CN117574269A - Intelligent identification method and system for natural cracks of land shale reservoir - Google Patents

Abstract

The embodiment of the invention provides a land shale reservoir natural fracture intelligent identification method and a system, which relate to the technical field of machine learning, and the method comprises the following steps: acquiring conventional logging data; the conventional logging data comprise depth data of sampling points of a plurality of preset logging curves and corresponding logging curve values; constructing graph data according to the depth data of the sampling points, the logging curve values and lithology recognition results; inputting the graph data into a pre-trained natural crack recognition model to obtain a natural crack recognition result corresponding to the sampling point; the natural crack identification model is obtained by training a model based on a graph convolution neural network by using corresponding natural crack labeling data of partial sampling points. The intelligent identification method and the intelligent identification system for the natural cracks of the land shale reservoir provided by the embodiment of the invention have the advantage that the accuracy of the identification of the natural cracks of the oil and gas reservoir is improved.

Description

Intelligent identification method and system for natural fractures in continental shale reservoirs

Technical field

Embodiments of the present invention relate to the field of machine learning technology, and specifically relate to an intelligent identification method and system for natural fractures in continental shale reservoirs.

Background technique

Natural fracture identification is the basis and key to the study of natural fracture distribution patterns in continental shale reservoirs, and can provide reliable geological basis for efficient exploration and development of unconventional reservoirs.

Natural fracture identification is a key issue in oil and gas exploration and development in unconventional reservoirs. The accuracy of natural fracture interpretation in core and imaging logging data is high, but the quantity is small and the cost is high, which is not conducive to understanding the development patterns of natural fractures in reservoirs in the entire region. Therefore, conventional logging curves are often used to explain the development of natural fractures in single wells. Since the logging response characteristics of natural fractures are weak and extremely complex, the interpretation accuracy of conventional logging natural fractures is low. Machine learning provides a good opportunity to solve this problem. Through machine learning methods, the nonlinear logging response characteristics of natural fractures can be better extracted, thereby improving the accuracy of identifying natural fractures in single wells.

However, the existing machine learning methods for identifying natural fractures in continental shale reservoirs still have low identification accuracy.

Contents of the invention

In view of the shortcomings of the existing technology, embodiments of the present invention provide an intelligent identification method and system for natural fractures in continental shale reservoirs.

Embodiments of the present invention provide a method for intelligent identification of natural fractures in continental shale reservoirs, which includes: obtaining conventional well logging data; wherein the conventional well logging data includes depth data of sampling points of multiple preset well logging curves and corresponding logging curve values; construct map data based on the depth data of the sampling point, the logging curve values and the lithology identification results; input the map data into a pre-trained natural fracture identification model to obtain The natural fracture identification results corresponding to the sampling points; wherein, the natural fracture identification model is based on a graph convolutional neural network and is obtained through model training with some of the natural fracture annotation data corresponding to the sampling points.

According to an intelligent identification method for natural fractures in continental shale reservoirs provided by an embodiment of the present invention, the map data is constructed based on the depth data of the sampling point, the logging curve value and the lithology identification result, including : The sampling point is used as the vertex of the graph, and the feature vector of the vertex is constructed based on the logging curve values of the multiple logging curves corresponding to the sampling point; based on the depth data, adjacent depths are Construct a first type of edge between the vertices; for the vertices corresponding to the sampling points corresponding to the same lithology identification result, construct a second type of edge between the vertices through similarity calculation; according to the constructed A graph structure is obtained from the vertices, the first type edge and the second type edge; graph data is obtained according to the graph structure and the feature vector of each vertex.

According to an intelligent identification method for natural fractures in continental shale reservoirs provided by an embodiment of the present invention, the method of constructing a second type of edge between the vertices through similarity calculation includes: calculating the Euclidean edge between two of the vertices. Reed distance; in response to the Euclidean distance being greater than the preset quantile, the second type edge is not constructed between the corresponding two vertices; in response to the Euclidean distance being less than or equal to The preset quantile is used to construct the second type edge between the corresponding two vertices.

According to an intelligent identification method for natural fractures in continental shale reservoirs provided by an embodiment of the present invention, the graph data is input into a pre-trained natural fracture identification model to obtain a natural fracture identification result corresponding to the sampling point, The method includes: inputting the graph data into a pre-trained natural crack identification model to obtain the natural crack identification result of the vertex; assigning the natural crack identification result of the vertex to the corresponding sampling point to obtain the sampling point Natural fracture identification results corresponding to points.

According to an embodiment of the present invention, an intelligent identification method for natural fractures in continental shale reservoirs is provided. The natural fracture identification model includes a multi-layer graph convolutional neural network and an output layer; the graph data input is pre-trained. A natural crack identification model to obtain the natural crack identification result of the vertex, including: inputting the graph data into the natural crack identification model, and using the graph convolutional neural network at each layer to identify the characteristics of the vertex. The vector is updated; wherein, each layer of the graph convolutional neural network performs feature vector aggregation on the neighbor vertices of the vertex, combines the aggregated feature vector with the feature vector of the vertex, and updates the characteristics of the vertex. Vector; according to the output result of the last layer of the graph convolutional neural network after being processed by the activation function of the output layer, the natural crack identification result of each of the vertices is obtained.

According to an embodiment of the present invention, a method for intelligent identification of natural fractures in continental shale reservoirs is provided. The method further includes: obtaining the natural fracture labeling data based on core description data and imaging logging data.

According to an intelligent identification method for natural fractures in continental shale reservoirs provided by an embodiment of the present invention, before constructing map data based on the depth data of the sampling point, the logging curve value and the lithology identification result , the method further includes: performing preset data preprocessing on the conventional well logging data.

According to an intelligent identification method for natural fractures in continental shale reservoirs provided by an embodiment of the present invention, the method further includes: identifying the natural fractures corresponding to the sampling points based on the depth data corresponding to the sampling points. Visual display of results.

Embodiments of the present invention also provide an intelligent identification system for natural fractures in continental shale reservoirs, including: an acquisition module, used to: acquire conventional well logging data; wherein the conventional well logging data includes a plurality of preset well logs The depth data of the sampling points of the curve and the corresponding logging curve values; a construction module, used to: construct map data according to the depth data of the sampling points, the logging curve values and the lithology identification results; an identification module, Used for: inputting the graph data into a pre-trained natural fracture identification model to obtain the natural fracture identification results corresponding to the sampling points; wherein the natural fracture identification model is based on a graph convolutional neural network, as described in part The natural fracture annotation data corresponding to the sampling points is obtained through model training.

An embodiment of the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, any one of the above-mentioned land-based operations is implemented. Steps of the intelligent identification method for natural fractures in facies shale reservoirs.

Embodiments of the present invention also provide a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the computer program can realize intelligent identification of natural fractures in any of the above-mentioned continental shale reservoirs. Method steps.

An embodiment of the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of any one of the above intelligent identification methods for natural fractures in continental shale reservoirs.

The method and system for intelligent identification of natural fractures in continental shale reservoirs provided by embodiments of the present invention obtain conventional well logging data; wherein the conventional well logging data includes depth data and corresponding sampling points of multiple preset well logging curves. According to the logging curve value of the sampling point, the graph data is constructed based on the depth data, logging curve value and lithology identification result of the sampling point. The graph data is input into the pre-trained natural fracture identification model to obtain the natural fracture identification result corresponding to the sampling point and improve Improved the accuracy of identification of natural fractures in oil and gas reservoirs.

Description of the drawings

In order to explain the technical solutions of the present invention more clearly, the drawings needed to be used in the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present invention. For those in the field, Ordinary technicians can also obtain other drawings based on these drawings without exerting creative work.

Figure 1 is one of the flow diagrams of the intelligent identification method for natural fractures in continental shale reservoirs provided by an embodiment of the present invention;

Figure 2 is a schematic diagram of the generation process of map data in the intelligent identification method for natural fractures in continental shale reservoirs provided by the embodiment of the present invention;

Figure 3 is the second schematic flow chart of the intelligent identification method for natural fractures in continental shale reservoirs provided by the embodiment of the present invention;

Figure 4 is the third schematic flow chart of the intelligent identification method for natural fractures in continental shale reservoirs provided by the embodiment of the present invention;

Figure 5 is a schematic structural diagram of an intelligent identification system for natural fractures in continental shale reservoirs provided by an embodiment of the present invention;

Figure 6 is a schematic structural diagram of an electronic device provided by an embodiment of the present invention.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present invention more clear, the technical solutions in the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention. , not all examples. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts fall within the scope of protection of the present invention.

Machine learning is a discipline that studies how to use computers to simulate human learning activities, that is, to acquire new knowledge and skills by analyzing and processing past data through machines. At this stage, methods for identifying natural fractures in single wells based on machine learning can be divided into three categories: unsupervised learning methods, supervised learning methods and semi-supervised learning methods. Unsupervised methods do not require labeled samples. Due to the lack of guidance from label information, the recognition accuracy is low. Algorithms used in identifying natural fractures in single wells include k-means, SOM neural network, Gaussian mixture model (GMM), etc.; supervised learning The method requires a large number of labeled samples for learning, and the recognition accuracy is relatively high. The algorithms used in single-well natural fracture identification include fuzzy inference system (FIS), tree model (decision tree, random forest, GBDT, XGBOOST, LightGBM), naive shell Yessian, Bayesian network, neural network (such as BP, CNN, RNN, CGAN), kernel method (such as SVM, KFD), etc.; semi-supervised learning method only requires a small number of labeled samples and is a way to solve the problem of labeled data volume. An effective method for small problems. The algorithms used in identifying natural fractures in single wells include semi-supervised support vector machines, Laplacian support vector machines, etc.

Generally, supervised learning methods are better at identifying natural cracks than unsupervised learning methods. When the number of labeled sample data is sufficient, supervised learning methods are the first choice. Semi-supervised learning methods are more suitable for model construction with limited labels. The embodiment of the present invention uses core observation and imaging logging interpretation results to obtain annotation data of natural fractures. Since the core observation and imaging logging interpretation results are limited, the embodiment of the present invention uses a semi-supervised machine learning method to develop a natural fracture identification model. of construction.

Figure 1 is one of the flow diagrams of the intelligent identification method for natural fractures in continental shale reservoirs provided by an embodiment of the present invention. As shown in Figure 1, the method includes:

Step S1: Obtain conventional well logging data; wherein the conventional well logging data includes depth data of sampling points of multiple preset well logging curves and corresponding well logging curve values.

Conventional well logging data is data obtained through conventional well logging methods. The so-called conventional logging methods mainly refer to the logging methods that are measured in oil and gas exploration and development, exploration well logging, evaluation well logging, and development well logging projects. Conventional well logging data includes multiple well logging curves, and the well logging curves include depth data of sampling points and corresponding well logging curve values. Among them, the logging curve values of different logging curves have different meanings.

In the development section of natural fractures, there are usually phenomena such as increased acoustic time difference, reduced resistivity, reduced density, increased compensation neutrons, and medium and high gamma values. However, natural fractures often have weak responses in well logging curves and their characteristic responses are different from those of fractures. The result does not correspond uniquely. On the basis of conventional well logging curves, a variety of characteristic curves that are more sensitive to the development of natural fractures can be selected and their sensitivities can be analyzed.

By analyzing the correlation between well log curves and natural fractures, in the embodiment of the present invention, acoustic transit time (AC), density log (DEN), neutron porosity (CNL), natural potential (SP), and natural gamma (GR) are selected. ), original formation resistivity (RT), invasion zone resistivity (RI), washout zone resistivity (RXO) and well diameter (CAL), a total of 9 logging curves were used to identify natural fractures.

Step S2: Construct map data based on the depth data of the sampling point, the logging curve value and the lithology identification result.

Graph data includes graph structures and feature vectors. The vertices in the graph structure correspond to sampling points in conventional well log data. Corresponding to each sampling point, it corresponds to the depth data and logging curve values of multiple preset well logging curves. The feature vector of the corresponding vertex can be constructed based on the logging curve values of multiple logging curves corresponding to the sampling point. Construct edges between sampling points based on the association between sampling points. In this way, graph data is obtained based on the vertices, edges and feature vectors of the vertices.

Step S3: Input the graph data into a pre-trained natural fracture identification model to obtain the natural fracture identification results corresponding to the sampling points; wherein, the natural fracture identification model is based on a graph convolutional neural network, as described in part The natural fracture annotation data corresponding to the sampling points is obtained through model training.

The natural fracture identification model is based on the graph convolutional neural network and is obtained through model training with the corresponding natural fracture annotation data of some sampling points.

The training process of the natural fracture identification model includes:

Construct training samples based on conventional well logging data; where conventional well logging data includes depth data and well logging curve values of preset sampling points of multiple well logging curves;

Construct map data based on the depth data, well logging curve values and lithology identification results of the sampling points;

The graph data is input into the neural network model constructed by the multi-layer graph convolutional neural network and the output layer. The natural fracture annotation data corresponding to some sampling points is used as the output label of the corresponding vertex to train the neural network model. After the training, the natural fractures are obtained. Identify the model.

Among them, during the training process, SGD or Adam can be used to iteratively optimize the error function according to the principle of minimizing the total error, thereby optimizing the parameters of the network. The grid search method can be used to find the optimal value of the hyperparameter. The basic idea is to divide the hyperparameter to be optimized into a grid within a certain spatial range, and find the optimal hyperparameter by traversing all the intersections in the grid. untie.

Rocks with different lithologies have large differences in brittleness. Usually, natural cracks are more likely to develop in areas where rocks are more brittle. Therefore, the embodiment of the present invention integrates lithological factors (lithology identification results of sampling points) into the construction process of the graph data of the natural fracture identification model, which can improve the model's ability to identify natural fractures. A graph, as a data structure, provides a clear representation of arbitrary relationships between entities. The calculation of graph data through the graph neural network algorithm can capture the topological information between different fracture segments and integrate geological understanding into the model. At the same time, the calculation of graph data has higher inductive paranoia than the calculation of original sequence data, and the training efficiency is higher, the training time is shorter, and the generalization ability is stronger.

The intelligent identification method of natural fractures in continental shale reservoirs provided by embodiments of the present invention obtains conventional well logging data; wherein the conventional well logging data includes depth data of sampling points of multiple preset well logging curves and corresponding logging data. Well curve values, map data is constructed based on the depth data, well logging curve values and lithology identification results of the sampling points. The map data is input into the pre-trained natural fracture identification model to obtain the natural fracture identification results corresponding to the sampling points, which improves the oil and gas Accuracy of identification of natural fractures in reservoirs.

According to an intelligent identification method for natural fractures in continental shale reservoirs provided by an embodiment of the present invention, the map data is constructed based on the depth data of the sampling point, the logging curve value and the lithology identification result, including : The sampling point is used as a vertex of the graph structure, and the feature vector of the vertex is constructed according to the logging curve values of the plurality of logging curves corresponding to the sampling point; according to the depth data, in the depth phase A first type of edge is constructed between adjacent vertices; for the vertices corresponding to the sampling points corresponding to the same lithology identification result, a second type of edge is constructed between the vertices through similarity calculation; according to the constructed The vertex, the first type edge and the second type edge are used to obtain a graph structure; graph data is obtained according to the graph structure and the feature vector of each of the vertices.

The construction of the graph structure needs to meet two conditions: (1) the constructed graph data set must have certain geological significance; (2) it must comply with the underlying logic of the graph convolution operation and improve the accuracy of natural fracture identification as much as possible.

Figure 2 is a schematic diagram of the generation process of map data in the intelligent identification method for natural fractures in continental shale reservoirs provided by the embodiment of the present invention. Based on the above two points, combined with Figure 2, the construction of the graph structure includes: (1) Vertex generation, taking the conventional well logging curve sampling point as the vertex, and the logging curve value corresponding to each vertex as the feature vector of the vertex, and the well logging curve The number of entries corresponds to the dimension of the vector, and the feature vectors of each vertex constitute the feature matrix. (2) Edge generation, depth-adjacent vertices are connected by edges to obtain the first type of edge (sequence edge in Figure 2); secondly, for the vertices corresponding to the sampling points corresponding to the same lithology identification result, the similarity is calculated in The second type of edge (distance edge in Figure 2) is constructed between vertices. The edges generated above are undirected edges, which together with all vertices and corresponding feature vectors constitute graph data.

It should be noted that the first type of edge and the second type of edge are only different in the construction methods used to represent edges.

The intelligent identification method for natural fractures in continental shale reservoirs provided by embodiments of the present invention uses the sampling point as the vertex of the graph structure, and constructs the feature vector of the vertex based on the logging curve values of multiple logging curves corresponding to the sampling point; Depth data, construct the first type of edge between depth-adjacent vertices; for the vertices corresponding to the sampling points corresponding to the same lithology identification result, construct the second type of edge between the vertices through similarity calculation; according to the constructed vertices, the third The first type of edges and the second type of edges obtain the graph structure; the graph data is obtained according to the graph structure and the characteristic vectors of each vertex, which improves the adaptability of the graph data for fracture identification, thereby further improving the accuracy of natural fracture identification in oil and gas reservoirs. sex.

According to an intelligent identification method for natural fractures in continental shale reservoirs provided by an embodiment of the present invention, constructing a second type of edge between the sampling points through similarity calculation includes: calculating the Euclidean distance between two of the vertices. metric distance; in response to the Euclidean distance being greater than a preset quantile, the second type edge is not constructed between the corresponding two vertices; in response to the Euclidean distance being less than or is equal to the preset quantile, then the second type edge is constructed between the corresponding two vertices.

When constructing a second type of edge between vertices through similarity calculation, calculate the Euclidean distance between two vertices and set a reasonable quantile. If the Euclidean distance is greater than the preset quantile, then The second type edge is not constructed between the corresponding two vertices. If the Euclidean distance is less than or equal to the preset quantile, the second type edge is constructed between the corresponding two vertices.

The intelligent identification method for natural fractures in continental shale reservoirs provided by embodiments of the present invention calculates the Euclidean distance between two vertices. In response to the Euclidean distance being greater than the preset quantile, the corresponding two vertices will not be identified. A second type of edge is constructed between vertices. In response to the Euclidean distance being less than or equal to the preset quantile, a second type of edge is constructed between the corresponding two vertices, thereby refining the relationship network between nodes, thereby The accuracy of identification of natural fractures in oil and gas reservoirs is further improved.

According to an intelligent identification method for natural fractures in continental shale reservoirs provided by an embodiment of the present invention, the graph data is input into a pre-trained natural fracture identification model to obtain a natural fracture identification result corresponding to the sampling point, The method includes: inputting the graph data into a pre-trained natural crack identification model to obtain the natural crack identification result of the vertex; assigning the natural crack identification result of the vertex to the corresponding sampling point to obtain the sampling point Natural fracture identification results corresponding to points.

When the graph data is input into the pre-trained natural fracture identification model to obtain the natural fracture identification results corresponding to the sampling points, the graph data is input into the pre-trained natural fracture identification model to obtain the natural fracture identification results for each vertex in the graph data. Since the vertices correspond to the sampling points, the natural fracture identification results of each vertex are assigned to the corresponding sampling points, thereby obtaining the natural fracture identification results corresponding to each sampling point.

The intelligent identification method for natural fractures in continental shale reservoirs provided by embodiments of the present invention obtains the natural fracture identification results at the vertex by inputting the graph data into the pre-trained natural fracture identification model; assigns the natural fracture identification results at the vertex to the corresponding Sampling point, and obtain the natural fracture identification result corresponding to the sampling point, realizing the rapid acquisition of the natural fracture identification result corresponding to the sampling point.

According to an embodiment of the present invention, an intelligent identification method for natural fractures in continental shale reservoirs is provided. The natural fracture identification model includes a multi-layer graph convolutional neural network and an output layer; the graph data input is pre-trained. A natural crack identification model to obtain the natural crack identification result of the vertex, including: inputting the graph data into the natural crack identification model, and using the graph convolutional neural network at each layer to identify the characteristics of the vertex. The vector is updated; wherein, each layer of the graph convolutional neural network performs feature vector aggregation on the neighbor vertices of the vertex, combines the aggregated feature vector with the feature vector of the vertex, and updates the characteristics of the vertex. Vector; according to the output result of the last layer of the graph convolutional neural network after being processed by the activation function of the output layer, the natural crack identification result of each of the vertices is obtained.

The structure of the crack identification model includes a multi-layer graph convolutional neural network and an output layer. Among them, for each layer of graph convolutional neural network, the aggregated feature vector is obtained by aggregating the feature vectors of the neighbor vertices of the vertex, and the aggregated feature vector is combined with the feature vector of the vertex to update the feature vector of the vertex. . Among them, the aggregation function can be used to aggregate the feature vectors of the neighbor vertices of the vertex, and the feature vector of the vertex is updated through the nonlinear activation function.

Therefore, when the graph data is input into the crack identification model and the natural fracture identification results corresponding to the vertices are output, the graph data is input into the crack identification model, and the feature vectors of the vertices are updated through each layer of the graph convolutional neural network, and the last layer The output of the graph convolutional neural network is processed by the activation function of the output layer and then outputs a probability value. The natural fracture identification result is obtained based on the output probability value. For example, during training, if there are cracks, the label is 1, and if there are no cracks, the label is 0. If there are cracks, the output probability value is close to 1; if there are no cracks, the output probability value is close to 0.

The feature vector of a vertex includes the measured curve value of the corresponding depth data. The feature vector of each vertex in the graph data can be expressed as a feature matrix. When the graph convolutional neural network of each layer updates the feature vectors of the vertices, the feature vectors of the vertices are updated based on the neighbor relationships between the vertices (that is, the graph structure in the graph data), that is, the feature matrix and the graph structure are implemented. information fusion.

Figure 3 is the second schematic flow chart of the intelligent identification method for natural fractures in continental shale reservoirs provided by an embodiment of the present invention. As shown in Figure 3, after inputting the graph data into the crack recognition model, the graph convolution layer (graph convolutional neural network) is used to integrate the graph structure and feature matrix for the first time, that is, the embedding process, to obtain the embedding matrix. The embedding matrix After being activated by the activation function, it is used as the input data of the next layer, and the second information integration is performed to obtain the next embedding matrix; after n updates, the final embedding matrix that needs to be output is obtained, and the embedding matrix is passed through the softmax function of the output layer. The classification result of the vertex is calculated, that is, the classification result of whether the crack identification result corresponding to the vertex has cracks or no cracks. Putting the above process at the micro-node scale, each embedding process, also known as the information transfer process, is divided into the aggregation of node features and the update of node features.

Among them, the graph neural network can use GraphSAGE, and of course other graph neural networks can also be used.

The intelligent identification method for natural fractures in continental shale reservoirs provided by embodiments of the present invention inputs graph data into the natural fracture identification model, and updates the feature vectors of vertices through each layer of graph convolutional neural networks, where each layer of graph The convolutional neural network aggregates the feature vectors of the neighbor vertices of the vertex, combines the aggregated feature vector with the feature vector of the vertex, updates the feature vector of the vertex, and passes the activation function of the output layer according to the last layer of graph convolutional neural network The processed output results obtain the natural fracture identification results of each vertex, further improving the accuracy of natural fracture identification in oil and gas reservoirs.

According to an embodiment of the present invention, a method for intelligent identification of natural fractures in continental shale reservoirs is provided. The method further includes: obtaining the natural fracture labeling data based on core description data and imaging logging data.

Natural fracture interpretation can be performed on core description data and imaging logging data, natural fracture conditions can be calibrated, and natural fracture labeling data corresponding to some sampling points can be obtained.

Embodiments of the present invention improve the accuracy of the labeling data by obtaining natural fracture labeling data based on core description data and imaging logging data.

According to an intelligent identification method for natural fractures in continental shale reservoirs provided by an embodiment of the present invention, before obtaining the lithology identification result of the sampling point, the method further includes: performing a test on the conventional well logging data. Default data preprocessing.

Since the numerical ranges of different conventional well logging curves are significantly different, before obtaining the lithology identification results of the sampling points, the conventional well logging data must be cleaned, unit standardized, depth corrected, data aligned and interpolated, detrended, and outliers processed. Pre-processing of at least one of standardization and normalization, as well as data preservation and backup, ensures the quality of logging data.

When training the crack identification model, the same data preprocessing process can be performed for the training samples.

The intelligent identification method for natural fractures in continental shale reservoirs provided by embodiments of the present invention improves the quality of conventional logging data by performing preset data preprocessing on conventional logging data, thereby improving the accuracy of natural fracture identification.

According to an intelligent identification method for natural fractures in continental shale reservoirs provided by an embodiment of the present invention, the method further includes: identifying the natural fractures corresponding to the sampling points based on the depth data corresponding to the sampling points. Visual display of results.

After obtaining the natural fracture identification results corresponding to the sampling points, the natural fracture identification results corresponding to the sampling points are visually displayed based on the depth data corresponding to the sampling points.

The intelligent identification method for natural fractures in continental shale reservoirs provided by embodiments of the present invention realizes the intuitive display of natural fracture identification results by visually displaying the natural fracture identification results corresponding to the sampling points based on the depth data corresponding to the sampling points.

Figure 4 is the third schematic flowchart of the intelligent identification method for natural fractures in continental shale reservoirs provided by the embodiment of the present invention. A specific embodiment is given below in conjunction with Figure 4.

The Fengcheng Formation in the Mahu Sag of the Junggar Basin was selected as an example to identify natural fractures. Nine conventional well logging curves of GR, CAL, SP, AC, CNL, DEN, RXO, Rt, and Ri were selected from 2 typical wells. Perform data standardization processing on conventional logging curves, and use core observation and imaging logging interpretation results to calibrate the processed conventional logging data. The vertices of the graph structure are generated based on the sampling points. The first type of edges is constructed based on the depth data sequence corresponding to the sampling points. The second type of edges is constructed by calculating the Euclidean distance between the same lithology logging sampling points. According to the first type Edges and second type edges are generated to obtain the global graph (graph data). The crack identification model (trained based on GraphSAGE) is used to classify the vertices of the constructed global graph to realize natural crack identification. Among them, the process of using the crack identification model to classify the vertices of the constructed global graph includes updating the embedding representation (feature vector) of the vertices and identifying natural cracks based on the softmax function normalization results.

The results show that the intelligent identification method of natural fractures in continental shale reservoirs provided by the embodiments of the present invention can better identify the sections with developed natural fractures, and can accurately identify natural fractures in oil and gas reservoirs.

It should be noted that the multiple preferred implementation modes provided in this embodiment can be freely combined as long as the logic or structure does not conflict with each other, and the present invention does not limit this.

The intelligent identification system for natural fractures in continental shale reservoirs provided by embodiments of the present invention is described below. The intelligent identification system for natural fractures in continental shale reservoirs described below is the same as the intelligent identification system for natural fractures in continental shale reservoirs described above. Methods can be compared to each other.

Figure 5 is a schematic structural diagram of an intelligent identification system for natural fractures in continental shale reservoirs provided by an embodiment of the present invention. As shown in Figure 5, the device includes an acquisition module 10, a construction module 20 and an identification module 30, wherein: the acquisition module 10 is used to: acquire conventional well logging data; wherein the conventional well logging data includes a plurality of preset logging data. The depth data of the sampling points of the well curve and the corresponding well log curve values; the construction module 20 is used to: construct map data according to the depth data of the sampling points, the well log curve values and the lithology identification results; the identification module 30 is used for: inputting the graph data into a pre-trained natural fracture identification model to obtain the natural fracture identification results corresponding to the sampling points; wherein the natural fracture identification model is based on a graph convolutional neural network, using some of the The natural fracture annotation data corresponding to the above sampling points was obtained through model training.

The intelligent identification system for natural fractures in continental shale reservoirs provided by embodiments of the present invention obtains conventional well logging data; wherein the conventional well logging data includes depth data of sampling points of multiple preset well logging curves and corresponding logging data. Well curve values, map data is constructed based on the depth data, well logging curve values and lithology identification results of the sampling points. The map data is input into the pre-trained natural fracture identification model to obtain the natural fracture identification results corresponding to the sampling points, which improves the oil and gas Accuracy of identification of natural fractures in reservoirs.

According to an intelligent identification system for natural fractures in continental shale reservoirs provided in an embodiment of the present invention, the construction module 20 is used to construct a system based on the depth data of the sampling point, the logging curve value and the lithology identification result. When graph data is used, it is specifically used to: use the sampling point as a vertex of the graph structure, and construct a feature vector of the vertex according to the logging curve values of the plurality of logging curves corresponding to the sampling point; The depth data is used to construct a first type of edge between the vertices adjacent in depth; for the vertices corresponding to the sampling points corresponding to the same lithology identification result, a second type of edge is constructed between the vertices through similarity calculation. type edge; obtain a graph structure based on the constructed vertices, the first type edge, and the second type edge; obtain graph data based on the graph structure and the feature vectors of each of the vertices.

The intelligent identification system for natural fractures in continental shale reservoirs provided by embodiments of the present invention uses the sampling point as the vertex of the graph structure, and constructs the feature vector of the vertex based on the logging curve values of multiple logging curves corresponding to the sampling point; Depth data, construct the first type of edge between depth-adjacent vertices; for the vertices corresponding to the sampling points corresponding to the same lithology identification result, construct the second type of edge between the vertices through similarity calculation; according to the constructed vertices, the third The first type of edges and the second type of edges obtain the graph structure; the graph data is obtained according to the graph structure and the characteristic vectors of each vertex, which improves the adaptability of the graph data for fracture identification, thereby further improving the accuracy of natural fracture identification in oil and gas reservoirs. sex.

According to an intelligent identification system for natural fractures in continental shale reservoirs provided by an embodiment of the present invention, when the building module 20 is used to construct a second type of edge between the vertices through similarity calculation, it is specifically used to: calculate the pairwise The Euclidean distance between the vertices; in response to the Euclidean distance being greater than the preset quantile, the second type edge is not constructed between the corresponding two vertices; in response to the Euclidean distance If the metric distance is less than or equal to the preset quantile, then the second type edge is constructed between the corresponding two vertices.

The intelligent identification system for natural fractures in continental shale reservoirs provided by embodiments of the present invention calculates the Euclidean distance between two vertices. In response to the Euclidean distance being greater than the preset quantile, the corresponding two vertices will not be identified. A second type of edge is constructed between vertices. In response to the Euclidean distance being less than or equal to the preset quantile, a second type of edge is constructed between the corresponding two vertices, thereby refining the relationship network between nodes, thereby The accuracy of identification of natural fractures in oil and gas reservoirs is further improved.

According to an intelligent identification system for natural fractures in continental shale reservoirs provided in an embodiment of the present invention, the identification module 30 is used to input the graph data into a pre-trained natural fracture identification model to obtain the natural fractures corresponding to the sampling points. The crack identification result is specifically used to: input the graph data into a pre-trained natural crack identification model to obtain the natural crack identification result of the vertex; assign the natural crack identification result of the vertex to the corresponding sampling points, and obtain the natural fracture identification results corresponding to the sampling points.

The intelligent identification system for natural fractures in continental shale reservoirs provided by embodiments of the present invention obtains the natural fracture identification results at the vertex by inputting the graph data into the pre-trained natural fracture identification model; assigns the natural fracture identification results at the vertex to the corresponding Sampling point, and obtain the natural fracture identification result corresponding to the sampling point, realizing the rapid acquisition of the natural fracture identification result corresponding to the sampling point.

According to an intelligent identification system for natural fractures in continental shale reservoirs provided by an embodiment of the present invention, the natural fracture identification model includes a multi-layer graph convolutional neural network and an output layer; the identification module 30 is used to convert the graph data When the pre-trained natural crack recognition model is input and the natural crack recognition result of the vertex is obtained, it is specifically used to: input the graph data into the natural crack recognition model, and use the graph convolutional neural network at each layer to The feature vector of the vertex is updated; wherein, each layer of the graph convolutional neural network performs feature vector aggregation on the neighbor vertices of the vertex, and combines the aggregated feature vector with the feature vector of the vertex. , update the feature vector of the vertex; obtain the natural crack identification result of each vertex according to the output result of the last layer of the graph convolutional neural network processed by the activation function of the output layer.

The intelligent identification system for natural fractures in continental shale reservoirs provided by embodiments of the present invention updates the feature vectors of vertices through each layer of graph convolutional neural networks by inputting graph data into the natural fracture identification model, where each layer of graph The convolutional neural network aggregates the feature vectors of the neighbor vertices of the vertex, combines the aggregated feature vector with the feature vector of the vertex, updates the feature vector of the vertex, and passes the activation function of the output layer according to the last layer of graph convolutional neural network The processed output results obtain the natural fracture identification results of each vertex, further improving the accuracy of natural fracture identification in oil and gas reservoirs.

According to an intelligent identification system for natural fractures in continental shale reservoirs provided by an embodiment of the present invention, the device further includes a labeling data acquisition module for: obtaining the natural fracture labeling data based on core description data and imaging logging data. .

Embodiments of the present invention improve the accuracy of the labeling data by obtaining natural fracture labeling data based on core description data and imaging logging data.

According to an intelligent identification system for natural fractures in continental shale reservoirs provided by an embodiment of the present invention, the device further includes a preprocessing module. In the building module 20, the depth data of the sampling point and the well logging curve are Before constructing map data using the values and lithology identification results, the preprocessing module is used to perform preset data preprocessing on the conventional logging data.

The intelligent identification system for natural fractures in continental shale reservoirs provided by embodiments of the present invention improves the quality of conventional well logging data by performing preset data preprocessing on conventional well logging data, thereby improving the accuracy of natural fracture identification.

According to an intelligent identification system for natural fractures in continental shale reservoirs provided by an embodiment of the present invention, the device further includes a display module for: according to the depth data corresponding to the sampling point, the depth data corresponding to the sampling point is displayed. The natural fracture identification results are visually displayed.

The intelligent identification system for natural fractures in continental shale reservoirs provided by embodiments of the present invention realizes the intuitive display of natural fracture identification results by visually displaying the natural fracture identification results corresponding to the sampling points based on the depth data corresponding to the sampling points.

Figure 6 is a schematic structural diagram of an electronic device provided by an embodiment of the present invention. As shown in Figure 6, the electronic device may include: a processor (processor) 410, a communications interface (Communications Interface) 420, a memory (memory) 430 and a communication bus. 440, in which the processor 410, the communication interface 420, and the memory 430 complete communication with each other through the communication bus 440. The processor 410 can call logical instructions in the memory 430 to execute a method for intelligent identification of natural fractures in continental shale reservoirs. The method includes: obtaining conventional well logging data; wherein the conventional well logging data includes a plurality of preset Depth data of the sampling points of the well logging curve and corresponding well logging curve values; construct map data based on the depth data of the sampling points, the well log curve values and the lithology identification results; input the map data in advance The trained natural fracture identification model obtains the natural fracture identification results corresponding to the sampling points; wherein, the natural fracture identification model is based on a graph convolutional neural network, and the natural fracture annotation data corresponding to some of the sampling points is passed through the model Obtained by training.

In addition, the above-mentioned logical instructions in the memory 430 can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present invention essentially or the part that contributes to the existing technology or the part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including Several instructions are used to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program code. .

On the other hand, embodiments of the present invention also provide a computer program product. The computer program product includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, The computer can execute the intelligent identification method of natural fractures in continental shale reservoirs provided by each of the above methods. The method includes: obtaining conventional well logging data; wherein the conventional well logging data includes sampling of multiple preset well logging curves. The depth data of the point and the corresponding well logging curve value; construct map data based on the depth data of the sampling point, the well log curve value and the lithology identification result; input the map data into the pre-trained natural fractures Identify the model to obtain the natural fracture identification results corresponding to the sampling points; wherein the natural fracture identification model is based on a graph convolutional neural network and is obtained through model training with the natural fracture annotation data corresponding to some of the sampling points.

On the other hand, embodiments of the present invention also provide a non-transitory computer-readable storage medium on which a computer program is stored. The computer program is implemented when executed by a processor to execute the continental shale reservoir provided by each of the above methods. An intelligent identification method for natural fractures, which method includes: obtaining conventional well logging data; wherein the conventional well logging data includes preset depth data of sampling points of multiple well logging curves and corresponding well logging curve values; according to the The depth data of the sampling point, the well logging curve value and the lithology identification result construct map data; input the map data into a pre-trained natural fracture identification model to obtain the natural fracture identification result corresponding to the sampling point; Wherein, the natural fracture identification model is based on a graph convolutional neural network and is obtained through model training with the natural fracture annotation data corresponding to some of the sampling points.

The device embodiments described above are only illustrative. The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they may be located in One location, or it can be distributed across multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Persons of ordinary skill in the art can understand and implement the method without any creative effort.

Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the part of the above technical solution that essentially contributes to the existing technology can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM/RAM, magnetic disc, optical disk, etc., including a number of instructions to cause a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in various embodiments or certain parts of the embodiments.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that it can still be used Modifications are made to the technical solutions described in the foregoing embodiments, or equivalent substitutions are made to some of the technical features; however, these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims (11)

1. An intelligent identification method for natural fractures in continental shale reservoirs, which is characterized by including: Obtain conventional well logging data; wherein the conventional well logging data includes depth data of sampling points of multiple preset well logging curves and corresponding well logging curve values; Construct map data based on the depth data of the sampling point, the logging curve value and the lithology identification result; Input the graph data into a pre-trained natural fracture identification model to obtain the natural fracture identification results corresponding to the sampling points; wherein, the natural fracture identification model is based on a graph convolutional neural network, with some of the sampling points corresponding to The natural fracture annotation data is obtained through model training. 2. The intelligent identification method of natural fractures in continental shale reservoirs according to claim 1, characterized in that the method is constructed based on the depth data of the sampling point, the logging curve value and the lithology identification result. Graph data, including: Use the sampling point as a vertex of the graph structure, and construct a feature vector of the vertex based on the logging curve values of the multiple well logging curves corresponding to the sampling point; Constructing a first type edge between the depth-adjacent vertices according to the depth data; For the vertices corresponding to the sampling points corresponding to the same lithology identification result, construct a second type of edge between the vertices through similarity calculation; Obtain a graph structure according to the constructed vertices, the first type of edges and the second type of edges; Graph data is obtained according to the graph structure and the feature vectors of each of the vertices. 3. The intelligent identification method of natural fractures in continental shale reservoirs according to claim 2, characterized in that the second type of edge is constructed between the vertices through similarity calculation, including: Calculate the Euclidean distance between the two vertices; In response to the Euclidean distance being greater than a preset quantile, the second type edge is not constructed between the corresponding two vertices; In response to the Euclidean distance being less than or equal to the preset quantile, the second type edge is constructed between the corresponding two vertices. 4. The method for intelligent identification of natural fractures in continental shale reservoirs according to claim 2, characterized in that the map data is input into a pre-trained natural fracture identification model to obtain the natural fractures corresponding to the sampling points. Crack identification results include: Input the graph data into a pre-trained natural crack recognition model to obtain the natural crack recognition result of the vertex; The natural fracture identification result of the vertex is assigned to the corresponding sampling point, and the natural fracture identification result corresponding to the sampling point is obtained. 5. The intelligent identification method of natural fractures in continental shale reservoirs according to claim 4, characterized in that the natural fracture identification model includes a multi-layer graph convolutional neural network and an output layer; The method of inputting the graph data into a pre-trained natural crack recognition model to obtain the natural crack recognition result of the vertex includes: The graph data is input into the natural fracture identification model, and the feature vectors of the vertices are updated through the graph convolutional neural network at each layer; wherein, the graph convolutional neural network at each layer is updated by Perform feature vector aggregation on the neighbor vertices of the vertex, combine the aggregated feature vector with the feature vector of the vertex, and update the feature vector of the vertex; According to the output result of the last layer of the graph convolutional neural network after being processed by the activation function of the output layer, the natural crack identification result of each of the vertices is obtained. 6. The method for intelligent identification of natural fractures in continental shale reservoirs according to claim 1, characterized in that the method further includes: The natural fracture labeling data is obtained based on core description data and imaging logging data. 7. The intelligent identification method of natural fractures in continental shale reservoirs according to claim 1, characterized in that, based on the depth data of the sampling point, the logging curve value and the lithology identification result Before constructing the graph data, the method also includes: Perform preset data preprocessing on the conventional well logging data. 8. The method for intelligent identification of natural fractures in continental shale reservoirs according to claim 1, characterized in that the method further includes: The natural fracture identification results corresponding to the sampling points are visually displayed based on the depth data corresponding to the sampling points. 9. An intelligent recognition system for natural fractures in continental shale reservoirs, which is characterized by including: An acquisition module, configured to: acquire conventional well logging data; wherein the conventional well logging data includes preset depth data of sampling points of multiple well logging curves and corresponding well logging curve values; A construction module, configured to: construct map data based on the depth data of the sampling point, the logging curve value and the lithology identification result; An identification module, configured to: input the graph data into a pre-trained natural fracture identification model to obtain the natural fracture identification results corresponding to the sampling points; wherein the natural fracture identification model is based on a graph convolutional neural network. The natural fracture annotation data corresponding to some of the sampling points was obtained through model training. 10. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that when the processor executes the program, it implements claim 1 Go to the steps of the intelligent identification method for natural fractures in continental shale reservoirs described in any one of 8. 11. A non-transitory computer-readable storage medium with a computer program stored thereon, characterized in that when the computer program is executed by a processor, the continental shale reservoir according to any one of claims 1 to 8 is realized. Steps of the intelligent identification method for natural cracks in layers.