Method and device for generating three-dimensional geological model based on geological prediction network

Abstract

The present application proposes a method and device for generating a three-dimensional geological model based on a geological prediction network, wherein the method includes: obtaining geological drilling data and / or geological profile data of the area to be modeled, and obtaining a sparse geological attribute point cloud through a geological attribute point cloud generation method; inputting the sparse geological attribute point cloud into the geological prediction network for processing, and the geological prediction network outputs a dense geological attribute point cloud; performing isosurface extraction on the dense geological attribute point cloud, and generating a three-dimensional geological model of the area to be modeled through geometric topological processing. This method can generate a three-dimensional geological model online and in real time by using geological drilling data and / or geological profile data and a geological prediction network, under the guarantee of hardware computing power.

Description

Technical Field

[0001] The present application relates to the field of coal mine technology, and in particular to a method and device for generating a three-dimensional geological model based on a geological prediction network. Background Art

[0002] With the development of information science and technology and its application in the field of geology, 3D geological modeling technology has been proposed and developed. Existing 3D geological modeling technology generally has problems such as large workload of human-computer interaction, low modeling efficiency, and difficulty in online real-time updating of models. Summary of the invention

[0003] This application proposes a method and device for generating a three-dimensional geological model based on a geological prediction network. The specific scheme is as follows:

[0004] In one aspect, an embodiment of the present application provides a method for generating a three-dimensional geological model based on a geological prediction network, comprising:

[0005] Acquire geological drilling data and / or geological profile data of the area to be modeled, and obtain a sparse geological attribute point cloud by using a geological attribute point cloud generation method;

[0006] Inputting the sparse geological attribute point cloud into a geological prediction network for processing, the geological prediction network outputting a dense geological attribute point cloud;

[0007] The dense geological attribute point cloud is subjected to isosurface extraction, and a three-dimensional geological model of the area to be modeled is generated through geometric topological processing.

[0008] In a possible implementation of an embodiment of one aspect of the present application, the sparse geological attribute point cloud is generated by the following steps:

[0009] voxelize the geological drilling data, the geological profile data, and four geological models of strata, faults, folds, and collapse columns to form corresponding voxel models;

[0010] Assigning corresponding geological attributes to each voxel in the voxel model according to the geological drilling data, the geological profile data, and the geological attribute information contained in the four geological models;

[0011] The sparse geological attribute point cloud is generated according to the center point of each voxel in the voxel model and the attribute value of the geological attribute.

[0012] In a possible implementation of an embodiment of one aspect of the present application, the geological prediction network is implemented by the following steps:

[0013] Acquire a training data set corresponding to the sample area, wherein the training data set includes sample geological attribute point clouds corresponding to four geological models of strata, faults, folds and collapse columns in the sample area respectively;

[0014] Using the sample geological attribute point cloud, the initial geological prediction network is trained by back propagation until the similarity between the geological attribute point cloud predicted and generated by the trained geological prediction network and the sample geological attribute point cloud is greater than a first threshold. In a possible implementation of an embodiment of one aspect of the present application, the back propagation training of the initial geological prediction network is implemented by the following steps:

[0015] Extracting a virtual drilling point cloud from the sample geological attribute point cloud, and inputting it into the initial geological prediction network for feature extraction to obtain a point cloud global feature vector;

[0016] Generate a predicted geological attribute point cloud using the point cloud global feature vector;

[0017] Determining a chamfer distance between the predicted geological attribute point cloud and the sample stratum geological attribute point cloud;

[0018] When the chamfer distance is greater than a threshold or the predicted geological attribute point cloud does not conform to geological laws, the parameters of the initial geological prediction network are adjusted through the back propagation training until the obtained chamfer distance is less than or equal to a second threshold and the predicted geological attribute point cloud conforms to geological laws, thereby generating the geological prediction network.

[0019] In a possible implementation of an embodiment of one aspect of the present application, the geological attribute is one of lithology, porosity, and mine pressure.

[0020] Another aspect of the present application provides a three-dimensional geological model generation device based on a geological prediction network, comprising:

[0021] The first acquisition module is used to acquire geological drilling data and / or geological profile data of the area to be modeled, and obtain a sparse geological attribute point cloud by a geological attribute point cloud generation method;

[0022] A second acquisition module is used to input the sparse geological attribute point cloud into a geological prediction network for processing, and the geological prediction network outputs a dense geological attribute point cloud;

[0023] The generation module is used to extract isosurfaces from the dense geological attribute point cloud and generate a three-dimensional geological model of the area to be modeled through geometric topology processing.

[0024] In a possible implementation manner of another aspect of the present application, the first acquisition module is used to:

[0025] voxelize the geological drilling data, the geological profile data, and four geological models of strata, faults, folds, and collapse columns to form corresponding voxel models;

[0026] Assigning corresponding geological attributes to each voxel in the voxel model according to the geological drilling data, the geological profile data, and the geological attribute information contained in the four geological models;

[0027] The sparse geological attribute point cloud is generated according to the center point of each voxel in the voxel model and the attribute value of the geological attribute.

[0028] In a possible implementation of another aspect of the present application, the present invention further includes:

[0029] The third acquisition module is used to acquire a training data set corresponding to the sample area, wherein the training data set includes sample geological attribute point clouds corresponding to four geological models of strata, faults, folds and collapse columns in the sample area respectively;

[0030] The training module is used to use the sample geological attribute point cloud to train the initial geological prediction network through back propagation until the similarity between the geological attribute point cloud predicted and generated by the trained geological prediction network and the sample geological attribute point cloud is greater than a first threshold.

[0031] In a possible implementation of another aspect of the present application, the training module is used to:

[0032] Extracting a virtual drilling point cloud from the sample geological attribute point cloud, and inputting it into the initial geological prediction network for feature extraction to obtain a point cloud global feature vector;

[0033] Generate a predicted geological attribute point cloud using the point cloud global feature vector;

[0034] Determining a chamfer distance between the predicted geological attribute point cloud and the sample stratum geological attribute point cloud;

[0035] When the chamfer distance is greater than a threshold or the predicted geological attribute point cloud does not conform to geological laws, the parameters of the initial geological prediction network are adjusted through the back propagation training until the obtained chamfer distance is less than or equal to a second threshold and the predicted geological attribute point cloud conforms to geological laws, thereby generating the geological prediction network.

[0036] In a possible implementation of another aspect of the present application, the geological attribute is one of lithology, porosity, and ore pressure.

[0037] Another aspect of the present application provides a computer device, including a processor and a memory;

[0038] The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to implement the method described in the above embodiment.

[0039] Another aspect of the present application provides a computer-readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method described in the above embodiment.

[0040] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0042] Figure 1 A schematic diagram of a process flow of a method for generating a three-dimensional geological model based on a geological prediction network provided in an embodiment of the present application;

[0043] Figure 2 A schematic diagram of a process for generating a geological attribute point cloud provided in an embodiment of the present application;

[0044] Figure 3 A schematic diagram of a flow chart of a geological prediction network training method provided in an embodiment of the present application;

[0045] Figure 4 A schematic diagram of a process of modeling a three-dimensional geological model of a coal mine based on a geological prediction network provided in an embodiment of the present application;

[0046] Figure 5 A schematic structural diagram of a three-dimensional geological model generating device based on a geological prediction network provided in an embodiment of the present application. DETAILED DESCRIPTION

[0047] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0048] The following describes a three-dimensional geological modeling method based on a geological prediction network according to an embodiment of the present application with reference to the accompanying drawings.

[0049] Figure 1 A schematic flow chart of a method for generating a three-dimensional geological model based on a geological prediction network is provided in an embodiment of the present application.

[0050] The three-dimensional geological model generation method based on the geological prediction network of the embodiment of the present application is executed by the three-dimensional geological model generation device based on the geological prediction network provided by the embodiment of the present application. The device can be configured in a computer device to realize the real-time generation of a three-dimensional geological model using the geological prediction network. The modeling efficiency is high, and it can match the mechanized coal mining and cutting speed to meet the dynamic planning and cutting requirements of the coal mining machine.

[0051] like Figure 1 As shown, the three-dimensional geological model generation method based on the geological prediction network includes:

[0052] Step 101, obtaining geological drilling data and / or geological profile data of the area to be modeled, and obtaining a sparse geological attribute point cloud by a geological attribute point cloud generation method.

[0053] In the present application, the area to be modeled may be a coal mining area, and the geological drilling data of the area to be modeled may include the location of the borehole in the area to be modeled, the borehole depth, the geological object through which the borehole passes, etc. Among them, the geological objects may include strata, faults, folds, collapse columns, etc.

[0054] In the present application, the geological drilling data and / or geological profile data of the area to be modeled may be input by a user, may be read from a file, etc., or may be acquired by other means.

[0055] After obtaining geological drilling data and / or geological profile data of the area to be modeled, the geological drilling data and / or geological profile data of the area to be modeled may be processed using a geological attribute point cloud generation method to obtain a sparse geological attribute point cloud.

[0056] Among them, the sparse geological attribute point cloud can include the spatial coordinate information of the point, the attribute value of the geological attribute, etc. The sparse geological attribute point cloud can be represented by a quaternion vector. For example, the spatial coordinates of the point are (x, y, z), and the attribute value of the geological attribute of the point is r, then it can be represented by a quaternion vector (x, y, z, r).

[0057] The geological attribute may be one of lithology, porosity, ore pressure, etc. For example, the geological attribute is lithology, and a sparse lithology point cloud may be obtained by a geological attribute point cloud generation method.

[0058] Step 102, input the sparse geological attribute point cloud into the geological prediction network for processing, and the geological prediction network outputs a dense geological attribute point cloud.

[0059] In this application, a geological prediction network can be pre-trained and deployed online. After obtaining the sparse geological attribute point cloud corresponding to the area to be modeled, the sparse geological attribute point cloud can be input into the geological prediction network, and the geological prediction network can complete the sparse geological attribute points and output a dense geological attribute point cloud.

[0060] Step 103, extracting isosurfaces from the dense geological attribute point cloud, and generating a three-dimensional geological model of the area to be modeled through geometric topology processing.

[0061] In this application, isosurface extraction can be performed on dense geological attribute point clouds, points with the same attributes can be classified into the same point cloud, the point cloud can be processed by MachingCube to generate geological interfaces, and the geological interfaces can be combined to form a three-dimensional geological model of the area to be modeled.

[0062] In the embodiment of the present application, by obtaining geological drilling data and / or geological profile data of the area to be modeled, and through the geological attribute point cloud generation method, a sparse geological attribute point cloud is obtained; the sparse geological attribute point cloud is input into the geological prediction network for processing, and the geological prediction network outputs a dense geological attribute point cloud; the dense geological attribute point cloud is extracted from the isosurface, and a three-dimensional geological model of the area to be modeled is generated through geometric topology processing. Thus, by inputting the sparse geological attribute point cloud of the area to be modeled into the geological prediction network, a dense geological attribute point cloud can be obtained, and a three-dimensional geological model of the area to be modeled is generated according to the dense geological attribute point cloud, so that the geological prediction network can be used to generate a three-dimensional geological model in real time, and the modeling efficiency is high, which can match the mechanized coal mining cutting speed and meet the dynamic planning cutting requirements of the coal mining machine. In addition, the three-dimensional geological model generation process based on the geological prediction network does not require human intervention, which reduces labor costs.

[0063] Figure 2 A schematic flow chart of a method for generating a geological attribute point cloud provided in an embodiment of the present application.

[0064] like Figure 2 As shown, the geological attribute point cloud generation method includes:

[0065] Step 201 , voxelize geological drilling data, geological profile data, and four geological models of strata, faults, folds, and collapse columns to form corresponding voxel models.

[0066] In the present application, a hexahedron voxel decomposition algorithm can be used to voxelize geological drilling data, geological profile data, and geological models of four geological objects, namely, strata, faults, folds, and collapse columns, to form a voxel model.

[0067] Step 202, assigning corresponding geological attributes to each voxel in the voxel model according to the geological drilling data, geological profile data, and geological attribute information contained in the four geological models.

[0068] In the present application, the geological attribute may be one of lithology, porosity, ore pressure, etc. For example, the assigned geological attribute may be lithology.

[0069] Step 203: Generate a sparse geological attribute point cloud according to the center point of each voxel in the voxel model and the attribute value of the geological attribute.

[0070] In the present application, the sparse geological attribute point cloud can be a point cloud corresponding to geological drilling data and geological profile data, or it can be a fusion point cloud of the two point clouds.

[0071] In one embodiment of the present application, it is possible to use Figure 3 The geological prediction network is trained by the method shown. Figure 3 A flowchart of a geological prediction network training method provided in an embodiment of the present application.

[0072] like Figure 3 As shown, the geological prediction network training method includes:

[0073] Step 301, obtaining a training data set corresponding to the sample area, wherein the training data set includes geological attribute point clouds corresponding to four geological models of strata, faults, folds and collapse columns in the sample area.

[0074] In this application, the initial geological prediction network can be implemented according to the principles of the classic point cloud completion algorithm, and the training point cloud generation method in the generator is modified to extract virtual drill holes, and the prediction effect evaluation criteria in the discriminator are modified to comprehensively consider the chamfer distance and geological laws.

[0075] In this application, classic point cloud completion algorithms include but are not limited to Point Completion Network (PCN), TopNet, Gridding Residual Network (GRNet), and Point Fractal Network (PF-Net). The generator refers to the algorithm for generating multi-size point clouds in the point cloud completion algorithm. This application is modified to generate virtual boreholes of varying numbers multiple times. The discriminator refers to the algorithm for evaluating the prediction effect based on the chamfer distance and loss function in the point cloud completion algorithm. This application is modified to comprehensively consider the chamfer distance and geological laws.

[0076] In this application, the training data set may include one or more of the sample geological attribute point clouds corresponding to the four geological models of strata, faults, folds and collapse columns. These four geological models should be verified on the spot to ensure that the model is consistent with the site to the greatest extent. Considering the complexity of the geological structure and stratum distribution law, the sample area and the area to be modeled belong to the same mining area.

[0077] Step 302: Using the sample geological attribute point cloud, an initial geological prediction network is trained by back propagation until the similarity between the geological attribute point cloud predicted and generated by the trained geological prediction network and the sample geological attribute point cloud is greater than a first threshold.

[0078] In this application, the generator of the initial geological prediction network can extract a virtual borehole point cloud from the sample geological attribute point cloud and input it into the network for feature extraction to obtain a global feature vector of the point cloud, and then use the feature vector to generate a predicted geological attribute point cloud. The discriminator of the initial geological prediction network can train the initial geological prediction network based on the difference between the predicted geological attribute point cloud and the sample geological attribute point cloud.

[0079] In this application, the chamfer distance between the predicted geological attribute point cloud and the formation geological attribute point cloud can be determined and calculated using the following formula:

[0080]

[0081] Among them, S1 represents the predicted geological attribute point cloud, S2 represents the sample geological attribute point cloud, CD(S1,S2) represents the chamfer distance between the above two point clouds, the first term on the right side of the equation represents the sum of the minimum distances from any point x in S1 to S2, and the second term represents the sum of the minimum distances from any point y in S2 to S1. If the chamfer distance is large, it means that the predicted geological attribute point cloud is far from the expectation; if the distance is small, it means that the predicted geological attribute point cloud is more in line with expectations.

[0082] When the chamfer distance is greater than the threshold or the predicted geological attribute point cloud does not conform to the geological law, the parameters of the initial geological prediction network are adjusted through back propagation training until the obtained chamfer distance is less than or equal to the second threshold and the predicted geological attribute point cloud conforms to the geological law, thereby completing the geological prediction network training and generating a geological prediction network, wherein the similarity between the geological attribute point cloud predicted and generated by the geological prediction network and the sample geological attribute point cloud is greater than the first threshold. Thus, the geological prediction network can be trained based on the comprehensive evaluation results of the chamfer distance and the geological law.

[0083] In this application, based on the geological prediction network for three-dimensional geological modeling of coal mines, the three-dimensional geological models of strata, faults, folds and collapse columns in the same coal mining area can be used as training samples and real models of the geological prediction network, and the coal mine geological drilling data and / or geological profile data can be used as the input of the geological prediction network, that is, the sparse point cloud as input, thereby converting the problem of geological condition inference in the three-dimensional geological modeling of coal mines into a problem of sparse point cloud completion to form a complete point cloud problem that approximates the real model, and then deep learning can be used to solve the problem.

[0084] In an embodiment of the present application, the initial geological prediction model is trained by using a training data set corresponding to a sample area belonging to the same mining area as the area to be modeled to obtain a geological prediction network. The geological prediction network can be used to generate a three-dimensional geological model in real time, thereby improving the modeling efficiency.

[0085] To further illustrate the above embodiment, the following takes the geological attribute as lithology, and the sample geological model includes a three-dimensional geological model of strata, faults, folds, collapse columns, etc. as an example, and combines Figure 4 Provide explanation. Figure 4 A schematic diagram of the process of modeling a three-dimensional geological model of a coal mine based on a geological prediction network is provided in an embodiment of the present application.

[0086] like Figure 4 As shown in the figure, the modeling process can be divided into three parts: coal mine geological information expression based on lithology point cloud, generation of geological prediction network and geological modeling application of geological prediction network.

[0087] For coal mine geological information expression based on lithology point cloud, the three-dimensional geological model of the strata, faults, folds, collapse columns, etc. of the sample area and the geological drilling and / or geological profile data of the area can be used as training samples.

[0088] When generating a geological prediction network, the initial geological prediction network can be trained based on the lithology point cloud obtained from the training samples to generate a geological prediction model. The detailed process can be referred to the above embodiment. Among them, the sample geological model for training and cross-validation can include three-dimensional geological models such as strata, faults, folds and collapse columns, and should be verified by field geology. In addition, considering the complexity of geological structure and stratum distribution law, the training samples should belong to the same mining area as the target area of ​​3D geological modeling.

[0089] When using the geological prediction network for geological modeling, the geological drilling data and / or geological profile data of the area to be modeled can be used to obtain a sparse geological attribute point cloud through a geological attribute point cloud generation method, and the coal mine geological drilling data and / or geological profile data and three-dimensional geological models such as strata, faults, folds and collapse columns are voxelized using a hexahedron voxel decomposition algorithm to form a voxel model, and the voxels at the spatial positions of the geological drilling data and / or geological profile data and three-dimensional geological models such as strata, faults, folds and collapse columns are assigned corresponding lithologies, and the center points and attribute values ​​of these voxels are extracted to form a sparse lithology point cloud. After that, the sparse lithology point cloud of the modeling data source is input into the trained geological prediction network to obtain a dense geological attribute point cloud, and the dense geological attribute point cloud is extracted by isosurface extraction. Through geometric topological processing, a three-dimensional geological model of the area to be modeled is generated.

[0090] The geological prediction network is obtained by training the geological model, and deployed online after the training is completed. It can support the real-time input and real-time modeling of the modeling data source, realize the online dynamic construction of the geological model, and support the dynamic planning and cutting of the coal mining machine. Therefore, according to the geological profile data and / or geological drilling data of the area to be modeled, the geological prediction network is used to generate a three-dimensional geological model in real time.

[0091] In this application, based on the principle of point cloud completion algorithm, three-dimensional geological models such as coal mine strata, faults, folds and collapse columns can be voxelized into point clouds with lithology, and geological law constraints can be integrated into the point cloud feature encoding and point cloud generation result evaluation process to form a lithology point cloud "encoder-decoder" mechanism to obtain a geological prediction network. In addition, the same coal mining area three-dimensional geological model samples are used for training. After the training is completed, the geological prediction network can be deployed online. Based on the sparse lithology point clouds formed by voxelization of geological drilling data, geological profile data, etc. in the coal mining area, an implicit geological model expressed by dense point clouds in the mining space is generated in real time to drive the coal mining machine to cut.

[0092] In order to implement the above-mentioned embodiment, the embodiment of the present application also proposes a three-dimensional geological model generation device based on a geological prediction network. Figure 5 A schematic structural diagram of a three-dimensional geological model generating device based on a geological prediction network provided in an embodiment of the present application.

[0093] like Figure 5 As shown, the three-dimensional geological model generation device 500 based on the geological prediction network includes:

[0094] The first acquisition module 510 is used to acquire geological drilling data and / or geological profile data of the area to be modeled, and obtain a sparse geological attribute point cloud by using a geological attribute point cloud generation method;

[0095] A second acquisition module 520 is used to input the sparse geological attribute point cloud into a geological prediction network for processing, and the geological prediction network outputs a dense geological attribute point cloud;

[0096] The generation module 530 is used to extract isosurfaces from the dense geological attribute point cloud and generate a three-dimensional geological model of the area to be modeled through geometric topology processing.

[0097] In a possible implementation of the embodiment of the present application, the first acquisition module 510 is used to:

[0098] voxelize the geological drilling data, the geological profile data, and four geological models of strata, faults, folds, and collapse columns to form corresponding voxel models;

[0099] Assigning corresponding geological attributes to each voxel in the voxel model according to the geological drilling data, the geological profile data, and the geological attribute information contained in the four geological models;

[0100] The sparse geological attribute point cloud is generated according to the center point of each voxel in the voxel model and the attribute value of the geological attribute.

[0101] In a possible implementation of the embodiment of the present application, the device may further include:

[0102] The third acquisition module is used to acquire a training data set corresponding to the sample area, wherein the training data set includes sample geological attribute point clouds corresponding to four geological models of strata, faults, folds and collapse columns in the sample area respectively;

[0103] The training module is used to use the sample geological attribute point cloud to train the initial geological prediction network through back propagation until the similarity between the geological attribute point cloud predicted and generated by the trained geological prediction network and the sample geological attribute point cloud is greater than a first threshold.

[0104] In a possible implementation of the embodiment of the present application, the training module is used to:

[0105] Extracting a virtual drilling point cloud from the sample geological attribute point cloud, and inputting it into the initial geological prediction network for feature extraction to obtain a point cloud global feature vector;

[0106] Generate a predicted geological attribute point cloud using the point cloud global feature vector;

[0107] Determining a chamfer distance between the predicted geological attribute point cloud and the sample stratum geological attribute point cloud;

[0108] When the chamfer distance is greater than a threshold or the predicted geological attribute point cloud does not conform to geological laws, the parameters of the initial geological prediction network are adjusted through the back propagation training until the obtained chamfer distance is less than or equal to a second threshold and the predicted geological attribute point cloud conforms to geological laws, thereby generating the geological prediction network.

[0109] In a possible implementation of the embodiment of the present application, the geological attribute is one of lithology, porosity, and mine pressure.

[0110] It should be noted that the above explanation of the embodiment of the method for generating a three-dimensional geological model based on a geological prediction network is also applicable to the device for generating a three-dimensional geological model of a coal mine in this embodiment, so it will not be repeated here.

[0111] In the embodiment of the present application, by obtaining geological drilling data and / or geological profile data of the area to be modeled, and through the geological attribute point cloud generation method, a sparse geological attribute point cloud is obtained; the sparse geological attribute point cloud is input into the geological prediction network for processing, and the geological prediction network outputs a dense geological attribute point cloud; the dense geological attribute point cloud is extracted from the isosurface, and a three-dimensional geological model of the area to be modeled is generated through geometric topology processing. Thus, by inputting the sparse geological attribute point cloud of the area to be modeled into the geological prediction network, a dense geological attribute point cloud can be obtained, and a three-dimensional geological model of the area to be modeled is generated according to the dense geological attribute point cloud, so that the geological prediction network can be used to generate a three-dimensional geological model in real time, and the modeling efficiency is high, which can match the mechanized coal mining cutting speed and meet the dynamic planning cutting requirements of the coal mining machine. In addition, the three-dimensional geological model generation process based on the geological prediction network does not require human intervention, which reduces labor costs.

[0112] In order to implement the above embodiment, the embodiment of the present application also proposes a computer device, including a processor and a memory;

[0113] The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to implement the three-dimensional geological model generation method based on the geological prediction network as described in the above embodiment.

[0114] In order to implement the above embodiment, the embodiment of the present application also proposes a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the three-dimensional geological model generation method based on the geological prediction network as described in the above embodiment is implemented.

[0115] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A method for generating a three-dimensional geological model based on a geological prediction network, characterized in that: include: Acquire geological drilling data and / or geological profile data of the area to be modeled, and obtain a sparse geological attribute point cloud by using a geological attribute point cloud generation method; Inputting the sparse geological attribute point cloud into a geological prediction network for processing, the geological prediction network outputting a dense geological attribute point cloud; Extracting isosurfaces from the dense geological attribute point cloud and generating a three-dimensional geological model of the area to be modeled through geometric topology processing; The sparse geological attribute point cloud is generated by the following steps: voxelize the geological drilling data, the geological profile data, and four geological models of strata, faults, folds, and collapse columns to form corresponding voxel models; Assigning corresponding geological attributes to each voxel in the voxel model according to the geological drilling data, the geological profile data, and the geological attribute information contained in the four geological models; The sparse geological attribute point cloud is generated according to the center point of each voxel in the voxel model and the attribute value of the geological attribute.

2. The method according to claim 1, characterized in that The geological prediction network is implemented by the following steps: Acquire a training data set corresponding to the sample area, wherein the training data set includes sample geological attribute point clouds corresponding to four geological models of strata, faults, folds and collapse columns in the sample area respectively; The sample geological attribute point cloud is used to train an initial geological prediction network through back propagation until the similarity between the geological attribute point cloud predicted and generated by the trained geological prediction network and the sample geological attribute point cloud is greater than a first threshold.

3. The method according to claim 2, characterized in that The back propagation training of the initial geological prediction network is achieved by the following steps: Extracting a virtual drilling point cloud from the sample geological attribute point cloud, inputting it into the initial geological prediction network for feature extraction, and obtaining a point cloud global feature vector; Generate a predicted geological attribute point cloud using the point cloud global feature vector; Determining a chamfer distance between the predicted geological attribute point cloud and the sample stratum geological attribute point cloud; When the chamfer distance is greater than the second threshold or the predicted geological attribute point cloud does not conform to geological laws, the parameters of the initial geological prediction network are adjusted through the back propagation training until the obtained chamfer distance is less than or equal to the second threshold and the predicted geological attribute point cloud conforms to geological laws, thereby generating the geological prediction network.

4. The method according to claim 1, characterized in that The geological attribute is one of lithology, porosity, and ore pressure.

5. A three-dimensional geological model generation device based on a geological prediction network, characterized in that: include: The first acquisition module is used to acquire geological drilling data and / or geological profile data of the area to be modeled, and obtain a sparse geological attribute point cloud by a geological attribute point cloud generation method; A second acquisition module is used to input the sparse geological attribute point cloud into a geological prediction network for processing, and the geological prediction network outputs a dense geological attribute point cloud; A generation module, used for performing isosurface extraction on the dense geological attribute point cloud and generating a three-dimensional geological model of the area to be modeled through geometric topological processing; The first acquisition module is further used to voxelize the geological drilling data, the geological profile data, and four geological models of strata, faults, folds, and collapse columns to form corresponding voxel models; Assigning corresponding geological attributes to each voxel in the voxel model according to the geological drilling data, the geological profile data, and the geological attribute information contained in the four geological models; The sparse geological attribute point cloud is generated according to the center point of each voxel in the voxel model and the attribute value of the geological attribute.

6. The device according to claim 5, characterized in that Also includes: The third acquisition module is used to acquire a training data set corresponding to the sample area, wherein the training data set includes sample geological attribute point clouds corresponding to four geological models of strata, faults, folds and collapse columns in the sample area respectively; The training module is used to use the sample geological attribute point cloud to train the initial geological prediction network through back propagation until the similarity between the geological attribute point cloud predicted and generated by the trained geological prediction network and the sample geological attribute point cloud is greater than a first threshold.

7. A computer device, characterized in that: including a processor and a memory; The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to implement the method according to any one of claims 1 to 4.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 4 is implemented.

Images