Mine geological drilling three-dimensional visualization method based on deep learning

Abstract

The invention relates to the technical field of surveying and mapping, and discloses a mine geological drilling three-dimensional visualization method based on deep learning, which comprises the following steps: S1, collecting drilling data, and extracting drilling sampling site data from the drilling data by adopting a three-point positioning method; s2, preprocessing the collected data to form a data set for model training; s3, constructing a three-dimensional deep learning model; s4, training the three-dimensional deep learning model by using the data set to realize optimization operation on the three-dimensional deep learning model; s5, performing data drilling by using the optimized three-dimensional deep learning model to form and generate a three-dimensional geologic model; and S6, inputting the three-dimensional geologic model after drilling into visual equipment, and carrying out display operation on the three-dimensional geologic model. According to the invention, efficient processing and three-dimensional visual display of mine geological drilling data are realized, the efficiency of mine exploration and development is improved, the safety risk is reduced, and powerful support is provided for decision making of mining companies.

Description

Technical Field

[0001] The present invention relates to the field of surveying and mapping technology, and in particular to a three-dimensional visualization method for mining geological drilling based on deep learning. Background Art

[0002] Following the digital earth and smart cities, the introduction of digital twins has further promoted the digitization, informatization and intelligence of various industries. In the geological industry, after the digital twins, industry or national research plans such as digital twin mines and petroleum geological twins have been proposed. Three-dimensional geological modeling technology is the basis and key technology of digital twin mines and petroleum digital twins, and is also an important part of digital earth and smart cities. Compared with traditional two-dimensional geological data, three-dimensional geological models can more completely express various geological phenomena, so as to better make decisions and geological analysis. Nowadays, three-dimensional geological modeling not only plays a major role in engineering exploration and mineral mining, but also develops rapidly in more and more aspects such as urban planning and construction, water resources management and geological disaster analysis. It is the key technical support for promoting the three-dimensional reconstruction of above-ground and underground entities and the development of urban digital twin technology.

[0003] Three-dimensional geological modeling technology has been fully developed, and has formed a number of mature modeling systems and a series of mature modeling methods. For example, explicit or implicit modeling methods supported by various interpolation techniques are widely used in engineering geology, and stochastic modeling methods supported by geostatistics are widely used in hydrogeological modeling and reservoir modeling. However, to date, based on limited geological survey data, accurately and intelligently establishing a three-dimensional visual fine geological model is still a considerable challenge in the field of geological modeling. With the rapid development of deep learning technology in recent years and its successful application in various fields, deep learning has injected fresh blood into the traditional field of geology and introduced new methods to geological modeling.

[0004] Existing deep learning-based 3D geological modeling algorithms do not consider the global correlation of geological models and cannot model the global relationship of 3D geological models. Existing deep learning-based conditional 3D geological modeling algorithms mostly use pre-trained unconditional generative adversarial network models combined with conditional constraint networks for modeling. The algorithms do not consider the relationship between all unknown points. The values ​​of unknown points are independent of each other. The correlation between simulated pixels and other global pixels cannot be considered. The dependence between pixels within the model cannot be modeled, so it is difficult to control the relationship between the overall and local aspects of the model, resulting in a lack of simulation capabilities for geological models.

[0005] A three-dimensional geological modeling method based on deep learning is recorded in publication number CN115619950 A, comprising: obtaining a three-dimensional geological model sample set and preprocessing the three-dimensional geological model sample set; performing mask processing on the preprocessed three-dimensional geological model sample set to extract drilling data and then construct a training data set; based on the training data set, training the three-dimensional geological modeling model to be trained to obtain a trained three-dimensional geological modeling model, wherein the three-dimensional geological modeling model at least includes a coding layer, a linear mapping layer, a Transformer decoding layer and an inverse mapping layer connected in sequence, wherein the coding layer is used to perform stratigraphic coding and position coding on the training data; obtaining the collected drilling data, and inputting the drilling data into the three-dimensional geological modeling model to generate a three-dimensional geological model. The present invention solves the current problem that the global relationship of the three-dimensional geological model cannot be modeled. The application uses the three-dimensional geological model to realize the three-dimensional geological construction operation. However, in the process of model data processing, the collected drilling data content is relatively messy, there is no correlation between the boreholes, the data set has a lot of data, the calculation efficiency is low, and the interference effect on the calculation results is relatively large. Therefore, a three-dimensional visualization method of mining geological boreholes based on deep learning is proposed. Summary of the invention

[0006] In order to solve the technical problems existing in the prior art, the present invention provides a three-dimensional visualization method of mine geological drilling based on deep learning.

[0007] The present invention is implemented by the following technical solution: a three-dimensional visualization method of mining geological drilling based on deep learning, comprising the following steps:

[0008] S1. Collecting drilling data, and extracting drilling sampling site data from the drilling data using a three-point positioning method;

[0009] S2. Preprocess the collected data to form a data set for model training;

[0010] S3, build a three-dimensional deep learning model;

[0011] S4. Using the data set to train the three-dimensional deep learning model to achieve optimization operations on the three-dimensional deep learning model;

[0012] S5. Using the optimized three-dimensional deep learning model to perform data drill to generate a three-dimensional geological model;

[0013] S6. Input the three-dimensional geological model after the drill into the visualization device to perform a display operation of the three-dimensional geological model.

[0014] As a further improvement of the above solution, the drilling data in step S1 includes the drilling number, the three-dimensional data of the rock formation of the drilling, and the rock formation number.

[0015] As a further improvement of the above scheme, the three-point positioning method in step S1 includes taking three adjacent drilling points that are not distributed in a straight line as a group of extraction, obtaining the points at the intersection of the top and bottom of the rock layer and the outer periphery of the borehole at the cross-section positions of two adjacent borehole axes as sampling site data, obtaining 8 sampling sites on each rock layer, and integrating the sampling site data and the borehole data to form a data set.

[0016] As a further improvement of the above solution, the data preprocessing in step S2 includes the following steps:

[0017] A1. Obtain the data set, create a rock formation coding comparison table based on the rock formation information, and implement the label coding operation for each rock formation;

[0018] A2. According to the rock layer coding comparison table, the attribute values ​​of the sampling sites determined by the three-point positioning method are converted;

[0019] A3. Convert the format of the original drilling data to obtain a sample set in a regular grid format.

[0020] As a further improvement of the above solution, the three-dimensional deep learning model in step S3 includes an input layer, a convolutional layer, a connection layer and an output layer connected in sequence;

[0021] The convolution layer includes convolution units connected in sequence, and the number of convolution units is at least 2, and the convolution unit is used to extract three-dimensional features of the data set to be tested;

[0022] The connection layer includes connection units connected in sequence, and the number of the connection units is at least 2.

[0023] As a further improvement of the above scheme, in step S4, the sample set is divided into a training set and a test set according to a set ratio, wherein the training set is used for training and adjusting the model parameters, and the test set is used for testing the accuracy of the model. During this period, the parameters of the model are continuously adjusted, and finally a model and parameters with the best accuracy are selected as the initialization model.

[0024] As a further improvement of the above solution, in step S5, the collected drilling data is input into the trained three-dimensional deep learning model to perform calculations and generate a three-dimensional geological model.

[0025] Compared with the prior art, the present invention has the following beneficial effects:

[0026] 1. The present invention uses a three-point positioning method to extract borehole sampling site data from the borehole data during the borehole sampling process. When constructing a three-dimensional model, the borehole data is collected according to the sampling points determined by the three-point positioning method, and then the three-dimensional model is constructed according to the collected sampling point data. On the one hand, relying on the three-point positioning, the individual borehole data is formed into a grid-like sampling point from a point to a line mode, so that the sampled borehole data are connected to facilitate the subsequent construction and verification of the three-dimensional model. On the other hand, the amount of data collection in the process of building the three-dimensional model is reduced, the calculation intensity in the process of building the three-dimensional model is reduced, and the efficiency of building the three-dimensional model is improved;

[0027] 2. The present invention realizes efficient processing and three-dimensional visualization of mine geological drilling data, improves the efficiency of mine exploration and development, reduces safety risks, and provides strong support for the decision-making of mining companies. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 A schematic diagram of the structure of the three-dimensional deep learning model in Example 4 provided by the present invention;

[0029] Figure 2 This is a simulation schematic diagram of the three-point positioning method in Example 2 provided by the present invention. DETAILED DESCRIPTION

[0030] The present invention is further described below in conjunction with the accompanying drawings and specific implementation methods. It should be noted that, under the premise of no conflict, the various embodiments or technical features described below can be arbitrarily combined to form a new embodiment.

[0031] Embodiment 1:

[0032] Please combine Figure 1-Figure 2 , a deep learning-based three-dimensional visualization method for mining geological drilling in this embodiment includes the following steps:

[0033] S1. Collecting drilling data, and extracting drilling sampling site data from the drilling data using a three-point positioning method;

[0034] S2. Preprocess the collected data to form a data set for model training;

[0035] S3, build a three-dimensional deep learning model;

[0036] S4. Using the data set to train the three-dimensional deep learning model to achieve optimization operations on the three-dimensional deep learning model;

[0037] S5. Using the optimized three-dimensional deep learning model to perform data drill to generate a three-dimensional geological model;

[0038] S6. Input the three-dimensional geological model after the drill into the visualization device to perform a three-dimensional display operation of the three-dimensional geological model.

[0039] Embodiment 2:

[0040] The drilling data includes the drilling number, the three-dimensional data of the rock formation of the drilling, and the rock formation number.

[0041] The three-point positioning method includes taking three adjacent borehole points that are not distributed in a straight line as a set of extraction groups, obtaining the points where the top and bottom of the rock layer and the outer periphery of the borehole at the cross-section positions of the two adjacent borehole axes intersect as sampling site data, obtaining 8 sampling sites on each rock layer, and integrating the sampling site data and the borehole data to form a data set;

[0042] When extracting sampling sites, three boreholes A, B and C that are not on the same celestial line are selected. At this time, the axes of boreholes A and B form section 2, the axes of boreholes A and C form section 1, and the axes of boreholes C and B form section 3. Each section forms four sampling points at the intersection of the top and bottom of the rock formation borehole and the outer periphery of the borehole, so eight sampling points are formed in the same borehole.

[0043] Embodiment 3:

[0044] The preprocessing of the collected data includes the following steps:

[0045] A1. Obtain the data set, create a rock formation coding comparison table based on the rock formation information, and implement the label coding operation for each rock formation;

[0046] A2. According to the rock layer coding comparison table, the attribute values ​​of the sampling sites determined by the three-point positioning method are converted;

[0047] A3. Convert the format of the original drilling data to obtain a sample set in a regular grid format.

[0048] Embodiment 4:

[0049] The three-dimensional deep learning model includes an input layer, a convolutional layer, a connection layer, and an output layer connected in sequence;

[0050] The convolution layer includes convolution unit 1, convolution unit 2 and convolution unit 3 connected in sequence, with a convolution kernel size of 3 and a feature dimension of 128; the convolution unit is used to extract three-dimensional features of the test data set;

[0051] The connection layer includes a fully connected layer 1, a fully connected layer 2, a fully connected layer 3, a fully connected layer 4 and a fully connected layer 5, which are connected in sequence and are used to extract features of eigenvalues ​​with a length of 536 connected in the connection layer. The number of neurons in each layer is 1024, 512, 128, 64 and 24 respectively.

[0052] The sample set is divided into a training set and a test set in a ratio of 8:2. The training set is used for model training and parameter adjustment, and the test set is used for model accuracy testing. During this period, the model parameters are continuously adjusted, and the optimizer (adam or sgd) and activation function (Relu or Liner) are used. Finally, a model with the best accuracy and parameters are selected as the initialization model.

[0053] The Relu function and Liner function are:

[0054]

[0055] Liner(x)=αx

[0056] Where Relu(x) represents the Relu loss function, Liner(x) represents the Liner loss function, x represents the output value of each layer, and α represents the slope of the linear function;

[0057] Based on the initialization model, the sample data set is continuously updated in subsequent iterative training to obtain the latest sample data set. The sample data set is also divided into a training set and a test set. The training set is used for model training, and the test set is used for model testing. The model is trained and predicted, and the accuracy of the iterative model is compared with the accuracy of the initialization model obtained in the previous step. If the accuracy is improved, the initialization model is replaced. If not, it is not replaced.

[0058] Among them, the RMSE and MAE formulas are used for accuracy evaluation, where:

[0059]

[0060] m is the number of test samples, Represents the actual value of the test sample, Represents the predicted value of the test sample.

[0061] Embodiment 5:

[0062] Real-time prediction and application:

[0063] Select measured drilling points in the mine, use drilling equipment to perform drilling operations, label the holes and sample the measured data of the holes;

[0064] Preprocess the sampled measured data to form a sample data set, and then input the sample data set into the trained three-dimensional deep learning model to perform model calculation to obtain a three-dimensional geological model;

[0065] The calculated three-dimensional geological model is input into the visualization device to perform three-dimensional display operation of the three-dimensional geological model.

[0066] The present invention uses a three-point positioning method to extract borehole sampling site data from borehole data during the borehole sampling process. When constructing a three-dimensional model, the borehole data is collected according to the sampling points determined by the three-point positioning method, and then the three-dimensional model is constructed according to the collected sampling point data. On the one hand, relying on the three-point positioning, the individual borehole data is formed into a grid-like sampling point in a point-to-line manner, so that the sampled borehole data are connected to facilitate the subsequent construction and verification of the three-dimensional model. On the other hand, the amount of data collection in the process of building the three-dimensional model is reduced, the calculation intensity in the process of building the three-dimensional model is reduced, and the efficiency of building the three-dimensional model is improved.

[0067] Realize efficient processing and 3D visualization of mine geological drilling data, improve the efficiency of mine exploration and development, reduce safety risks, and provide strong support for mining companies' decision-making.

[0068] The above-mentioned embodiments are only preferred embodiments of the present invention and cannot be used to limit the scope of protection of the present invention. Any non-substantial changes and substitutions made by technicians in this field on the basis of the present invention shall fall within the scope of protection required by the present invention.

Claims

1. A deep learning-based three-dimensional visualization method for mining geological drilling, characterized in that: The following steps are involved: S1. Collecting drilling data, and extracting drilling sampling site data from the drilling data using a three-point positioning method; S2. Preprocess the collected data to form a data set for model training; S3, build a three-dimensional deep learning model; S4. Using the data set to train the three-dimensional deep learning model to achieve optimization operations on the three-dimensional deep learning model; S5. Using the optimized three-dimensional deep learning model to perform data drill to generate a three-dimensional geological model; S6. Input the three-dimensional geological model after the drill into the visualization device to perform a display operation of the three-dimensional geological model.

2. A method for three-dimensional visualization of mining geological drilling based on deep learning as claimed in claim 1, characterized in that: The drilling data in step S1 includes the drilling number, the three-dimensional data of the rock layer of the drilling, and the rock layer number.

3. The method for three-dimensional visualization of mining geological drilling based on deep learning according to claim 1, characterized in that: The three-point positioning method in step S1 includes taking three adjacent drilling points that are not distributed in a straight line as a group of extraction, obtaining the points at the intersection of the top and bottom of the rock layer and the outer periphery of the borehole at the cross-section positions of two adjacent borehole axes as sampling site data, obtaining 8 sampling sites on each rock layer, and integrating the sampling site data and the borehole data to form a data set.

4. The method for three-dimensional visualization of mining geological drilling based on deep learning according to claim 1, characterized in that: The data preprocessing in step S2 includes the following steps: A1. Obtain the data set, create a rock formation coding comparison table based on the rock formation information, and implement the label coding operation for each rock formation; A2. According to the rock layer coding comparison table, the attribute values ​​of the sampling sites determined by the three-point positioning method are converted; A3. Convert the format of the original drilling data to obtain a sample set in a regular grid format.

5. The method for three-dimensional visualization of mining geological drilling based on deep learning according to claim 1, characterized in that: The three-dimensional deep learning model in step S3 includes an input layer, a convolutional layer, a connection layer and an output layer connected in sequence; The convolution layer includes convolution units connected in sequence, and the number of convolution units is at least 2, and the convolution unit is used to extract three-dimensional features of the data set to be tested; The connection layer includes connection units connected in sequence, and the number of the connection units is at least 2.

6. The method for three-dimensional visualization of mining geological drilling based on deep learning according to claim 1, characterized in that: In step S4, the sample set is divided into a training set and a test set according to a set ratio, wherein the training set is used for model training and parameter adjustment, and the test set is used for model accuracy testing. During this period, the model parameters are continuously adjusted, and finally a model with the best accuracy and parameters are selected as the initialization model.

7. The method for three-dimensional visualization of mining geological drilling based on deep learning according to claim 1, characterized in that: In step S5, the collected drilling data is input into the trained three-dimensional deep learning model to perform calculations and generate a three-dimensional geological model.

Images