Three-dimensional multi-parameter fusion magmatic rock positioning method and system

Abstract

The invention discloses a magmatic rock positioning method and system based on three-dimensional multi-parameter fusion, and relates to the field of geological prospecting, and the method comprises the steps: obtaining the three-dimensional data of each geological parameter of a target region, and carrying out the preprocessing of the three-dimensional data to obtain three-dimensional multi-parameter data; identifying a probability prediction value of a type to which each three-dimensional grid belongs in the three-dimensional multi-parameter data by using the trained magmatic rock classification and identification model; and according to the probability prediction value of the belonging type of the three-dimensional grid and the probability prediction value of the belonging type of the adjacent three-dimensional grid, carrying out smoothing processing to determine the final types of all the three-dimensional grids, and obtaining the magmatic rock body type and three-dimensional positioning. According to the method, the rock mass types of the three-dimensional grids are determined based on the various parameters by aligning and fusing the various three-dimensional data of the target area, and the rock mass types of the positions can be accurately determined, so that the rock mass positioning accuracy of the magmatic rock is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of geological exploration, and more specifically, to a method and system for positioning magmatic rocks by three-dimensional multi-parameter fusion. Background Art

[0002] Magmatic rocks are rocks formed by the cooling and solidification of magma ejected onto the earth's surface or intruded into the earth's crust. Common magmatic rocks include granite, syenite, diorite, andesite, gabbro, and basalt, etc. The positioning of magmatic rocks is of great significance for geological structure research, mineral exploration, and geological age determination.

[0003] Existing magmatic rock positioning methods usually directly invert and locate the position of underground rock masses through electrical methods and electromagnetic methods, or invert the distribution of underground rock layers through seismic exploration technology, or depict the gravity field distribution through gravity measurement instruments to invert and determine the general position of magmatic rocks. However, there are problems of insufficient positioning accuracy and poor accuracy when positioning magmatic rock masses only through a single method.

[0004] Therefore, how to improve the positioning accuracy of magmatic rock masses is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0005] In view of this, the present invention provides a method and system for positioning magmatic rocks by three-dimensional multi-parameter fusion, which improves the positioning accuracy of magmatic rock masses by obtaining various three-dimensional parameters of the target area.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] The present invention discloses a method for positioning magmatic rocks by three-dimensional multi-parameter fusion, and the specific steps are as follows:

[0008] Obtain three-dimensional data of various geological parameters of the target area, and preprocess to obtain three-dimensional multi-parameter data;

[0009] Use the trained magmatic rock classification and recognition model to determine the probability prediction value of the type to which each three-dimensional grid in the three-dimensional multi-parameter data belongs;

[0010] According to the probability prediction value of the type to which the three-dimensional grid belongs and the probability prediction value of the type to which the adjacent three-dimensional grid belongs, perform smoothing processing to determine the final type of all three-dimensional grids, and obtain the magmatic rock mass type and three-dimensional positioning.

[0011] Further, the three-dimensional geological parameter data includes: three-dimensional density data, three-dimensional magnetic susceptibility data, three-dimensional resistivity data, three-dimensional polarizability data, three-dimensional elastic modulus data, three-dimensional mineral alteration data, and three-dimensional radioactive gamma energy spectrum measurement data, which are obtained through the following methods:

[0012] Divide the target area into multiple sub-areas;

[0013] Obtain the gravity field data of each of the sub-areas through a gravity measuring instrument, and invert the three-dimensional underground density data;

[0014] Measure the geomagnetic field of each of the sub-areas using a magnetometer, and invert the three-dimensional underground magnetic susceptibility data;

[0015] Adopt electrical methods and electromagnetic methods to measure and invert the three-dimensional resistivity data and three-dimensional polarizability data of each of the sub-areas;

[0016] Obtain the seismic wave data of each of the sub-areas through seismic exploration, and invert the three-dimensional underground elastic modulus data;

[0017] Obtain the satellite hyperspectral remote sensing data of each of the sub-areas, and invert the three-dimensional mineral alteration data;

[0018] Conduct radioactive exploration on each of the sub-areas, and invert the three-dimensional underground radioactive gamma energy spectrum measurement data.

[0019] Further, the dividing the target area into multiple sub-areas includes: measuring radon gas and mercury gas in the target area to determine the position and trend of deep fractures in the target area; dividing the target area into multiple sub-areas with independent structures according to the position and trend of the deep fractures.

[0020] Further, the preprocessing to obtain three-dimensional multi-parameter data includes:

[0021] Clean the three-dimensional data of geological parameters, remove incorrect data and interpolate and supplement missing data;

[0022] Set the resampling three-dimensional grid size, and resample the three-dimensional data of geological parameters after data cleaning;

[0023] Align and register each three-dimensional grid in the resampled three-dimensional density data, three-dimensional magnetic susceptibility data, three-dimensional resistivity data, three-dimensional polarizability data, three-dimensional elastic modulus data, three-dimensional mineral alteration data, and three-dimensional radioactive gamma energy spectrum measurement data, and fuse them to obtain the three-dimensional multi-parameter data.

[0024] Further, the magmatic rock classification and recognition model is obtained by stacking multiple restricted Boltzmann machines. The specific training method is:

[0025] Determine the multi-parameter data of the actual materials corresponding to the types through experiments, use the parameter values of each parameter as sample features, and use the material type as the sample label to organize and obtain a training sample set;

[0026] Pre-train each of the multiple restricted Boltzmann machines using the training sample set;

[0027] Stack the multiple pre-trained restricted Boltzmann machines to form an initial igneous rock classification and recognition model;

[0028] Fine-tune the igneous rock classification and recognition model using the backpropagation algorithm, update the network parameters of the initial igneous rock classification and recognition model, and obtain the final recognition model.

[0029] Further, use the contrastive divergence algorithm to perform the pre-training on each of the multiple restricted Boltzmann machines, specifically:

[0030] Input the training sample set into the restricted Boltzmann machine, and determine the states of the training sample set in the visible units of the restricted Boltzmann machine and the probabilities in the hidden units of the restricted Boltzmann machine;

[0031] Based on the states and probabilities, calculate the first joint activation probability and the second joint activation probability; the first joint activation probability is used to represent the joint activation probability of the visible units and the hidden units under the data distribution; the second joint activation probability is used to represent the joint activation probability of the reconstruction distribution;

[0032] Based on the first joint activation probability and the second joint activation probability, update the weights of the restricted Boltzmann machine from the visible units to the hidden units.

[0033] Further, when updating the weights from the visible units to the hidden units, the incremental formula for weight update is:

[0034] Δw qxy =∈ ra (<v x h y > data -<v x h y > recon );

[0035] where, Δw qxy is the weight update increment of the restricted Boltzmann machine, w qxy represents the weight from the x-th visible unit to the y-th hidden unit in the q-th layer; ∈ ra is the learning rate; <> data represents the joint activation probability symbol of the visible units and the hidden units under the data distribution; v x represents the state of the x-th visible unit, h y represents the probability of the y-th hidden unit given the visible state v x ; <> reconSymbol for the joint activation probability under the reconstructed distribution.

[0036] Furthermore, the backpropagation algorithm is a backpropagation algorithm based on a loss function, and the loss function is:

[0037]

[0038] where C k is the loss function of the magmatic rock classification and recognition model, N is the total number of samples, and K is the total number of categories; t nk represents the true label of the k-th target category of the n-th sample; y nk represents the predicted probability that the n-th sample belongs to the k-th category; λ bh is the regularization coefficient of the loss function, and w qxy represents the weight from the x-th visible unit to the y-th hidden unit in the q-th layer of the recognition model, is the square of the weight.

[0039] Furthermore, the smoothing process is specifically as follows:

[0040] Step 1: Sort the probability prediction values of the types to which each three-dimensional grid belongs;

[0041] Step 2: Select a three-dimensional grid to be determined, and judge whether the difference between the maximum value and the second maximum value after sorting is greater than a set threshold;

[0042] Step 3: If it is greater, take the type corresponding to the maximum value of the probability prediction value as the final type of the three-dimensional grid, otherwise go to Step 4;

[0043] Step 4: Calculate the probability prediction mean and variance of the types to which the neighboring three-dimensional grids of the three-dimensional grid to be determined belong, and determine the probability correction value of the three-dimensional grid to be determined;

[0044] Step 5: Sort all the probability correction values of the three-dimensional grid to be determined, and select the type corresponding to the maximum value as the final type of the three-dimensional grid;

[0045] Step 6: Judge whether the final types of all three-dimensional grids are determined. If so, end the smoothing process. If not, return to Step 2;

[0046] The formula for the probability correction value is:

[0047]

[0048] where, represents the probability correction value of the j-th type of the three-dimensional grid to be determined; P oj represents the probability prediction value of the j-th type of the three-dimensional grid to be determined; Represents the average probability prediction of the j-th type in the adjacent three-dimensional grid; e is the natural constant; Represents the variance of the probability prediction of the j-th type in the adjacent three-dimensional grid, and the formula is:

[0049]

[0050] Among them, P ij Represents the probability prediction value of the j-th type in the i-th adjacent three-dimensional grid, where i = 1, 2, 3... M, and M is the total number of adjacent three-dimensional grids.

[0051] The present invention also discloses a three-dimensional multi-parameter fusion magma rock positioning system, including:

[0052] Data acquisition module: Acquire the three-dimensional data of various geological parameters in the target area and preprocess it to obtain three-dimensional multi-parameter data;

[0053] Model recognition module: Use the trained magma rock classification and recognition model to determine the probability prediction value of the type to which each three-dimensional grid in the three-dimensional multi-parameter data belongs;

[0054] Rock mass positioning module: According to the probability prediction value of the type to which the three-dimensional grid belongs and the probability prediction value of the type to which the adjacent three-dimensional grid belongs, perform smoothing processing to determine the final type of all three-dimensional grids, and obtain the magma rock mass type and three-dimensional positioning.

[0055] Through the above technical solutions, compared with the prior art, the present invention discloses a three-dimensional multi-parameter fusion magma rock positioning method and system. By measuring fractures and dividing the target area, it can adapt to complex geological environments, improve its flexibility and effectiveness in practical applications; by fusing multiple geological parameters, it can more comprehensively reflect the geological characteristics of the target area, help eliminate the limitations of single methods, enhance the reliability of overall analysis, and thus significantly improve the positioning accuracy of magma rock masses; and using the trained magma rock classification and recognition model can not only locate magma rock masses but also realize the classification and recognition of different magma rocks. The present invention can accurately confirm the rock mass type at each position by aligning and fusing various three-dimensional data in the target area and determining the rock mass type of each three-dimensional grid based on multiple parameters, thereby improving the positioning accuracy of magma rock masses. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only the embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to the provided drawings.

[0057] Figure 1 This is the overall flow schematic diagram of the embodiment of the present invention. Specific embodiments

[0058] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0059] The embodiment of the present invention discloses a method for locating magmatic rocks by three-dimensional multi-parameter fusion, as Figure 1 shown, the specific steps are as follows:

[0060] Obtain the three-dimensional data of various geological parameters in the target area, and preprocess to obtain three-dimensional multi-parameter data;

[0061] Use the trained magmatic rock classification and recognition model to determine the probability prediction values of the types to which each three-dimensional grid in the three-dimensional multi-parameter data belongs;

[0062] According to the probability prediction values of the types to which the three-dimensional grids belong and the probability prediction values of the types to which the adjacent three-dimensional grids belong, perform smoothing processing to determine the final types of all three-dimensional grids, and obtain the magmatic rock mass type and three-dimensional positioning.

[0063] In a specific embodiment, the three-dimensional geological parameter data includes: three-dimensional density data, three-dimensional magnetic susceptibility data, three-dimensional resistivity data, three-dimensional polarizability data, three-dimensional elastic modulus data, three-dimensional mineral alteration data, and three-dimensional radioactive gamma energy spectrum measurement data, and is obtained through the following methods:

[0064] Divide the target area into multiple sub-areas;

[0065] Obtain the gravity field data of each sub-area through a gravity measurement instrument, and invert the underground three-dimensional density data;

[0066] Use a magnetometer to measure the geomagnetic field of each sub-area, and invert the underground three-dimensional magnetic susceptibility data;

[0067] Adopt electrical methods and electromagnetic methods to measure and invert the three-dimensional resistivity data and three-dimensional polarizability data of each sub-area;

[0068] Obtain the seismic wave data of each sub-area through seismic exploration, and invert the underground three-dimensional elastic modulus data;

[0069] Obtain the satellite hyperspectral remote sensing data of each sub-area, and invert the three-dimensional mineral alteration data;

[0070] Radioactive exploration is carried out on each sub-region, and the underground three-dimensional radioactive gamma energy spectrum measurement data is inverted. The above inversion methods based on the gravity field, geomagnetic field, electrical method, electromagnetic method, and seismic waves are common methods in the field of geological exploration and will not be elaborated here.

[0071] In a specific embodiment, the target area is divided into multiple sub-regions, including: measuring radon gas and mercury gas in the target area to determine the location and trend of deep fractures in the target area; dividing the target area into multiple sub-regions with independent structures according to the location and trend of the deep fractures. Radon gas and mercury gas often release to the surface along the deep fracture zone. By measuring the distribution and concentration of these gases, the location and trend of the deep fractures can be outlined. By dividing multiple sub-regions, the geological characteristics of each region can be analyzed in more detail, and specific geological anomalies or characteristics, especially those related to magmatic activities, can be better identified, thereby reducing the interference of fracture fissures on inversion and improving the accuracy of inverting three-dimensional density, magnetic susceptibility, resistivity, polarizability, and elastic modulus data.

[0072] In a specific embodiment, three-dimensional multi-parameter data is preprocessed, including:

[0073] Clean the three-dimensional data of geological parameters, remove the error data and interpolate and supplement the missing data;

[0074] Set the resampling three-dimensional grid size and resample the three-dimensional data of geological parameters after data cleaning;

[0075] Align and register each three-dimensional grid in the resampled three-dimensional density data, three-dimensional magnetic susceptibility data, three-dimensional resistivity data, three-dimensional polarizability data, three-dimensional elastic modulus data, three-dimensional mineral alteration data, and three-dimensional radioactive gamma energy spectrum measurement data, and fuse them to obtain three-dimensional multi-parameter data.

[0076] In a specific embodiment, the magmatic rock classification and recognition model is obtained by stacking multiple restricted Boltzmann machines. The specific training method is as follows:

[0077] Determine the multi-parameter data of the actual material corresponding to the type through experiments, use the parameter values of each parameter as sample features, and use the type to which the material belongs as the sample label to organize and obtain a training sample set;

[0078] Pre-train multiple restricted Boltzmann machines respectively using the training sample set;

[0079] Stack the pre-trained multiple restricted Boltzmann machines to form an initial magmatic rock classification and recognition model;

[0080] Use the backpropagation algorithm to fine-tune the magmatic rock classification and recognition model, update the network parameters of the initial magmatic rock classification and recognition model, and obtain the final recognition model.

[0081] In a specific embodiment, the types of the three-dimensional grids include: granite, syenite, diorite, andesite, gabbro, basalt, as well as air, water, and soil.

[0082] In a specific embodiment, the contrastive divergence algorithm is used to pre-train multiple restricted Boltzmann machines respectively, specifically:

[0083] Input the training sample set into the restricted Boltzmann machine, and determine the states of the training sample set in the visible units of the restricted Boltzmann machine and the probabilities in the hidden units of the restricted Boltzmann machine;

[0084] Based on the states and probabilities, calculate the first joint activation probability and the second joint activation probability; the first joint activation probability is used to represent the joint activation probability of the visible units and the hidden units under the data distribution; the second joint activation probability is used to represent the joint activation probability of the reconstructed distribution;

[0085] Based on the first joint activation probability and the second joint activation probability, update the weights of the restricted Boltzmann machine from the visible units to the hidden units.

[0086] In a specific embodiment, when updating the weights from the visible units to the hidden units, the incremental formula for weight update is:

[0087] Δw qxy =∈ ra (<v x h y > data -<v x h y > recon );

[0088] Among them, Δw qxy is the weight update increment of the restricted Boltzmann machine, w qxy represents the weight from the x-th visible unit to the y-th hidden unit in the q-th layer; ∈ ra is the learning rate; <> data represents the joint activation probability symbol of the visible units and the hidden units under the data distribution; v x represents the state of the x-th visible unit, h y represents the probability of the y-th hidden unit given the visible state v x ; <> recon represents the joint activation probability symbol under the reconstructed distribution.

[0089] The formula for the joint activation probability of the visible units and the hidden units under the data distribution is:

[0090]

[0091] Where N is the total number of samples, v nx represents the state of the x-th visible unit in the n-th sample, and v nx is the state of the x-th visible unit; h ny (v nx ) represents the probability of the n-th sample for the y-th hidden unit given the visible state v nx . The formula is:

[0092]

[0093] Where Sig represents the sigmoid activation function, w xy represents the weight from the x-th visible unit to the y-th hidden unit, and b y is the bias of the y-th hidden unit.

[0094] The formula for the joint activation probability of the reconstruction distribution is:

[0095]

[0096] Where represents the state of the x-th visible unit in the n-th sample reconstructed through the hidden state; represents the probability of the n-th sample for the y-th hidden unit given the hidden state .

[0097] In a specific embodiment, the backpropagation algorithm is the backpropagation algorithm based on the loss function, and the loss function is:

[0098]

[0099] Where C k is the loss function of the magmatic rock classification and recognition model, N is the total number of samples, and K is the total number of categories; t nk represents the true label of the k-th target category of the n-th sample; y nk represents the predicted probability that the n-th sample belongs to the k-th category; λ bh is the regularization coefficient of the loss function, and w qxy represents the weight from the x-th visible unit to the y-th hidden unit in the q-th layer of the recognition model, is the square of the weight.

[0100] In a specific embodiment, the smoothing process is specifically:

[0101] Step 1: Sort the probability prediction values of the types to which each three-dimensional grid belongs;

[0102] Step 2: Select a three-dimensional grid to be determined, and judge whether the difference between the maximum value and the second maximum value after sorting is greater than the set threshold;

[0103] Step 3: If it is greater, use the type corresponding to the maximum value of the probability prediction value as the final type of the 3D grid; otherwise, go to Step 4;

[0104] Step 4: Calculate the probability prediction mean and variance of the types to which the neighboring 3D grids of the 3D grid to be determined belong, and determine the probability correction value of the 3D grid to be determined;

[0105] Step 5: Sort all the probability correction values of the 3D grid to be determined, and select the type corresponding to the maximum value as the final type of the 3D grid;

[0106] Step 6: Determine whether the final types of all 3D grids are determined. If so, end the smoothing process; if not, return to Step 2;

[0107] The formula for the probability correction value is:

[0108]

[0109] where represents the probability correction value of the j-th type of the 3D grid to be determined; P oj represents the probability prediction value of the j-th type of the 3D grid to be determined; represents the average probability prediction of the j-th type of the neighboring 3D grids; e is the natural constant; represents the probability prediction variance of the j-th type of the neighboring 3D grids, and the formula is:

[0110]

[0111] where P ij represents the probability prediction value of the j-th type of the i-th neighboring 3D grid, i = 1, 2, 3... M, and M is the total number of neighboring 3D grids.

[0112] By performing threshold judgment and correction on the prediction probabilities of each type output by the magmatic rock classification and recognition model, it is possible to perform probability correction and smoothing processing on each grid according to the neighboring 3D grids, further improving the accuracy of the recognition and positioning of various magmatic rocks.

[0113] The embodiment of the present invention also discloses a 3D multi-parameter fusion magmatic rock positioning system, including:

[0114] Data acquisition module: Acquire the 3D data of each geological parameter in the target area and preprocess it to obtain 3D multi-parameter data;

[0115] Model recognition module: Use the trained magmatic rock classification and recognition model to determine the probability prediction values of the types to which each 3D grid in the 3D multi-parameter data belongs;

[0116] Rock mass positioning module: Based on the probability prediction values of the types to which the three-dimensional grids belong and the probability prediction values of the types to which the adjacent three-dimensional grids belong, perform smoothing processing to determine the final types of all three-dimensional grids, and obtain the types of magmatic rock masses and three-dimensional positioning.

[0117] In the present specification, the various embodiments are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the same or similar parts among the various embodiments, reference can be made to each other. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple. For the relevant parts, reference can be made to the description in the method part.

[0118] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A three-dimensional multi-parameter fusion igneous rock positioning method, characterized in that: The specific steps are as follows: Obtain three-dimensional data of geological parameters of the target area, and preprocess to obtain three-dimensional multi-parameter data; Using the trained igneous rock classification and recognition model, determine the probability prediction value of the type to which each three-dimensional grid in the three-dimensional multi-parameter data belongs; According to the probability prediction value of the type of the three-dimensional grid and the probability prediction value of the type of the adjacent three-dimensional grid, smoothing processing is performed to determine the final type of all three-dimensional grids, and the type and three-dimensional location of the igneous rock mass are obtained.

2. A three-dimensional multi-parameter fusion igneous rock positioning method according to claim 1, characterized in that: The three-dimensional geological parameter data include: three-dimensional density data, three-dimensional magnetic susceptibility data, three-dimensional resistivity data, three-dimensional polarizability data, three-dimensional elastic modulus data, three-dimensional mineral alteration data and three-dimensional radioactive gamma spectrum measurement data, which are obtained by the following methods: Divide the target area into multiple sub-areas; Acquiring gravity field data of each of the sub-areas through a gravity measuring instrument, and inverting underground three-dimensional density data; Using a magnetometer to measure the geomagnetic field of each of the sub-areas, and inverting underground three-dimensional magnetic susceptibility data; Using electrical and electromagnetic methods to measure and invert three-dimensional resistivity data and three-dimensional polarizability data of each of the sub-areas; Acquire the vibration wave data of each sub-area through seismic exploration, and invert the underground three-dimensional elastic modulus data; Acquire satellite hyperspectral remote sensing data of each of the sub-areas and invert three-dimensional mineral alteration data; Radioactive exploration is carried out on each of the sub-areas, and underground three-dimensional radioactive gamma spectrum measurement data is inverted.

3. A three-dimensional multi-parameter fusion igneous rock positioning method according to claim 2, characterized in that: The dividing the target area into multiple sub-areas includes: measuring radon gas and mercury gas in the target area to determine the location and direction of deep fractures in the target area; and dividing the target area into multiple sub-areas with independent structures according to the location and direction of the deep fractures.

4. A three-dimensional multi-parameter fusion igneous rock positioning method according to claim 2, characterized in that: The preprocessing obtains three-dimensional multi-parameter data, including: Cleaning the three-dimensional data of the geological parameters, removing erroneous data and interpolating and supplementing missing data; Set the resampling 3D grid size and resample the 3D data of geological parameters after data cleaning; The three-dimensional grids in the resampled three-dimensional density data, the three-dimensional magnetic susceptibility data, the three-dimensional resistivity data, the three-dimensional polarizability data, the three-dimensional elastic modulus data, the three-dimensional mineral alteration data and the three-dimensional radioactive gamma spectrum measurement data are aligned and registered, and fused to obtain the three-dimensional multi-parameter data.

5. The three-dimensional multi-parameter fusion igneous rock positioning method according to claim 1 is characterized in that: The igneous rock classification and recognition model is obtained by stacking multiple restricted Boltzmann machines. The specific training method is: Determine the multi-parameter data of the actual material corresponding to the type through experiments, take the parameter value of each parameter as the sample feature, take the type of the material as the sample label, and sort out to obtain a training sample set; Pre-training the plurality of restricted Boltzmann machines respectively using the training sample set; Superimposing the plurality of restricted Boltzmann machines after pre-training to form an initial igneous rock classification and recognition model; The igneous rock classification and recognition model is fine-tuned using a back-propagation algorithm, the network parameters of the initial igneous rock classification and recognition model are updated, and a final recognition model is obtained.

6. A three-dimensional multi-parameter fusion igneous rock positioning method according to claim 5, characterized in that: The pre-training is performed on the plurality of restricted Boltzmann machines respectively using the contrast divergence algorithm, specifically: Inputting the training sample set into the restricted Boltzmann machine, determining the state of the training sample set in the visible unit of the restricted Boltzmann machine and the probability of the hidden unit of the restricted Boltzmann machine; Based on the state and probability, a first joint activation probability and a second joint activation probability are calculated; the first joint activation probability is used to characterize the joint activation probability of visible units and hidden units under the data distribution; the second joint activation probability is used to characterize the joint activation probability of the reconstructed distribution; Based on the first joint activation probability and the second joint activation probability, weights from visible units to hidden units of the restricted Boltzmann machine are updated.

7. A three-dimensional multi-parameter fusion igneous rock positioning method according to claim 6, characterized in that: When the weights from the visible unit to the hidden unit are updated, the incremental formula for weight update is: Δw qxy =∈ ra (<v x h y > data -<v x h y > recon ); Among them, Δw qxy is the weight update increment of the restricted Boltzmann machine, w qxy represents the weight from the xth visible unit to the yth hidden unit in the qth layer; ∈ ra is the learning rate; data Represents the joint activation probability symbol of visible units and hidden units under data distribution; v x represents the state of the xth visible unit, h y Represents a given visible state v x The probability of the yth hidden unit when recon Represents the symbol of the joint activation probability under the reconstructed distribution.

8. The three-dimensional multi-parameter fusion igneous rock positioning method according to claim 5, characterized in that: The back propagation algorithm is a back propagation algorithm based on a loss function, and the loss function is: Among them, C k is the loss function of the igneous rock classification and recognition model, N is the total number of samples, K is the total number of categories; t nk represents the true label of the kth target category of the nth sample; y nk represents the predicted probability that the nth sample belongs to the kth category; λ bh is the regularization coefficient of the loss function, w qxy represents the weight from the xth visible unit to the yth hidden unit in the qth layer of the recognition model, is the square of the weight.

9. The three-dimensional multi-parameter fusion igneous rock positioning method according to claim 1 is characterized in that: The smoothing process is specifically as follows: Step 1: sorting the probability prediction values ​​of the types of the three-dimensional grids; Step 2: Select a three-dimensional grid to be determined, and determine whether the difference between the maximum value and the second largest value after sorting is greater than a set threshold; Step 3: If it is greater than, the type corresponding to the maximum value of the probability prediction value is used as the final type of the three-dimensional grid, otherwise, proceed to step 4; Step 4: Calculate the probability prediction mean and variance of each type of the adjacent three-dimensional grid of the three-dimensional grid to be determined, and determine the probability correction value of the three-dimensional grid to be determined; Step 5: sorting all probability correction values ​​of the three-dimensional grid to be determined, and selecting the type corresponding to the maximum value as the final type of the three-dimensional grid; Step 6: Determine whether the final type of all three-dimensional grids is determined, if yes, end the smoothing process, if no, return to step 2; The formula for the probability correction value is: in, represents the probability correction value of the j-th type of the three-dimensional grid to be determined; P oj represents the probability prediction value of the j-th type of the three-dimensional grid to be determined; represents the average value of the probability prediction of the type j in the adjacent three-dimensional grid; e is a natural constant; It represents the probability prediction variance of the j-th type of the adjacent three-dimensional grid, and the formula is: Among them, P ij Represents the probability prediction value of the j-th type of the i-th adjacent three-dimensional grid, i = 1, 2, 3...M, M is the total number of adjacent three-dimensional grids.

10. A three-dimensional multi-parameter fusion magmatic rock positioning system, characterized in that: include: Data acquisition module: obtains three-dimensional data of geological parameters of the target area, and preprocesses to obtain three-dimensional multi-parameter data; Model recognition module: using the trained igneous rock classification recognition model to determine the probability prediction value of the type to which each three-dimensional grid in the three-dimensional multi-parameter data belongs; Rock mass positioning module: According to the probability prediction value of the type of the three-dimensional grid and the probability prediction value of the type of the adjacent three-dimensional grid, smoothing processing is performed to determine the final type of all three-dimensional grids, and the igneous rock mass type and three-dimensional positioning are obtained.

Images