the three-dimensional geological modeling model to generate a three-dimensional geological model. The method solves the problem that the global relation of the three-dimensional geological model cannot be modeled at present.
Description
Technical Field
The invention relates to the technical field of mapping, in particular to a three-dimensional geological modeling method, device, electronic equipment and storage medium based on deep learning.
Background
In recent years, digital twinning has further driven the digitization, informatization and intellectualization of various industries, following the digital earth and smart cities. In the geological industry, after digital twinning, industry or national grade research plans of digital twinning mine, petroleum geological twinning and the like are proposed, and the three-dimensional geological modeling technology is a foundation and key technology of digital twinning mine and petroleum digital twinning and is also an important component of digital earth and smart city. Compared with the traditional two-dimensional geological data, the three-dimensional geological model can more completely express various geological phenomena, so that decision making and geological analysis can be better carried out. The three-dimensional geological modeling plays an important role in engineering exploration and mineral exploitation, rapidly develops in more and more aspects such as urban planning and construction, water resource management and geological disaster analysis, and is a key technical support for promoting the three-dimensional reconstruction of overground and underground entities and the development of urban digital twin technology.
Through the development of the past decades, three-dimensional geologic modeling techniques have been fully developed, and a plurality of mature modeling systems and a series of mature modeling methods have been formed. For example, explicit or implicit modeling methods supported by various interpolation techniques are widely applied to engineering geology, and stochastic modeling methods supported by geostatistics are widely applied to hydrogeology modeling and reservoir modeling. However, to date, the accurate and intelligent creation of three-dimensional visual fine geologic models based on limited geologic survey data remains a significant challenge in the field of geologic modeling. With the rapid development of deep learning technology in recent years and the successful application of the deep learning technology in various fields, the deep learning technology injects fresh blood into the traditional geology field, and introduces a new method for geology modeling.
However, existing three-dimensional geologic modeling algorithms based on deep learning do not consider the global relevance of geologic models and cannot model the global relationships of three-dimensional geologic models. The existing conditional three-dimensional geological modeling algorithm based on deep learning mostly adopts pre-trained unconditional generation countermeasure network model to combine with conditional constraint network for modeling, the algorithm does not consider the relation among all unknown points, the values of the unknown points are mutually independent, the correlation between the simulated pixel points and the global other pixel points cannot be considered, and the dependence among all pixels in the model cannot be modeled, so that the overall and local relation of the model is difficult to control, and the simulation capability of the geological model is deficient.
Disclosure of Invention
The invention aims to overcome the technical defects, and provides a three-dimensional geological modeling method, device, electronic equipment and storage medium based on deep learning, which solve the technical problem that the global relation of a three-dimensional geological model cannot be modeled in the prior art.
In order to achieve the technical purpose, the invention adopts the following technical scheme:
in a first aspect, the present invention provides a three-dimensional geologic modeling method, comprising the steps of:
acquiring a three-dimensional geological model sample set, and preprocessing the three-dimensional geological model sample set;
performing mask processing on the preprocessed three-dimensional geological model sample set to extract drilling data, and then constructing a training data set;
training a three-dimensional geological modeling model to be trained based on the training data set to obtain a trained three-dimensional geological modeling model, wherein the three-dimensional geological modeling model at least comprises a coding layer, a linear mapping layer, a transducer decoding layer and an inverse mapping layer which are connected in sequence, and the coding layer is used for carrying out stratum coding and position coding on training data;
acquiring acquired drilling data, and inputting the drilling data into the three-dimensional geological modeling model to generate a three-dimensional geological model.
In one embodiment, the acquiring a three-dimensional geological model sample set and preprocessing the three-dimensional geological model sample set includes:
acquiring a three-dimensional geological model sample set, and manufacturing a stratum coding comparison table according to stratum in the three-dimensional geological model data set so as to carry out label coding on each stratum;
converting attribute values of each three-dimensional geological model in the three-dimensional geological model sample set according to the stratum coding comparison table;
and carrying out format conversion on the three-dimensional geological model sample set to obtain a three-dimensional geological model sample set with a regular grid format.
In one embodiment, the masking the preprocessed three-dimensional geological model sample set to extract drilling data, and then constructing a training data set includes:
carrying out random mask processing on the three-dimensional geological model sample set in the regular grid format to obtain a three-dimensional model set for extracting part of virtual drilling holes;
normalizing the three-dimensional model set of the extracted part of the virtual drilling holes;
and dividing the three-dimensional model set obtained after the processing into a training set and a testing set.
In one embodiment, the manner in which the coding layer performs the layer coding and the position coding on the input training data is specifically:
three-dimensional convolution is carried out on an input three-dimensional model by adopting a three-dimensional convolution kernel, and the three-dimensional model is segmented to obtain a three-dimensional model block, wherein the size of the three-dimensional convolution kernel is the size of the three-dimensional model block, and the step size of each three-dimensional convolution kernel is the same as the size of the convolution kernel;
expanding the segmented three-dimensional model into a sequence to obtain stratum codes of the three-dimensional model, wherein the length of the sequence is the number of blocks into which the three-dimensional model is divided, and word vectors in the sequence are the number of pixels in the segmented three-dimensional model;
and (3) performing position coding on the three-dimensional model after the blocking by adopting a method of combining [0,1] range coding with linear mapping.
In one embodiment, the linear mapping layer is configured to map the formation code and the position code into word vectors with the same shape, the transform decoding layer is configured to decode and output an output result of the linear mapping layer, and the transform decoding layer is specifically configured to:
and adding the preset random noise, the mapped stratum code and the position code, and then decoding, wherein the shape of the preset random noise is the same as that of the mapped stratum code and the position code.
In one embodiment, the inverse mapping layer is configured to restore an output result of the transform decoding layer to the pixel layer through a fully-connected layer structure, where the number of input channels of the fully-connected layer is the number of channels of the transform decoding layer, and the number of output channels is a product of a total number of pixels and a number of formation categories of the three-dimensional geological model divided by a word vector sequence length of the transform decoding layer.
In one embodiment, the three-dimensional geologic modeling method further comprises:
based on a preset multi-weight cross entropy loss function, performing convergence training on the three-dimensional geological modeling model by adopting a backward propagation and random gradient descent method so as to construct a trained three-dimensional geological modeling model.
In a second aspect, the present invention further provides a three-dimensional geological modeling apparatus based on deep learning, including:
the preprocessing module is used for acquiring a three-dimensional geological model sample set and preprocessing the three-dimensional geological model sample set;
the training data acquisition module is used for carrying out mask processing on the preprocessed three-dimensional geological model sample set so as to construct a training data set after extracting drilling data;
the training module is used for training the three-dimensional geologic modeling model to be trained based on the training data set to obtain a trained three-dimensional geologic modeling model, wherein the three-dimensional geologic modeling model at least comprises a coding layer, a linear mapping layer, a transducer decoding layer and an inverse mapping layer which are connected in sequence, and the coding layer is used for carrying out stratum coding and position coding on training data;
the modeling module is used for acquiring the acquired drilling data, and inputting the drilling data into the three-dimensional geological modeling model to generate the three-dimensional geological model.
In a third aspect, the present invention also provides an electronic device, including: a processor and a memory;
the memory has stored thereon a computer program executable by the processor;
the processor, when executing the computer program, implements the steps in the three-dimensional geological modeling method based on deep learning as described above.
In a fourth aspect, the present invention also provides a computer readable storage medium storing one or more programs executable by one or more processors to implement the steps in the deep learning based three-dimensional geologic modeling method as described above.
Compared with the prior art, the three-dimensional geological modeling method, the device, the electronic equipment and the storage medium based on the deep learning provided by the invention have the advantages that firstly, a three-dimensional geological model sample set is obtained, and the three-dimensional geological model sample set is preprocessed; then carrying out mask processing on the preprocessed three-dimensional geological model sample set to extract drilling data, and constructing a training data set; then training a three-dimensional geological modeling model to be trained based on the training data set to obtain a trained three-dimensional geological modeling model, wherein the three-dimensional geological modeling model at least comprises a coding layer, a linear mapping layer, a transducer decoding layer and an inverse mapping layer which are connected in sequence, and the coding layer is used for carrying out stratum coding and position coding on training data; and finally acquiring the acquired drilling data, and inputting the drilling data into the three-dimensional geological modeling model to generate the three-dimensional geological model. The method has the advantages that the method is capable of applying the Transformer which is good at establishing global correlation in the field of three-dimensional geologic modeling, stratum and position information is extracted through a coding method for constructing stratum and position, the three-dimensional geologic model which is attached to the known drilling hole is generated in a full-automatic mode, the quality and the refinement degree of the generated three-dimensional geologic model are improved, and the application value of the three-dimensional geologic model is further improved.
Drawings
FIG. 1 is a flow chart of a three-dimensional geologic modeling method based on deep learning provided by an embodiment of the invention;
FIG. 2 is a schematic diagram of a three-dimensional geologic modeling model provided by an embodiment of the invention;
FIG. 3 is a schematic diagram of functional modules of a three-dimensional geological modeling apparatus based on deep learning according to an embodiment of the present invention;
fig. 4 is a schematic diagram of a hardware structure of an electronic device according to an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Referring to fig. 1 and 2, the three-dimensional geological modeling method based on deep learning provided by the invention comprises the following steps:
s100, acquiring a three-dimensional geological model sample set, and preprocessing the three-dimensional geological model sample set;
s200, performing mask processing on the preprocessed three-dimensional geological model sample set to extract drilling data, and performing stratum coding and position coding on the drilling data to obtain a training data set;
s300, training a three-dimensional geologic modeling model to be trained based on the training data set to obtain a trained three-dimensional geologic modeling model, wherein the three-dimensional geologic modeling model at least comprises a linear mapping layer, a transducer decoding layer and an inverse mapping layer which are connected in sequence;
s400, acquiring acquired drilling data, performing stratum coding and position coding on the acquired drilling data, and inputting the processed drilling data into the three-dimensional geological modeling model to generate a three-dimensional geological model.
In the embodiment, the coding layer and the linear mapping layer are constructed, so that the application of a transducer which is good at establishing global correlation in the field of three-dimensional geologic modeling is possible, the formation and position information is extracted through a coding method for constructing the formation and the position, the three-dimensional geologic model which is attached to the known drilling hole is automatically generated, the quality and the refinement degree of the generated three-dimensional geologic model are improved, and the application value of the three-dimensional geologic model is further improved.
In one embodiment, the step S100 specifically includes:
acquiring a three-dimensional geological model sample set, and manufacturing a stratum coding comparison table according to stratum in the three-dimensional geological model data set so as to carry out label coding on each stratum;
converting attribute values of each three-dimensional geological model in the three-dimensional geological model sample set according to the stratum coding comparison table;
and carrying out format conversion on the three-dimensional geological model sample set to obtain a three-dimensional geological model sample set with a regular grid format.
In this embodiment, firstly, a formation coding comparison table is made according to all the strata appearing in the current three-dimensional geologic model data set, tag coding is performed on each stratum, then, according to the formation coding comparison table, attribute values of the three-dimensional geologic model are set according to the formation coding comparison table, then, the geologic model is arranged into a grid format, and a data set for deep learning is conveniently constructed, wherein when the original geologic model is stratum surface data of a triangular network structure, triangular network data are converted into regular grid data by means of regular sampling and judging which stratum is located with sampling points.
In one embodiment, the step S200 specifically includes:
carrying out random mask processing on the three-dimensional geological model sample set in the regular grid format to obtain a three-dimensional model set for extracting part of virtual drilling holes;
normalizing the three-dimensional model set of the extracted part of the virtual drilling holes;
and dividing the three-dimensional model set obtained after the processing into a training set and a testing set.
In this embodiment, a random mask is made on a three-dimensional geological model which is arranged by using a three-dimensional geological model with a regular grid format, a part of virtual drilling is reserved, the rest is assigned to 0, then the three-dimensional model is normalized by adopting a maximum and minimum normalization method, then a three-dimensional model set obtained after processing is divided into a training set and a test set, and an exemplary method is that a specified number of three-dimensional geological models are randomly extracted from the constructed three-dimensional model set, 80% of the extracted three-dimensional geological models are used as the training set, and 20% of the extracted three-dimensional geological models are used as the test set, so that a training data set for deep learning is constructed.
After the training data set is constructed, the number of blocks of the transform decoding layer and the number of multi-head attention modules are set, the number of blocks of the three-dimensional geological model is preset, the dimension of word vectors required to be input by the transform decoding layer is calculated, and the encoding layer can be used for carrying out stratum encoding and position encoding on the input three-dimensional data.
In some embodiments, referring to fig. 2, in step S300, the manner of performing the layer coding and the position coding on the training data input by the coding layer is specifically:
three-dimensional convolution is carried out on an input three-dimensional model by adopting a three-dimensional convolution kernel, and the three-dimensional model is segmented to obtain a three-dimensional model block, wherein the size of the three-dimensional convolution kernel is the size of the three-dimensional model block, and the step size of each three-dimensional convolution kernel is the same as the size of the convolution kernel;
expanding the segmented three-dimensional model into a sequence to obtain stratum codes of the three-dimensional model, wherein the length of the sequence is the number of blocks into which the three-dimensional model is divided, and word vectors in the sequence are the number of pixels in the segmented three-dimensional model;
and (3) performing position coding on the three-dimensional model after the blocking by adopting a method of combining [0,1] range coding with linear mapping.
In this embodiment, the 3DCNN is adopted to perform three-dimensional convolution on the input three-dimensional model, the three-dimensional model is segmented in a convolution manner, the size of the three-dimensional convolution kernel is the size of each three-dimensional model block, the step size of the three-dimensional convolution kernel is the same as the size of the convolution kernel, so that each segmented three-dimensional model is mapped into a word vector independently, then the three-dimensional model feature vector after the three-dimensional convolution is unfolded into a sequence, the length of the sequence is the number of blocks into which the three-dimensional model is divided, and the dimension of the word vector in the sequence is the number of pixels in the model after the segmentation, so that the stratum coding feature of the three-dimensional model can be obtained.
When encoding three-dimensional stratum information, the three-dimensional model needs to be split into more word vectors as much as possible according to hardware equipment, namely, a word vector represents a model block smaller, so that a self-attention module in a decoding layer of a transducer extracts more three-dimensional geologic model features.
In the position coding, 0,1 is used]Performing position coding on the segmented three-dimensional model by combining range coding with a linear mapping method; firstly, the coordinate with the smallest actual geographic coordinate in three directions in each model block is used for replacing the whole model block, and secondly, the absolute position coordinate (x, y, z) of each model block in the three-dimensional model after the block is divided is subtracted by the smallest coordinate (x min ,y min ,z min ) Dividing the coordinates of each model block by the actual length, width and height (l, w, h) of the model block to obtain a preliminary position code (i, j, k), and normalizing by using the maximum and minimum values to obtain [0,1] of the three-dimensional geological model]The calculation formula of the position code, the preliminary position code (i, j, k) is as follows:
in some embodiments, referring to fig. 2, the linear mapping layer is configured to map the formation code and the position code into word vectors with the same shape, the transform decoding layer is configured to decode and output an output result of the linear mapping layer, and the transform decoding layer is specifically configured to:
and adding the preset random noise, the mapped stratum code and the position code, and then decoding, wherein the shape of the preset random noise is the same as that of the mapped stratum code and the position code.
In this embodiment, a linear mapping layer is constructed before the decoding layer of the transducer, and the extracted stratum coding information and position coding information of the virtual drilling three-dimensional model are mapped into one-dimensional word vectors with the same shape. Illustratively, three-dimensional position information after [0,1] range coding is input into a full-connection layer for linear mapping, and the number of channels of the position coding is mapped from 3 to the number of channels of a transducer decoding layer.
And then mapping a random vector into random noise with the same shape as the stratum code and the position code, adding the random noise with the stratum code and the position code, and then decoding, wherein the random noise is obtained by linear mapping of the random vector, and the random noise, the position code and the stratum code are input into a transducer decoding layer together so as to improve the anti-noise capability of the three-dimensional geological model and reduce the noise. The length of the random vector is the word vector length of the transform decoding layer, the channel number is 1 and accords with normal distribution. Illustratively, the random vector is first input into a fully connected linear mapping layer, the number of channels is mapped from 1 to the number of channels input into a transform decoding layer, and then the linear mapped stratum code, position code and random noise are added and then decoded, so that the characteristics of the three-dimensional geological model are fully extracted.
In some embodiments, referring to fig. 2, the inverse mapping layer is configured to restore an output result of the transform decoding layer to a pixel layer through a fully-connected layer structure, where the number of input channels of the fully-connected layer is the number of channels of the transform decoding layer, and the number of output channels is a product of a total number of pixels and a number of formation categories of the three-dimensional geological model divided by a word vector sequence length of the transform decoding layer.
In this embodiment, an inverse mapping layer is constructed after the transform decoding layer, and the information after decoding is restored to the pixel layer using the full connection layer. Specifically, the inverse mapping layer is constructed by adopting a full-connection layer, and is used for restoring the three-dimensional stratum characteristics decoded by the transform decoding layer into a three-dimensional geological model, the number of input channels of the full-connection layer is the number of channels of the transform decoding layer, the number of output channels is the product of the total pixel number and the stratum category number of the three-dimensional geological model divided by the word vector sequence length of the transform decoding layer, then the result of the full-connection layer is input into an argmax function, a one-dimensional three-dimensional geological model array is obtained, and finally the one-dimensional three-dimensional geological model array is restored into three dimensions, so that the three-dimensional geological model is obtained.
In some embodiments, to increase the accuracy of model building, the method further comprises:
based on a preset multi-weight cross entropy loss function, performing convergence training on the three-dimensional geological modeling model by adopting a backward propagation and random gradient descent method so as to construct a trained three-dimensional geological modeling model.
In this embodiment, a multi-weighted loss function is constructed, and the loss function is used to calculate l oss of the model output result compared with the original three-dimensional geological model, and the method of backward propagation and random gradient descent is adopted to train the network, so as to obtain a trained deep learning model.
Specifically, constructing a multi-weight cross entropy loss function, so that the network more pays attention to drilling data and the neighborhood of the drilling data; loss function of the area to be drilled 0 Weight set to 0.5, loss function of first order neighborhood of borehole/ n1 Weight is set to 0.25, and the loss function l of the secondary neighborhood is set to be 0.25 n2 Weight is set to 0.15, and the remaining area loss function l n3 The weight is set to 0.1, i.e. the loss function can be written as loss=0.5 l 0 +0.25l n1 +0.15l n2 +0.1l n3 The method comprises the steps of carrying out a first treatment on the surface of the The first-order neighborhood of the borehole is a set of planar eight neighbors of each pixel on the borehole; the secondary neighborhood of the borehole is a set of planar twenty-five neighbors of each pixel on the borehole; and training the network by calculating a loss function of the prediction result and the original model and adopting a backward propagation and random gradient descent method, and finally obtaining the deep learning model of the three-dimensional geological modeling.
According to the technical scheme provided by the invention, a three-dimensional geological model sample set is firstly obtained, and the three-dimensional geological model sample set is preprocessed; then carrying out mask processing on the preprocessed three-dimensional geological model sample set to extract drilling data, and constructing a training data set; then training a three-dimensional geological modeling model to be trained based on the training data set to obtain a trained three-dimensional geological modeling model, wherein the three-dimensional geological modeling model at least comprises a coding layer, a linear mapping layer, a transducer decoding layer and an inverse mapping layer which are connected in sequence, and the coding layer is used for carrying out stratum coding and position coding on training data; and finally acquiring the acquired drilling data, and inputting the drilling data into the three-dimensional geological modeling model to generate the three-dimensional geological model. By using the 3DCNN to construct the linear mapping layer, the application of a transducer which is good at establishing global correlation in the field of three-dimensional geologic modeling is possible, stratum and position information is extracted by a coding method for constructing stratum and position, the training of drilling data and neighborhood thereof is enhanced by constructing a multi-weight cross entropy loss function, training and denoising are assisted by constructing random noise, a three-dimensional geologic model which is attached to a known drilling hole is automatically generated, the quality and the refinement degree of the generated three-dimensional geologic model are improved, and the application value of the three-dimensional geologic model is further improved.
In another embodiment of the present invention, referring to fig. 3, the three-dimensional geological modeling apparatus includes a preprocessing module 11, a training data obtaining module 12, a training module 13, and a modeling module 14.
The preprocessing module 11 is used for acquiring a three-dimensional geological model sample set and preprocessing the three-dimensional geological model sample set.
The training data acquisition module 12 is configured to perform mask processing on the preprocessed three-dimensional geological model sample set to extract drilling data, and then construct a training data set.
The training module 13 is configured to train the three-dimensional geologic modeling model to be trained based on the training data set, and obtain a trained three-dimensional geologic modeling model, where the three-dimensional geologic modeling model at least includes a coding layer, a linear mapping layer, a transducer decoding layer, and an inverse mapping layer that are sequentially connected, and the coding layer is configured to perform formation coding and position coding on training data.
The modeling module 14 is configured to obtain acquired borehole data, and input the borehole data into the three-dimensional geologic modeling model to generate a three-dimensional geologic model.
In the embodiment, the coding layer and the linear mapping layer are constructed, so that the application of a transducer which is good at establishing global correlation in the field of three-dimensional geologic modeling is possible, the formation and position information is extracted through a coding method for constructing the formation and the position, the three-dimensional geologic model which is attached to the known drilling hole is automatically generated, the quality and the refinement degree of the generated three-dimensional geologic model are improved, and the application value of the three-dimensional geologic model is further improved.
It should be noted that, the modules referred to in the present invention refer to a series of computer program instruction segments capable of performing specific functions, and specific embodiments of each module refer to the above corresponding method embodiments, which are not described herein again.
In one embodiment, the preprocessing module 11 is specifically configured to:
acquiring a three-dimensional geological model sample set, and manufacturing a stratum coding comparison table according to stratum in the three-dimensional geological model data set so as to carry out label coding on each stratum;
converting attribute values of each three-dimensional geological model in the three-dimensional geological model sample set according to the stratum coding comparison table;
and carrying out format conversion on the three-dimensional geological model sample set to obtain a three-dimensional geological model sample set with a regular grid format.
In one embodiment, the training data acquisition module 12 is specifically configured to:
carrying out random mask processing on the three-dimensional geological model sample set in the regular grid format to obtain a three-dimensional model set for extracting part of virtual drilling holes;
normalizing the three-dimensional model set of the extracted part of the virtual drilling holes;
and dividing the three-dimensional model set obtained after the processing into a training set and a testing set.
In one embodiment, the manner in which the coding layer performs the layer coding and the position coding on the input training data is specifically:
three-dimensional convolution is carried out on an input three-dimensional model by adopting a three-dimensional convolution kernel, and the three-dimensional model is segmented to obtain a three-dimensional model block, wherein the size of the three-dimensional convolution kernel is the size of the three-dimensional model block, and the step size of each three-dimensional convolution kernel is the same as the size of the convolution kernel;
expanding the segmented three-dimensional model into a sequence to obtain stratum codes of the three-dimensional model, wherein the length of the sequence is the number of blocks into which the three-dimensional model is divided, and word vectors in the sequence are the number of pixels in the segmented three-dimensional model;
and (3) performing position coding on the three-dimensional model after the blocking by adopting a method of combining [0,1] range coding with linear mapping.
In one embodiment, the linear mapping layer is configured to map the formation code and the position code into word vectors with the same shape, the transform decoding layer is configured to decode and output an output result of the linear mapping layer, and the transform decoding layer is specifically configured to:
and adding the preset random noise, the mapped stratum code and the position code, and then decoding, wherein the shape of the preset random noise is the same as that of the mapped stratum code and the position code.
In one embodiment, the inverse mapping layer is configured to restore an output result of the transform decoding layer to the pixel layer through a fully-connected layer structure, where the number of input channels of the fully-connected layer is the number of channels of the transform decoding layer, and the number of output channels is a product of a total number of pixels and a number of formation categories of the three-dimensional geological model divided by a word vector sequence length of the transform decoding layer.
In one embodiment, the training module 13 is further configured to:
based on a preset multi-weight cross entropy loss function, performing convergence training on the three-dimensional geological modeling model by adopting a backward propagation and random gradient descent method so as to construct a trained three-dimensional geological modeling model.
Another embodiment of the present invention provides an electronic device, as shown in fig. 4, the electronic device 10 includes:
one or more processors 110 and a memory 120, one processor 110 being illustrated in fig. 4, the processors 110 and the memory 120 being coupled via a bus or other means, the bus coupling being illustrated in fig. 4.
The processor 110 is configured to implement various control logic for the electronic device 10, which may be a general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA), a single-chip microcomputer, ARM (Acorn RISC Machine) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof. Also, the processor 110 may be any conventional processor, microprocessor, or state machine. The processor 110 may also be implemented as a combination of computing devices, e.g., a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a DSP, and/or any other such configuration.
The memory 120 is used as a non-volatile computer readable storage medium for storing non-volatile software programs, non-volatile computer executable programs and modules, such as program instructions corresponding to the three-dimensional geologic modeling method in the embodiment of the invention. The processor 110 executes various functional applications of the electronic device 10 and data processing, i.e., implements the three-dimensional geologic modeling methods in the method embodiments described above, by running non-volatile software programs, instructions, and units stored in the memory 120.
The memory 120 may include a storage program area that may store an operating platform, at least one application program required for a function, and a storage data area; the storage data area may store data created from the use of the electronic device 10, and the like. In addition, memory 120 may include high-speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, memory 120 may optionally include memory located remotely from processor 110, which may be connected to electronic device 10 via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
One or more units are stored in memory 120 that, when executed by one or more processors 110, perform the three-dimensional geologic modeling method in any of the method embodiments described above, e.g., perform method steps S100 through S400 in fig. 1 described above.
Another embodiment of the present invention provides a computer-readable storage medium storing computer-executable instructions for execution by one or more processors, e.g., to perform the method steps S100 through S400 of fig. 1 described above.
By way of example, computer-readable storage media can comprise read-only memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically erasable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM), which acts as external cache memory. By way of illustration and not limitation, RAM may be available in many forms such as Synchronous RAM (SRAM), dynamic RAM, (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchlink DRAM (SLDRAM), and Direct Rambus RAM (DRRAM). The disclosed memory components or memories of the operating environments described herein are intended to comprise one or more of these and/or any other suitable types of memory.
In summary, according to the three-dimensional geological modeling method, the device, the electronic equipment and the storage medium based on deep learning provided by the invention, a three-dimensional geological model sample set is firstly obtained, and the three-dimensional geological model sample set is preprocessed; then carrying out mask processing on the preprocessed three-dimensional geological model sample set to extract drilling data, and constructing a training data set; then training a three-dimensional geological modeling model to be trained based on the training data set to obtain a trained three-dimensional geological modeling model, wherein the three-dimensional geological modeling model at least comprises a coding layer, a linear mapping layer, a transducer decoding layer and an inverse mapping layer which are connected in sequence, and the coding layer is used for carrying out stratum coding and position coding on training data; and finally acquiring the acquired drilling data, and inputting the drilling data into the three-dimensional geological modeling model to generate the three-dimensional geological model. By using the 3DCNN to construct the linear mapping layer, the application of a transducer which is good at establishing global correlation in the field of three-dimensional geologic modeling is possible, stratum and position information is extracted by a coding method for constructing stratum and position, the training of drilling data and neighborhood thereof is enhanced by constructing a multi-weight cross entropy loss function, training and denoising are assisted by constructing random noise, a three-dimensional geologic model which is attached to a known drilling hole is automatically generated, the quality and the refinement degree of the generated three-dimensional geologic model are improved, and the application value of the three-dimensional geologic model is further improved.
The above-described embodiments of the present invention do not limit the scope of the present invention. Any other corresponding changes and modifications made in accordance with the technical idea of the present invention shall be included in the scope of the claims of the present invention.
Claims (9)
1. The three-dimensional geological modeling method based on deep learning is characterized by comprising the following steps of:
acquiring a three-dimensional geological model sample set, and preprocessing the three-dimensional geological model sample set;
performing mask processing on the preprocessed three-dimensional geological model sample set to extract drilling data, and then constructing a training data set;
training a three-dimensional geological modeling model to be trained based on the training data set to obtain a trained three-dimensional geological modeling model, wherein the three-dimensional geological modeling model at least comprises a coding layer, a linear mapping layer, a transducer decoding layer and an inverse mapping layer which are connected in sequence, and the coding layer is used for carrying out stratum coding and position coding on training data;
acquiring acquired drilling data, and inputting the drilling data into the three-dimensional geological modeling model to generate a three-dimensional geological model;
the coding layer carries out stratum coding and position coding on the input training data specifically comprises the following modes:
three-dimensional convolution is carried out on an input three-dimensional model by adopting a three-dimensional convolution kernel, and the three-dimensional model is segmented to obtain a three-dimensional model block, wherein the size of the three-dimensional convolution kernel is the size of the three-dimensional model block, and the step size of each three-dimensional convolution kernel is the same as the size of the convolution kernel;
expanding the segmented three-dimensional model into a sequence to obtain stratum codes of the three-dimensional model, wherein the length of the sequence is the number of blocks into which the three-dimensional model is divided, and word vectors in the sequence are the number of pixels in the segmented three-dimensional model;
position coding is carried out on the three-dimensional model after the blocking by adopting a method combining [0,1] range coding and linear mapping;
firstly, carrying out three-dimensional convolution on an input three-dimensional model by adopting a 3DCNN (three-dimensional computational complexity), partitioning the three-dimensional model in a convolution mode, wherein the size of a three-dimensional convolution kernel is the size of each three-dimensional model block, the step size of the three-dimensional convolution kernel is the same as the size of the convolution kernel so as to map each partitioned three-dimensional model into a word vector independently, then expanding the feature vector of the three-dimensional model after the three-dimensional convolution into a sequence, wherein the length of the sequence is the number of blocks into which the three-dimensional model is partitioned, and the dimension of the word vector in the sequence is the number of pixels in the model after the partition, so that the stratum coding feature of the three-dimensional model can be obtained;
when position coding is carried out, firstly, the coordinate with the smallest actual geographic coordinate in three directions in each model block is used for replacing the whole model block, and secondly, the absolute position coordinate of each model block in the three-dimensional model after the segmentation is carried outSubtracting the minimum coordinate +.>Secondly, the coordinates of each model block are divided by the actual length, width and height of the model block>Obtaining the preliminaryPosition coding->Finally, using maximum and minimum value normalization to obtain [0,1] of the three-dimensional geological model]And (5) position coding.
2. The deep learning-based three-dimensional geologic modeling method of claim 1, wherein the obtaining a three-dimensional geologic model sample set and preprocessing the three-dimensional geologic model sample set comprises:
acquiring a three-dimensional geological model sample set, and manufacturing a stratum coding comparison table according to stratum in the three-dimensional geological model data set so as to carry out label coding on each stratum;
converting attribute values of each three-dimensional geological model in the three-dimensional geological model sample set according to the stratum coding comparison table;
and carrying out format conversion on the three-dimensional geological model sample set to obtain a three-dimensional geological model sample set with a regular grid format.
3. The deep learning based three-dimensional geologic modeling method of claim 2, wherein masking the preprocessed three-dimensional geologic model sample set to extract borehole data, constructing a training data set, comprises:
carrying out random mask processing on the three-dimensional geological model sample set in the regular grid format to obtain a three-dimensional model set for extracting part of virtual drilling holes;
normalizing the three-dimensional model set of the extracted part of the virtual drilling holes;
and dividing the three-dimensional model set obtained after the processing into a training set and a testing set.
4. The three-dimensional geological modeling method based on deep learning according to claim 3, wherein the linear mapping layer is configured to map a formation code and a position code into word vectors with the same shape, the transform decoding layer is configured to decode and output an output result of the linear mapping layer, and the transform decoding layer is specifically configured to:
adding the preset random noise, the mapped stratum code and the position code, and then decoding, wherein the shape of the preset random noise is the same as the shape of the mapped stratum code and the position code;
constructing a linear mapping layer before a decoding layer of a transducer, and mapping the extracted stratum coding information and position coding information of the three-dimensional model of the virtual drilling into one-dimensional word vectors with the same shape;
and then mapping a random vector into random noise with the same shape as the stratum code and the position code, and adding the random noise with the stratum code and the position code for decoding, wherein the random noise is obtained by linear mapping of the random vector, the random noise is input into a transducer decoding layer together with the position code and the stratum code, so that the noise resistance of the three-dimensional geological model can be improved and the noise can be reduced, the length of the random vector is the word vector length of the transducer decoding layer, and the channel number is 1 and accords with normal distribution.
5. The deep learning-based three-dimensional geologic modeling method of claim 4, wherein the inverse mapping layer is configured to restore an output result of the transform decoding layer to the pixel layer through a fully connected layer structure, wherein the number of input channels of the fully connected layer is the number of channels of the transform decoding layer, and the number of output channels is a product of a total number of pixels and a number of formation categories of the three-dimensional geologic model divided by a word vector sequence length of the transform decoding layer.
6. The deep learning based three-dimensional geologic modeling method of claim 1, further comprising:
based on a preset multi-weight cross entropy loss function, performing convergence training on the three-dimensional geological modeling model by adopting a backward propagation and random gradient descent method so as to construct a trained three-dimensional geological modeling model.
7. A three-dimensional geologic modeling apparatus based on deep learning, comprising:
the preprocessing module is used for acquiring a three-dimensional geological model sample set and preprocessing the three-dimensional geological model sample set;
the training data acquisition module is used for carrying out mask processing on the preprocessed three-dimensional geological model sample set so as to construct a training data set after extracting drilling data;
the training module is used for training the three-dimensional geologic modeling model to be trained based on the training data set to obtain a trained three-dimensional geologic modeling model, wherein the three-dimensional geologic modeling model at least comprises a coding layer, a linear mapping layer, a transducer decoding layer and an inverse mapping layer which are connected in sequence, and the coding layer is used for carrying out stratum coding and position coding on training data;
the modeling module is used for acquiring the acquired drilling data, and inputting the drilling data into the three-dimensional geological modeling model to generate a three-dimensional geological model;
the coding layer carries out stratum coding and position coding on the input training data specifically comprises the following modes:
three-dimensional convolution is carried out on an input three-dimensional model by adopting a three-dimensional convolution kernel, and the three-dimensional model is segmented to obtain a three-dimensional model block, wherein the size of the three-dimensional convolution kernel is the size of the three-dimensional model block, and the step size of each three-dimensional convolution kernel is the same as the size of the convolution kernel;
expanding the segmented three-dimensional model into a sequence to obtain stratum codes of the three-dimensional model, wherein the length of the sequence is the number of blocks into which the three-dimensional model is divided, and word vectors in the sequence are the number of pixels in the segmented three-dimensional model;
position coding is carried out on the three-dimensional model after the blocking by adopting a method combining [0,1] range coding and linear mapping;
firstly, carrying out three-dimensional convolution on an input three-dimensional model by adopting a 3DCNN (three-dimensional computational complexity), partitioning the three-dimensional model in a convolution mode, wherein the size of a three-dimensional convolution kernel is the size of each three-dimensional model block, the step size of the three-dimensional convolution kernel is the same as the size of the convolution kernel so as to map each partitioned three-dimensional model into a word vector independently, then expanding the feature vector of the three-dimensional model after the three-dimensional convolution into a sequence, wherein the length of the sequence is the number of blocks into which the three-dimensional model is partitioned, and the dimension of the word vector in the sequence is the number of pixels in the model after the partition, so that the stratum coding feature of the three-dimensional model can be obtained;
when position coding is carried out, firstly, the coordinate with the smallest actual geographic coordinate in three directions in each model block is used for replacing the whole model block, and secondly, the absolute position coordinate of each model block in the three-dimensional model after the segmentation is carried outSubtracting the minimum coordinates +.>Secondly, the coordinates of each model block are divided by the actual length, width and height of the model block>Obtaining a preliminary position code +.>Finally, using maximum and minimum value normalization to obtain [0,1] of the three-dimensional geological model]And (5) position coding.
8. An electronic device, comprising: a processor and a memory;
the memory has stored thereon a computer program executable by the processor;
the processor, when executing the computer program, implements the steps of the deep learning based three-dimensional geologic modeling method of any of claims 1-6.
9. A computer readable storage medium storing one or more programs executable by one or more processors to implement the steps in the deep learning based three-dimensional geologic modeling method of any of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211258278.XA CN115619950B (en) | 2022-10-13 | 2022-10-13 | Three-dimensional geological modeling method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211258278.XA CN115619950B (en) | 2022-10-13 | 2022-10-13 | Three-dimensional geological modeling method based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115619950A CN115619950A (en) | 2023-01-17 |
| CN115619950B true CN115619950B (en) | 2024-01-19 |
Family
ID=84862370
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211258278.XA Active CN115619950B (en) | 2022-10-13 | 2022-10-13 | Three-dimensional geological modeling method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115619950B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113658322A (en) * | 2021-07-30 | 2021-11-16 | 哈尔滨工业大学 | Visual transform-based three-dimensional voxel reconstruction method |
| CN113902861A (en) * | 2021-10-11 | 2022-01-07 | 安徽理工大学 | Three-dimensional geological modeling method based on machine learning |
| US11238307B1 (en) * | 2020-09-24 | 2022-02-01 | Eagle Technology, Llc | System for performing change detection within a 3D geospatial model based upon semantic change detection using deep learning and related methods |
| CN114117906A (en) * | 2021-11-19 | 2022-03-01 | 山东大学 | Multi-scale unsupervised seismic wave velocity inversion method based on observation data self-encoding |
| WO2022042865A1 (en) * | 2020-08-31 | 2022-03-03 | Huawei Technologies Co., Ltd. | Lifting 2d representations to 3d using attention models |
| CN114140601A (en) * | 2021-12-13 | 2022-03-04 | 杭州师范大学 | Three-dimensional grid reconstruction method and system based on single image under deep learning framework |
| CN114429529A (en) * | 2022-04-01 | 2022-05-03 | 国家电投集团科学技术研究院有限公司 | Three-dimensional geological model generation method and device, electronic equipment and storage medium |
| CN114943178A (en) * | 2022-05-19 | 2022-08-26 | 中国地质大学（武汉） | Three-dimensional geological model modeling method and device and computer equipment |
-
2022
- 2022-10-13 CN CN202211258278.XA patent/CN115619950B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2022042865A1 (en) * | 2020-08-31 | 2022-03-03 | Huawei Technologies Co., Ltd. | Lifting 2d representations to 3d using attention models |
| US11238307B1 (en) * | 2020-09-24 | 2022-02-01 | Eagle Technology, Llc | System for performing change detection within a 3D geospatial model based upon semantic change detection using deep learning and related methods |
| CN113658322A (en) * | 2021-07-30 | 2021-11-16 | 哈尔滨工业大学 | Visual transform-based three-dimensional voxel reconstruction method |
| CN113902861A (en) * | 2021-10-11 | 2022-01-07 | 安徽理工大学 | Three-dimensional geological modeling method based on machine learning |
| CN114117906A (en) * | 2021-11-19 | 2022-03-01 | 山东大学 | Multi-scale unsupervised seismic wave velocity inversion method based on observation data self-encoding |
| CN114140601A (en) * | 2021-12-13 | 2022-03-04 | 杭州师范大学 | Three-dimensional grid reconstruction method and system based on single image under deep learning framework |
| CN114429529A (en) * | 2022-04-01 | 2022-05-03 | 国家电投集团科学技术研究院有限公司 | Three-dimensional geological model generation method and device, electronic equipment and storage medium |
| CN114943178A (en) * | 2022-05-19 | 2022-08-26 | 中国地质大学（武汉） | Three-dimensional geological model modeling method and device and computer equipment |
Non-Patent Citations (6)
| Title |
|---|
| New method for estimating strike and dip based on structural expansion orientation for 3D geological modeling;Zhao, YB等;Frontiers of Earth Science;第15卷(第3期);第676-691页 * |
| Research Status of and Trends in 3D Geological Property Modeling Method:A Review;Yuyang Liu;Applied Science;第12卷(第11期);全文 * |
| Three-Dimensional Reconstruction from a Single RGB Image Using Deep Learning:A Review;Khan Muhanmmad Saif Ulah;Journal of Imaging;第8卷(第9期);全文 * |
| 一种基于钻孔数据的优化三维地质建模方法;贾志宾;曹凯;杨志强;;科学技术与工程;第17卷(第17期);第178-183页 * |
| 区域三维地质建模方法与建模系统研究;冉祥金;中国博士学位论文全文数据库 基础科学辑;第A011-320页 * |
| 基于TIN-Octree的三维地质模型构建方法研究;尚福华;杨彦彬;杜睿山;;计算技术与自动化;第38卷(第4期);第121-125页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115619950A (en) | 2023-01-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110222373B (en) | Geotechnical engineering full life cycle oriented engineering investigation data processing method | |
| EP3535607B1 (en) | Seismic data processing artificial intelligence | |
| CN112381937B (en) | Multi-source geological data coupling modeling method based on drilling and complex geological profile | |
| US10380793B2 (en) | Geobody surface reconstruction | |
| US11531127B2 (en) | Methods and systems for reference-based inversion of seismic image volumes | |
| CN116245013A (en) | Geological prediction model construction method, modeling method, equipment and storage medium | |
| Iqbal et al. | Groundwater level prediction model using correlation and difference mechanisms based on boreholes data for sustainable hydraulic resource management | |
| CN113359212A (en) | Reservoir characteristic prediction method and model based on deep learning | |
| Kim et al. | History matching of a channelized reservoir using a serial denoising autoencoder integrated with ES-MDA | |
| Chen et al. | Deep convolutional generative adversarial networks for modeling complex hydrological structures in Monte-Carlo simulation | |
| CN114661744B (en) | Terrain database updating method and system based on deep learning | |
| Li et al. | Simultaneous estimation of geologic and reservoir state variables within an ensemble-based multiple-point statistic framework | |
| CN112001435A (en) | Method and system for constructing training sample set in regional landslide early warning and storage medium | |
| CN110197013B (en) | Riverbed bedrock surface modeling method based on Morphing | |
| Zhang et al. | Multi-source information fused generative adversarial network model and data assimilation based history matching for reservoir with complex geologies | |
| Zhang et al. | GeoScope: Full 3D geospatial information system case study | |
| CN115619950B (en) | Three-dimensional geological modeling method based on deep learning | |
| CN113779881A (en) | Method, device and equipment for predicting capacity of dense water-containing gas reservoir | |
| US20240127039A1 (en) | Geologic learning framework | |
| CN114581556A (en) | Digital map filling method in regional geological survey | |
| Cao et al. | Self-supervised temporal analysis of spatiotemporal data | |
| CN112084553A (en) | Surveying method for tunnel planning | |
| Ahmadi et al. | A sensitivity study of FILTERSIM algorithm when applied to DFN modeling | |
| CN112307541A (en) | Urban underground space rock-soil informatization comprehensive integrated digital delivery method | |
| CN109272042B (en) | Small-scale crack modeling method, device, equipment and medium based on PR model |
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 |