Method for delineating favorable uranium metallogenic area based on machine learning algorithm

Abstract

The invention belongs to the technical field of sandstone type uranium resource exploration and evaluation, and particularly relates to a method for delineating a uranium metallogenic favorable area based on a machine learning algorithm, and the method comprises the steps: 1, selecting uranium metallogenic indication elements; 2, preprocessing uranium metallogenic indication element value data; 3, establishing a machine learning model, and calculating a uranium mineralization probability value; and 4, delineating a uranium metallogenic favorable area according to the uranium metallogenic probability value. According to the method, the existing physical and chemical exploration information and drilling information can be fully utilized, accurate delineation of the uranium metallogenic favorable area is achieved, then accurate delineation of the uranium metallogenic target area is assisted, fine arrangement of uranium mine drilling holes is guided, the probability of occurrence of non-ore drilling holes is reduced, and the effect of efficiently and economically finding the uranium mine is achieved.

Description

Technical Field

[0001] The invention belongs to the technical field of sandstone-type uranium resource exploration and evaluation, and specifically relates to a method for delineating favorable uranium mineralization areas based on a machine learning algorithm. Background Art

[0002] At present, in the case of radioactive exploration data, geochemical exploration data, ground high-precision magnetic survey data and other basic exploration data and a small number of drill holes in the uranium exploration area, the scope of the favorable uranium mineralization area is simply determined based on the correlation between radioactive anomalies, geochemical anomalies, magnetic anomalies and the ore-bearing conditions in the drill holes. In this case, the in-depth connection between radioactive anomalies, geochemical anomalies, magnetic anomalies and the ore-bearing conditions in the drill holes is not fully explored, resulting in the inaccuracy of the scope of the favorable uranium mineralization area delineated, making it difficult to economically and efficiently lock the uranium mineralization target area.

[0003] Therefore, it is urgent to develop a method for delineating favorable uranium mineralization areas to solve the problem that existing technologies are difficult to economically and accurately delineate favorable uranium mineralization areas. Summary of the invention

[0004] The purpose of the present invention is to provide a method for delineating favorable uranium mineralization areas based on a machine learning algorithm. The method is based on radioactive exploration data, geochemical exploration data, ground high-precision magnetic survey data and known drilling information, and applies a machine learning algorithm to deeply explore the intrinsic relationship between radioactive information, geochemical information, magnetic information and uranium mineralization, calculate the uranium mineralization probability in the uranium exploration area, and then delineate favorable uranium mineralization areas, thereby providing technical support for accurately locating uranium mineralization target areas and saving drilling costs.

[0005] The technical solution to achieve the purpose of the present invention is:

[0006] A method for delineating favorable uranium mineralization areas based on a machine learning algorithm, the method comprising:

[0007] Step 1: Select uranium mineralization indicator elements;

[0008] Step 2: Preprocessing of uranium mineralization indicator value data;

[0009] Step 3: Establish a machine learning model to calculate the probability of uranium mineralization;

[0010] Step 4: Based on the uranium mineralization probability value, delineate the favorable uranium mineralization area.

[0011] The step 1 specifically comprises: based on the existing radioactive exploration data, geochemical exploration data, high-precision magnetic survey data and drilling ore-seeking conditions in the uranium exploration area, and according to the mineralization type of the uranium ore in the uranium exploration area, determining the radioactive indicator elements, soil geochemical indicator elements, soil metal activity indicator elements, magnetic indicator elements, and soil gas content indicator elements that can indicate the uranium mineralization conditions.

[0012] The radioactive exploration elements include: CR U , CR Th , CR K , C Rn 、TLD、C Po-210 ;Soil geochemical elements include: CG U , CG Th , CG K , CG Mo , CG V , CG Pb , CG Zn , CG Cu ; The active elements of soil metals include: CP U , CP Th , CP K , CP Mo , CP V , CP Pb , CP Zn , CP Cu ; High-precision ground magnetic survey data include: ΔT A , ΔT R ; Elements of gas content in soil include: CQ H2S , CP S02 , CP CH4 , CP He .

[0013] The step 2 comprises:

[0014] Step 2.1: Use the radial basis function method to interpolate the values ​​of each uranium mineralization indicator element, and then extract the values ​​of each uranium mineralization indicator element to form a set of uranium mineralization indicator element values ​​ptest;

[0015] Step 2.2: Extract the values ​​of each indicator element at the drilling position, and use the piecewise cubic Hermitian polynomial interpolation method to interpolate the values ​​of each indicator element at the drilling position, and then extract the value t of each element at the drilling position.

[0016] The step 3 comprises:

[0017] Step 3.1: Select a machine learning method based on the drill hole distribution density value in the uranium exploration area;

[0018] Step 3.2: Define the label dataset according to the uranium drill hole type;

[0019] Step 3.3: Use BP neural network method or S4VM method to train and test the model and calculate the uranium mineralization probability value.

[0020] The specific step 3.1 is: when the borehole distribution density value in the uranium exploration area is ≥ 2 / km2 , machine learning uses BP neural network method; when the drill hole distribution density value in the uranium exploration area is less than 2 / km 2 ,Machine learning adopts S4VM semi-supervised support vector machine learning method.

[0021] The step 3.2 is specifically as follows: when the borehole is an industrial hole of a uranium mine, the label data value of the borehole is 1; when the borehole is a uranium mineralized hole, the label data value of the borehole is 0.8; when the borehole is an abnormal uranium mine hole, the label data value of the borehole is 0.5; when the borehole is a uranium mine hole without ore, the label data value of the borehole is 0; the set of label data values ​​of the boreholes is formed into a label data set.

[0022] In step 3.3, the BP neural network method is used to train and test the model, and the calculation of the uranium mineralization probability value includes:

[0023] Step 3.3.1a: Establish an initial training data set p, where p is the set of all uranium mineralization indicator element values ​​at the drilling location;

[0024] Step 3.3.2a: Create a label dataset t, where t is the label dataset corresponding to p;

[0025] Step 3.3.3a: Create a training data set ptest to be tested, where ptest is a set of all uranium mineralization indicator factor values;

[0026] Step 3.3.4a: Based on the MATLAB platform, use the premnmx normalization function to normalize the p and t data sets to the range of [-1,1], and obtain the normalized data pn of p, the minimum value minp of each row in p, the maximum value maxp of each row in p, the normalized data tn of t, the minimum value mint of each row in t, and the maximum value maxt of each row in t;

[0027] Step 3.3.5a: Set the network parameters of the newff function as follows: Set the NodeNum1 value to 15, the NodeNum2 value to 30, the TypeNum value to 1, the TF1 value to 'tansig', the TF2 value to 'tansig', and the TF3 value to 'tansig';

[0028] Step 3.3.6a: Use the newff function, set the parameters net.trainParam.show to 50, net.trainParam.epochs to 60000, net.trainParam.goal to 1e-10, net.trainParam.lr=0.005, and establish the BP neural network net;

[0029] Step 3.3.7a: Use the train function to train the net;

[0030] Step 3.3.8a: Use the premnmx function to normalize ptest to obtain the normalized data set ptestn, the data set minptest of the minimum value in each row of ptestn, and the data set maxptest of the maximum value in each row of ptestn;

[0031] Step 3.3.9a: Based on the ptest, minptest, and maxptest data, use the tramnmx function to calculate the anti-normalized data set p2n of ptest;

[0032] Step 3.3.10a: Based on the trained net and p2n, use the sim function to calculate the simulation data set an;

[0033] Step 3.3.11a: Based on an, use the postmnmx function to calculate the uranium mineralization probability value a.

[0034] In step 3.3, the S4VM method is used to calculate the uranium mineralization probability value, including:

[0035] Step 3.3.1b: Establish an initial training data set p, where p is the set of all uranium mineralization indicator element values ​​at the drilling location;

[0036] Step 3.3.2b: Create a label dataset t, where t is the label dataset corresponding to p;

[0037] Step 3.3.3b: Create an unlabeled instance set unlabelInstance, where unlabelInstance is a set of uranium mineralization indicator element values ​​other than p;

[0038] Step 3.3.4b: Using the MATLAB platform, set the parameter kernel value to 'RBF', C1 value to 100, C2 value to 0.1, and sampleTime value to 100;

[0039] Step 3.3.5b: Based on p, t, unlabelInstance, kernel, C1, C2, and sampleTime, use the S4VM function to establish a prediction model prediction;

[0040] Step 3.3.6b: Based on the prediction model, use the size function and length function to calculate the uranium mineralization probability value a.

[0041] The step 4 is specifically as follows: the improved Shepard method is used to perform plane interpolation on the uranium mineralization probability value a at each measuring point, and the area with a value ≥ 0.5 is delineated as a favorable area for uranium mineralization.

[0042] The beneficial technical effects of the present invention are:

[0043] 1. A method for delineating favorable uranium mineralization areas based on a machine learning algorithm provided by the present invention is used. By selecting uranium mineralization indicator factors that reflect the mineralization conditions of different types of uranium ore, based on the uranium ore findings of a small number of existing drill holes in the uranium exploration area, a model training and prediction of the uranium mineralization indicator factors and the uranium ore findings are carried out with the help of a machine learning algorithm, and a potential connection between the uranium mineralization indicator factors and the uranium ore findings is established, and the uranium mineralization probability in the section without drilling control is obtained in particular, so as to delineate favorable uranium mineralization areas.

[0044] 2. The method provided by the present invention is a method for delineating favorable uranium mineralization areas based on a machine learning algorithm. The method is suitable for situations where there are a small number of known drill holes in the uranium exploration area and conventional uranium exploration geophysical and geochemical methods such as ground radioactivity measurement, geochemical measurement and magnetic measurement have been carried out, but the distribution information of uranium ore bodies in the uranium exploration area has not yet been mastered.

[0045] 3. Compared with the conventional method of determining favorable uranium mineralization areas mainly by delineating superimposed abnormal areas of uranium mineralization elements, the method for delineating favorable uranium mineralization areas based on a machine learning algorithm provided by the present invention can make full use of existing physical and chemical prospecting information and drilling information to achieve accurate delineation of favorable uranium mineralization areas, and then assist in accurately delineating uranium mineralization target areas, guide the fine layout of uranium ore drilling, reduce the probability of the occurrence of boreholes without ore, and achieve the effect of efficient and economical uranium prospecting. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 A flow chart of a method for delineating favorable uranium mineralization areas based on a machine learning algorithm provided by the present invention;

[0047] Figure 2 This is a distribution map of uranium mineralization probability values ​​in a certain uranium exploration area obtained by the S4VM method in Example 1 of the present invention, in which: 1-corrosion source area; 2-floodplain subphase; 3-oxidized sand body in the river channel; 4-residual gray sand body in the river channel; 5-corrosion source area boundary; 6-ancient river channel boundary; 7-inferred redox front line; 8-nuclear industry uranium industrial hole; 9-nuclear industry uranium mineralized hole; 10-nuclear industry uranium abnormal hole; 11-nuclear industry non-ore hole; 12-coalfield uranium mineralized hole; 13-coalfield uranium abnormal hole; 14-coalfield non-ore hole; 15-nuclear industry uranium industrial hole and number in 2019. DETAILED DESCRIPTION

[0048] The present invention is further described in detail below with reference to the accompanying drawings and embodiments.

[0049] like Figure 1 As shown, the present invention provides a method for delineating favorable uranium mineralization areas based on a machine learning algorithm, which specifically includes the following steps:

[0050] Step 1: Select uranium mineralization indicator elements.

[0051] Based on the uranium mineralization conditions in the uranium exploration area, the existing radioactive exploration data, geochemical exploration data, high-precision magnetic survey data and drilling ore conditions in the exploration area are compared and comprehensively analyzed to determine the radioactive indicator elements, soil geochemical indicator elements, soil metal activity indicator elements, magnetic indicator elements, and soil gas content indicator elements that can indicate the uranium mineralization conditions. The indicator elements can be multiple or one. Among them, the radioactive exploration elements include: CR U , CR Th , CR K , C Rn 、TLD、C Po-210 (representing ground gamma spectroscopy measurement of uranium content, thorium content, potassium content, soil radon content, soil thermoluminescence dose, and soil polonium-210 content); soil geochemical elements include: CG U , CG Th , CG K , CG Mo , CG V , CG Pb , CG Zn , CG Cu (representing soil geochemical measurement of uranium content, thorium content, potassium content, molybdenum content, vanadium content, lead content, zinc content, and copper content in turn); soil metal active elements include: CP U , CP Th , CP K , CP Mo , CP V , CP Pb , CP Zn , CP Cu (represents soil metal activity state measurement uranium content, thorium content, potassium content, molybdenum content, vanadium content, lead content, zinc content, copper content); ground high-precision magnetic measurement data include: ΔT A , ΔT R (representing total magnetic anomaly and residual magnetic anomaly of cover layer respectively); the gas content elements in the soil include: CQ H2S , CP S02 , CP CH4 , CP He (representing the hydrogen sulfide gas content, sulfur dioxide gas content, methane content, and helium content in the soil respectively).

[0052] For different uranium mineralization types, the selection of uranium mineralization indicator factor E is as follows:

[0053] For interlayer oxidation zone sandstone type uranium deposits: CR U , C Rn 、TLD、CP U , CP Mo , CP V , ΔT A , ΔT R .

[0054] For exudative sandstone uranium deposits: C Rn 、TLD、CP U , CQ H2S , CP S02 , CP CH4 , CP He .

[0055] For paleo-river channel sandstone-type uranium deposits: C Rn 、TLD、CP U , ΔT R .

[0056] For volcanic rock type uranium deposits: CR U , CR Th , CR K , Tc, C Rn , C Po-210 , CG U , CG Th , CG K , CG Mo , CG V , CG Pb , CG Zn , CG Cu , CP U , CP Th , CP Mo , CP V , CP Pb , CP Zn , CP Cu , CP He .

[0057] For granite-type uranium deposits: CR U , CR Th , CR K , C Rn 、TLD、C Po-210 , CP U , CP Th , CP K ,,CP He , ΔT A .

[0058] Step 3.3.1b: Establish an initial training data set p, where p is the set of all uranium mineralization indicator element values ​​at the drill hole location.

[0059] Step 3.3.2b: Create a labeled dataset t, where t is the labeled dataset corresponding to p.

[0060] Step 3.3.3b: Create an unlabeled instance set unlabelInstance, where unlabelInstance is a set of uranium mineralization indicator factor values ​​other than p.

[0061] Step 3.3.4b: Using the MATLAB platform, set the parameter kernel value to 'RBF' (RBF is the abbreviation of Radial Basis Function, which means "radial basis function"), C1 value to 100, C2 value to 0.1, and sampleTime value to 100.

[0062] Step 3.3.5b: Based on p, t, unlabelInstance, kernel, C1, C2, and sampleTime, use the S4VM function to establish a prediction model prediction.

[0063] Step 3.3.6b: Based on the prediction model, use the size function and length function to calculate the uranium mineralization probability value a.

[0064] Step 2.1: Use the “radial basis function method” to interpolate the values ​​of each uranium mineralization indicator element, and then extract the values ​​of each uranium mineralization indicator element according to the grid size of 10 meters × 10 meters to form the value set ptest of each uranium mineralization indicator element.

[0065] Step 2.2: Extract the values ​​of each indicator element at the drilling position, and use the "piecewise cubic Hermitian polynomial interpolation method" to interpolate the values ​​of each indicator element at the drilling position, and then extract the value t of each element at the drilling position.

[0066] Step 3: Establish a machine learning model to calculate the probability value of uranium mineralization.

[0067] Step 3.1: Select a machine learning method based on the drill hole distribution density value in the uranium exploration area.

[0068] When the drilling density in the uranium exploration area is ≥ 2 / km 2 , machine learning uses BP neural network method; when the drill hole distribution density value in the uranium exploration area is less than 2 / km 2 ,Machine learning adopts S4VM semi-supervised support vector machine learning method.

[0069] Step 3.2: Define the label dataset based on the uranium drill hole type.

[0070] The label data set is a collection of label data values ​​at the borehole. When the borehole is an industrial uranium hole, the label data value of the borehole is 1; when the borehole is a uranium mineralized hole, the label data value of the borehole is 0.8; when the borehole is an abnormal uranium hole, the label data value of the borehole is 0.5; when the borehole is a uranium non-ore hole, the label data value of the borehole is 0.

[0071] Step 3.3: Use BP neural network method or S4VM method to train and test the model and calculate the uranium mineralization probability value.

[0072] (A) Using BP neural network method to train and test the model and calculate the probability value of uranium mineralization

[0073] Step 3.3.1a: Establish an initial training data set p, where p is the set of all uranium mineralization indicator element values ​​at the drill hole location.

[0074] Step 3.3.2a: Create a labeled dataset t, where t is the labeled dataset corresponding to p.

[0075] Step 3.3.3a: Create a training data set ptest to be tested, where ptest is a set of all uranium mineralization indicator factor values.

[0076] Step 3.3.4a: Based on the MATLAB platform, use the premnmx normalization function to normalize the p and t data sets to the range of [-1,1] to obtain pn, minp, maxp, tn, mint, and maxt.

[0077] Where, pn: the normalized data of matrix p;

[0078] tn: normalized data of matrix t;

[0079] minp: column vector, the minimum value of each row in p;

[0080] mint: column vector, the minimum value of each row in t;

[0081] maxp: column vector, the maximum value of each row in p;

[0082] mint: Column vector, the maximum value of each row in t.

[0083] Step 3.3.5a: Set the network parameters of the newff function as follows:

[0084] NodeNum1 value is 15;

[0085] NodeNum2 value is 30;

[0086] TypeNum value is 1;

[0087] TF1 value is 'tansig';

[0088] TF2 value is 'tansig';

[0089] The TF3 value is 'tansig'.

[0090] Among them, NodeNum1: the number of nodes in the first hidden layer;

[0091] NodeNum2: the number of nodes in the second hidden layer;

[0092] TypeNum: Output dimension

[0093] TF1: input layer transfer function 1;

[0094] TF2: input layer transfer function 2;

[0095] TF3: input layer transfer function 3;

[0096] tansig: Hyperbolic tangent activation function.

[0097] Step 3.3.6a: Use the newff function, set the parameters net.trainParam.show to 50, net.trainParam.epochs to 60000, net.trainParam.goal to 1e-10, net.trainParam.lr=0.005, and establish the BP neural network net.

[0098] Step 3.3.7a: Use the train function to train the net.

[0099] Step 3.3.8a: Use the premnmx function to normalize ptest to obtain the values ​​of ptestn, minptest, and maxptest. Among them, ptestn is the normalized data set; minptest is a column vector, which is the data set of the minimum value in each row of ptestn; maxptest is a column vector, which is the data set of the maximum value in each row of ptestn.

[0100] Step 3.3.9a: Based on the ptest, minptest, and maxptest data, use the tramnmx function to calculate the anti-normalized data set p2n of ptest.

[0101] Step 3.3.10a: Based on the trained net and p2n, use the sim function to calculate the simulation data set an.

[0102] Step 3.3.11a: Based on an, use the postmnmx function to calculate the uranium mineralization probability value a.

[0103] (B) Calculation of uranium mineralization probability using the S4VM method

[0104] Step 3.3.1b: Establish an initial training data set p, where p is the set of all uranium mineralization indicator element values ​​at the drill hole location.

[0105] Step 3.3.2b: Create a labeled dataset t, where t is the labeled dataset corresponding to p.

[0106] Step 3.3.3b: Create an unlabeled instance set unlabelInstance, where unlabelInstance is a set of uranium mineralization indicator factor values ​​other than p.

[0107] Step 3.3.4b: Using the MATLAB platform, set the parameter kernel value to 'RBF' (RBF is the abbreviation of Radial Basis Function, which means "radial basis function"), C1 value to 100, C2 value to 0.1, and sampleTime value to 100.

[0108] Step 3.3.5b: Based on p, t, unlabelInstance, kernel, C1, C2, and sampleTime, use the S4VM function to establish a prediction model prediction.

[0109] Step 3.3.6b: Based on the prediction model, use the size function and length function to calculate the uranium mineralization probability value a.

[0110] Step 4: Based on the uranium mineralization probability value, delineate the favorable uranium mineralization area.

[0111] The “improved Shepard method” is used to perform plane interpolation on the uranium mineralization probability value a at each measuring point, and the area with a value ≥ 0.5 is identified as a favorable area for uranium mineralization.

[0112] Example 1 Taking a certain interlayer oxidation zone type sandstone uranium ore exploration area as an example, the method of the present invention is used to delineate the favorable uranium mineralization area

[0113] The exploration area of ​​a certain interlayer oxidation zone sandstone uranium deposit is 693km 2 There are 32 known boreholes in the area, with a borehole density of 0.046 / km 2 , there are soil radon gas measurement and metal activity measurement data in the exploration area. According to the method provided by the present invention, the preferred uranium mineralization indicator factor is C Rn , CP U , CP Mo, CP V , using the S4VM method, the uranium mineralization probability distribution map is obtained, such as Figure 2 As shown in Figure 2, the favorable areas for uranium mineralization are Figure 2 The section with medium uranium mineralization probability greater than or equal to 0.5.

[0114] The present invention is described in detail above with reference to the accompanying drawings and embodiments, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge of ordinary technicians in the field without departing from the purpose of the present invention. The contents not described in detail in the present invention can adopt the existing technology.

Claims

1. A method for delineating favorable uranium mineralization areas based on a machine learning algorithm, characterized in that: The method comprises: Step 1: Select uranium mineralization indicator elements; Step 2: Preprocessing of uranium mineralization indicator value data; Step 3: Establish a machine learning model to calculate the probability of uranium mineralization; Step 4: Based on the uranium mineralization probability value, delineate the favorable uranium mineralization area.

2. The method for delineating favorable uranium mineralization areas based on a machine learning algorithm according to claim 1, characterized in that: The step 1 specifically comprises: based on the existing radioactive exploration data, geochemical exploration data, high-precision magnetic survey data and drilling ore-seeking conditions in the uranium exploration area, and according to the mineralization type of the uranium ore in the uranium exploration area, determining the radioactive indicator elements, soil geochemical indicator elements, soil metal activity indicator elements, magnetic indicator elements, and soil gas content indicator elements that can indicate the uranium mineralization conditions.

3. The method for delineating favorable uranium mineralization areas based on a machine learning algorithm according to claim 2, characterized in that: The radioactive exploration elements include: CR U , CR Th , CR K , C Rn 、TLD、C Po-210 ;Soil geochemical elements include: CG U , CG Th , CG K , CG Mo , CG V , CG Pb , CG Zn , CG Cu ; The active elements of soil metals include: CP U , CP Th , CP K , CP Mo , CP V , CP Pb , CP Zn , CP Cu ; High-precision ground magnetic survey data include: ΔT A , ΔT R ; Elements of gas content in soil include: CQ H2S , CP S02 , CP CH4 , CP He .

4. The method for delineating favorable uranium mineralization areas based on a machine learning algorithm according to claim 1, characterized in that: The step 2 comprises: Step 2.1: Use the radial basis function method to interpolate the values ​​of each uranium mineralization indicator element, and then extract the values ​​of each uranium mineralization indicator element to form a set of uranium mineralization indicator element values ​​ptest; Step 2.2: Extract the values ​​of each indicator element at the drilling position, and use the piecewise cubic Hermitian polynomial interpolation method to interpolate the values ​​of each indicator element at the drilling position, and then extract the value t of each element at the drilling position.

5. The method for delineating favorable uranium mineralization areas based on a machine learning algorithm according to claim 1, characterized in that: The step 3 comprises: Step 3.1: Select a machine learning method based on the drill hole distribution density value in the uranium exploration area; Step 3.2: Define the label dataset according to the uranium drill hole type; Step 3.3: Use BP neural network method or S4VM method to train and test the model and calculate the uranium mineralization probability value.

6. The method for delineating favorable uranium mineralization areas based on a machine learning algorithm according to claim 5, characterized in that: The specific step 3.1 is: when the borehole distribution density value in the uranium exploration area is ≥ 2 / km 2 , machine learning uses BP neural network method; when the drill hole distribution density value in the uranium exploration area is less than 2 / km 2 ,Machine learning adopts S4VM semi-supervised support vector machine learning method.

7. The method for delineating favorable uranium mineralization areas based on a machine learning algorithm according to claim 5, characterized in that: The step 3.2 is specifically as follows: when the borehole is an industrial hole of a uranium mine, the label data value of the borehole is 1; when the borehole is a uranium mineralized hole, the label data value of the borehole is 0.8; when the borehole is an abnormal uranium mine hole, the label data value of the borehole is 0.5; when the borehole is a uranium mine hole without ore, the label data value of the borehole is 0; the set of label data values ​​of the boreholes is formed into a label data set.

8. The method for delineating favorable uranium mineralization areas based on a machine learning algorithm according to claim 5, characterized in that: In step 3.3, the BP neural network method is used to train and test the model, and the calculation of the uranium mineralization probability value includes: Step 3.3.1a: Establish an initial training data set p, where p is the set of all uranium mineralization indicator element values ​​at the drilling location; Step 3.3.2a: Create a label dataset t, where t is the label dataset corresponding to p; Step 3.3.3a: Create a training data set ptest to be tested, where ptest is a set of all uranium mineralization indicator factor values; Step 3.3.4a: Based on the MATLAB platform, use the premnmx normalization function to normalize the p and t data sets to the range of [-1,1], and obtain the normalized data pn of p, the minimum value minp of each row in p, the maximum value maxp of each row in p, the normalized data tn of t, the minimum value mint of each row in t, and the maximum value maxt of each row in t; Step 3.3.5a: Set the network parameters of the newff function as follows: Set the NodeNum1 value to 15, the NodeNum2 value to 30, the TypeNum value to 1, the TF1 value to 'tansig', the TF2 value to 'tansig', and the TF3 value to 'tansig'; Step 3.3.6a: Use the newff function, set the parameters net.trainParam.show to 50, net.trainParam.epochs to 60000, net.trainParam.goal to 1e-10, net.trainParam.lr=0.005, and establish the BP neural network net; Step 3.3.7a: Use the train function to train the net; Step 3.3.8a: Use the premnmx function to normalize ptest to obtain the normalized data set ptestn, the data set minptest of the minimum value in each row of ptestn, and the data set maxptest of the maximum value in each row of ptestn; Step 3.3.9a: Based on the ptest, minptest, and maxptest data, use the tramnmx function to calculate the anti-normalized data set p2n of ptest; Step 3.3.10a: Based on the trained net and p2n, use the sim function to calculate the simulation data set an; Step 3.3.11a: Based on an, use the postmnmx function to calculate the uranium mineralization probability value a.

9. The method for delineating favorable uranium mineralization areas based on a machine learning algorithm according to claim 5, characterized in that: In step 3.3, the S4VM method is used to calculate the uranium mineralization probability value, including: Step 3.3.1b: Establish an initial training data set p, where p is the set of all uranium mineralization indicator element values ​​at the drilling location; Step 3.3.2b: Create a label dataset t, where t is the label dataset corresponding to p; Step 3.3.3b: Create an unlabeled instance set unlabelInstance, where unlabelInstance is a set of uranium mineralization indicator element values ​​other than p; Step 3.3.4b: Using the MATLAB platform, set the parameter kernel value to 'RBF', C1 value to 100, C2 value to 0.1, and sampleTime value to 100; Step 3.3.5b: Based on p, t, unlabelInstance, kernel, C1, C2, and sampleTime, use the S4VM function to establish a prediction model prediction; Step 3.3.6b: Based on the prediction model, use the size function and length function to calculate the uranium mineralization probability value a.

10. The method for delineating favorable uranium mineralization areas based on machine learning algorithm according to claim 1, characterized in that: The step 4 is specifically as follows: the improved Shepard method is used to perform plane interpolation on the uranium mineralization probability value a at each measuring point, and the area with a value ≥ 0.5 is delineated as a favorable area for uranium mineralization.

Images