A risk assessment method for concealed mineral exploration based on tectonic geochemical collaborative simulation

Abstract

The present invention provides a hidden mine exploration risk assessment method of tectonic geochemical collaborative simulation, which obtains the main variable by tectonic geochemical data; extracts the structural line density value with the same position as the tectonic rock sampling point as the covariate; calculates the semivariogram of the main variable and the covariate, and its cross semivariogram, integrates the main variable and the covariate, and performs collaborative kriging estimation; performs sequential inertial Gaussian collaborative simulation, obtains the simulation realization of equal probability distribution, and then obtains the mean type simulation realization; uses the P-A method to draw the change relationship curve between the mean type simulation realization and the cumulative area percentage and the cumulative mineral point percentage, and determines the optimal threshold; uses the local and spatial uncertainty algorithm to calculate the probability expression mode of tectonic geochemical anomalies with simulation values ​​greater than the threshold, and performs risk assessment. The beneficial effects of the present invention are: providing technical means and decision-making services for the identification of ore-causing anomalies, the location prediction of hidden ore deposits, and the risk assessment of ore deposit exploration.

Description

Technical Field

[0001] The invention relates to the technical field of mineral exploration, and in particular to a hidden mine exploration risk assessment method based on structural geochemical collaborative simulation. Background Art

[0002] In recent years, with the implementation of the national crisis mine replacement resource prospecting work and the urgent need for deep geological prospecting, a new upsurge of geological prospecting in crisis mines, deep and peripheral areas of old mines has been set off, promoting the development of structural geochemical prospecting theory and exploration technology and its application in hidden mine prediction and deep prospecting. Studies have shown that structural geochemical prospecting methods have achieved significant prospecting results in deep and peripheral geological prospecting of polymetallic deposits such as gold, copper, lead, zinc, and tin that are clearly controlled by structures.

[0003] Tectonic rocks contain rich information about the aggregation and dispersion of mineralizing materials. They are the only way to connect deep ore bodies with shallow surface mineralized primary halos, and the two are corresponding and consistent. In tectonic geochemical prospecting activities, the focus is still on the processing and analysis of tectonic rock geochemical data, and the tectonic system has not been effectively integrated into prospecting theory and mathematical analysis. In addition, due to the sparse characteristics of tectonic geochemical sampling data, the use of deterministic modeling technology will lead to the problem of data smoothing effect. The modeling results cannot truly reflect the spatial distribution structure and morphological characteristics of the original information, which brings great uncertainty to the prediction and evaluation work. This uncertainty will be propagated through the subsequent analysis and decision-making process, and will have a profound impact on the final result, and will not be effectively identified by tectonic geochemical anomalies. Summary of the invention

[0004] In order to solve the above problems, the present invention provides a hidden mine exploration risk assessment method based on tectonic geochemical collaborative simulation, which mainly includes:

[0005] First, the structural geochemical data has the characteristics of compositional data and is a kind of "fixed sum" data, that is, the sum of the contents of all elements tested in each sample is equal to a fixed value; then, the central logarithmic ratio transformation is used to convert the compositional geochemical data into an open Euclidean space, and the transformed data obeys an approximate normal distribution to meet the assumptions of the sequential Gaussian co-simulation (the data obeys or approximately obeys a normal distribution); in the co-simulation process, a target variable after the central logarithmic ratio transformation is defined as the main variable; through the GIS spatial analysis technology, the spatial distribution map of the structural line density is drawn, and the structural line density value at the same location as the structural rock sampling point is extracted and defined as the covariate. If the variable does not obey the normal distribution, a logarithmic transformation is performed; the semivariogram of the main variable and the covariate is calculated , and the cross-semivariogram between the two, and obtain relevant geostatistical parameters, such as nugget value, sill value, range, etc. On this basis, the main variables and covariates are integrated through the global co-kriging method to perform co-kriging estimation; sequential Gaussian co-simulation can be performed to obtain a series of simulation realizations of equal probability distribution and mean-type simulation realizations; based on the mean-type simulation realization, cumulative area percentage and cumulative mineral point percentage, the PA (prediction-area) method is used to draw the change relationship curve between the three, and the optimal threshold is determined; the local uncertainty and spatial uncertainty algorithms are used to calculate the probability expression pattern of structural geochemical anomalies with simulation values ​​greater than the optimal threshold, providing technical support and decision-making services for the delineation of prospecting target areas and the risk assessment of concealed mineral exploration.

[0006] Furthermore, the centered logarithmic ratio transformation is used to transform the structural geochemical data into an open Euclidean space to eliminate the closure effect of the data. The transformed data approximately obey the normal distribution to meet the assumptions of the sequential-inertial Gaussian co-simulation. The centered logarithmic ratio transformation formula is as follows:

[0007]

[0008] Among them, clr(X) is the matrix composed of D variables after the central logarithmic ratio transformation, y i is the original geochemical variable, is the geometric mean of D original geochemical variables, i = 1, 2,…, D.

[0009] Furthermore, the tectonic line density distribution map was drawn through GIS spatial analysis technology, and the tectonic line density value at the same location as the tectonic rock sampling point was extracted as a covariate. If the covariate does not obey the normal distribution, a central logarithmic transformation is performed to meet the assumptions of the sequential inertial Gaussian co-simulation.

[0010] Furthermore, the semivariogram formula is:

[0011]

[0012] Among them, γ is the semivariogram; h is the spatial separation distance of the sample points, that is, the lag distance; n(h) is the total number of sample pairs under the interval h; z(x k ) is the spatial position x k The value at z(x k +h) is the spatial position (x k +h) value.

[0013] Furthermore, the cross-semivariogram formula is:

[0014]

[0015] Among them, γ ij (h) is the cross semivariogram of variables i and j; h is the spatial separation distance of sample points, i.e., the lag distance; n(h) is the total number of sample pairs under interval h; z i (x k ) is the spatial position x k The value of variable i at position z i (x k +h) is the spatial position (x k +h) the value of variable i; z j (x k ) is the spatial position x k The value of variable j at z j (x k +h) is the spatial position (x k +h) the value of variable j.

[0016] Furthermore, by establishing the neighboring point principal variable z1(u i ) and covariate z2(u j ) to calculate the co-kriging estimate of the main variable z1 Its expression is as follows:

[0017]

[0018] Constraints:

[0019]

[0020]

[0021] Among them, is the estimated value of the main variable z1 at u; λ i and λ j are the weights of the main variable z1 and the covariate z2 respectively; n(u) and m(u) are the number of neighboring points of the main variable z1 and the covariate z2 at u, that is, the number of neighboring points within the range; u i and uj are the sampling positions of z1 and z2 respectively.

[0022] Furthermore, under the condition that both the main variable z1 and the covariate z2 obey or approximately obey the normal distribution, the sequence-inertial Gaussian co-simulation process is as follows:

[0023] (1) Let the number of unsampled grid nodes be p, and define a random path so that each grid node at the unsampled point position u0 is calculated only once;

[0024] (2) At the kth node u0(k), 1≤k≤p, the conditional data is the neighboring points of the grid node u0(k), that is, the main variable z1(u i ), covariate z2(u j ), and the first (k-1) simulation values composition;

[0025] (3) Estimate the parameters of the local Gaussian conditional cumulative distribution function of the main variable at u0(k), namely the mean and variance, where the mean is equal to the estimated value at u0(k) The variance is equal to the variance σ at u0(k) 2 ;

[0026] (4) Using Monte Carlo simulation, a simulation value z(u0(k)) is randomly extracted from the local Gaussian conditional cumulative distribution function, and then added to the k-th conditional simulation data set for simulation of other grid nodes;

[0027] (5) If k<p, repeat steps (2)-(4) until k=p, that is, all nodes u0(k) are simulated;

[0028] The above steps only complete the first simulation implementation. In order to generate a series of simulation implementations, steps (1)-(5) will be simulated using different random paths.

[0029] Furthermore, the mean-valued simulation achieves The formula is:

[0030]

[0031] in, is the cokriging estimate z * The l-th simulation realization of (u0), l = 1, 2, ..., L, L is the total number of simulation realizations, and u0 is the grid node in the sequential-inertial Gaussian collaborative simulation process.

[0032] Furthermore, the formula of the local uncertainty algorithm is:

[0033]

[0034] Among them, z(x * ) is the simulation point x * The value at z c is the optimal threshold achieved by the mean-value simulation; N is the total number of conditional simulations; n(x * ) is the number of times the simulation value of the point exceeds the optimal threshold in all N simulations.

[0035] Furthermore, the formula of the spatial uncertainty algorithm is:

[0036]

[0037] Among them, z c The best threshold achieved for mean-type simulations; is the spatial position x k The simulated value at, k = 1, 2, ..., m; n (x1 * , x2 * , ..., x m * ) is the number of times the simulated values ​​of all m simulation points exceed the optimal threshold; N is the total number of conditional simulations.

[0038] The beneficial effects of the technical solution provided by the present invention are as follows: the present invention has a guiding role in inferring the migration direction of ore-forming fluids during tectonic activities and studying the activation, migration, enrichment and dispersion laws of ore-forming elements, and provides technical means and decision-making services for identifying ore-causing anomalies, locating and predicting hidden ore deposits, and assessing the risks of ore exploration. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

[0040] Figure 1 It is a flow chart of a hidden mine exploration risk assessment method of structural geochemical collaborative simulation in an embodiment of the present invention.

[0041] Figure 2 In the embodiment of the present invention, (a) is the original Pb element content distribution diagram, that is, the original variable Raw(Pb), and (b) is the Pb element distribution diagram after the central logarithmic ratio transformation, that is, the main variable z1(Pb).

[0042] Figure 3 It is a structural line density map of the ore-concentrated area structural system in the embodiment of the present invention.

[0043] Figure 4 In the figure, (a) is the semivariogram of the main variable z1(Pb), (b) is the semivariogram of the covariate z2(FD), and (c) is the cross semivariogram of the main variable z1(Pb) and the covariate z2(FD).

[0044] Figure 5 1 and 2. This is the spatial distribution pattern of the tectonic geochemical collaborative simulation implementation in the embodiment of the present invention: (a) is the 100th simulation implementation, and (b) is the 134th simulation implementation.

[0045] Figure 6 This is the mean value simulation implementation in the embodiment of the present invention. E The spatial distribution pattern of Pb.

[0046] Figure 7 The ROC curve in the embodiment of the present invention evaluates the classification performance of the three variables of the lead element.

[0047] Figure 8 The PA curve evaluation mean value simulation is implemented in the embodiment of the present invention. E (Pb) prediction performance.

[0048] Fig. 9 It is the probability expression mode of structural geochemical anomaly under the condition that the simulation value is greater than 0.67 in the embodiment of the present invention. DETAILED DESCRIPTION

[0049] In order to have a clearer understanding of the technical features, purposes and effects of the present invention, specific embodiments of the present invention are now described in detail with reference to the accompanying drawings.

[0050] Aiming at the prospecting needs of crisis mines, periphery of mining areas and deep mines, the present invention provides a hidden mine exploration risk assessment method based on structural geochemical collaborative simulation. First, in view of the closure effect problem of structural geochemical data, the central logarithmic ratio transformation method is used to open the data and make the data obey the approximate normal distribution; secondly, the co-kriging method is used to calculate the main variable, covariate and the semi-variogram between the two; then, the sequential inertial Gaussian collaborative simulation is used to add structural ore-controlling elements related to the main variable as covariates to achieve the objective expression of structural geochemical collaborative simulation and its spatial structure; finally, through the spatial uncertainty and local uncertainty modeling technology, the uncertainty measurement of structural rock geochemical anomalies at unsampled points is achieved.

[0051] Please refer to Figure 1 , Figure 1 Flow chart of a method for risk assessment of concealed mine exploration by tectonic geochemical collaborative simulation in an embodiment of the present invention, comprising:

[0052] S1: Perform central logarithmic ratio transformation on structural geochemical data to obtain the main variables;

[0053] S2: Draw the spatial distribution map of tectonic line density and extract the tectonic line density value with the same location as the tectonic rock sampling point as the covariate;

[0054] S3: Calculate the semivariogram of the main variable and the covariate, and the cross semivariogram between the two, and obtain relevant geostatistical parameters, wherein the relevant geostatistical parameters include nugget value, sill value and range;

[0055] S4: Based on step S3, the main variables and covariates are integrated through the global cokriging method to perform cokriging estimation;

[0056] S5: Execute sequential Gaussian co-simulation to obtain a series of simulation realizations with equal probability distribution and mean-valued simulation realizations;

[0057] S6: Based on the mean-value simulation realization, cumulative area percentage and cumulative mineral point percentage, the PA method is used to draw the change relationship curve between the three and determine the optimal threshold;

[0058] S7: Adopt local uncertainty and spatial uncertainty algorithms to calculate the probability expression pattern of structural geochemical anomalies with simulation values ​​greater than the optimal threshold, conduct risk assessment for concealed mineral exploration, and provide technical and decision-making services for the delineation of prospecting targets and risk assessment for concealed mineral exploration.

[0059] The specific content of this method is as follows:

[0060] Based on multivariate structural geochemical data, the centered log-ratio (CLR) transformation is first used to obtain the main variable z1. The CLR transformation not only eliminates the closure effect of the composition data, but also makes the main variable z1 obey or approximately obey the normal distribution to meet the assumptions of the sequential Gaussian co-simulation. The formula for the centered log-ratio transformation is as follows:

[0061]

[0062] Where clr(X) is the matrix of D variables after the central logarithmic ratio transformation, y i is the original geochemical variable; is the geometric mean of D original geochemical variables, i = 1, 2, ..., D.

[0063] In order to calculate the cross semivariogram between the main variable and the covariate, the tectonic line density distribution map is drawn through GIS spatial analysis technology. It is necessary to extract the density value of the tectonic line at the same position as the tectonic rock sample (that is, the same tectonic rock sampling position) in the map to obtain the auxiliary variable z2, also called the covariate. Check whether the covariate obeys or approximately obeys the normal distribution. If it does not obey the normal distribution, a logarithmic transformation is performed.

[0064] The semivariogram of the main variable z1 and the covariate z2, as well as the cross-semivariogram between the two, are calculated to obtain relevant geostatistical parameters, such as nugget value, sill value, and range, which are used to describe the spatial correlation characteristics of the variables and the subsequent sequential Gaussian co-simulation. The sequential Gaussian co-simulation technology introduces the co-kriging algorithm, which integrates the main variable and the covariate to perform co-kriging estimation.

[0065] ①Semivariogram function formula:

[0066]

[0067] Among them, γ is the semivariogram; h is the spatial separation distance of the sample points, that is, the lag distance; n(h) is the total number of sample pairs under the interval h; z(x k ) is the spatial position x k The value at z(x k +h) is the spatial position (x k +h) value.

[0068] ②Cross-semivariogram function formula:

[0069]

[0070] Among them, γ ij is the cross semivariogram of variables i and j; h is the spatial separation distance of sample points, i.e., the lag distance; n(h) is the total number of sample pairs under interval h; z i (x k ) is the spatial position x k The value of variable i at position z i (x k +h) is the spatial position (x k +h) the value of variable i; z j (x k ) is the spatial position x k The value of variable j at z j (x k +h) is the spatial position (x k +h) the value of variable j; when i = j, γ ij (h) is the semivariogram of variable i or variable j.

[0071] In this part, the global co-kriging method is used to estimate the semivariogram of the main variable and the covariate, taking into account the spatial variability characteristics of the two. By estimating the cross semivariogram between the main variable and the covariate, the joint spatial variability characteristics of the two are taken into account, and the recognition ability of the main variable for the prediction target is improved.

[0072] The global co-kriging method is used to integrate the main variables z1(u i ) and covariate z2(u j ), and obtain the co-kriging estimate. This method establishes the main variable z1(u i ) and covariate z2(u j ) linear combination model, calculate the estimated value of the main variable z1 Its expression is as follows:

[0073]

[0074] Constraints:

[0075]

[0076]

[0077] Among them, is the estimated value of the main variable z1 at u; λ i and λ j are the weights of the main variable z1 and the covariate z2 respectively; n(u) and m(u) are the number of neighboring points of the main variable z1 and the covariate z2 at u, that is, the number of neighboring points within the range; u i and u j are the sampling positions of z1 and z2 respectively. Under the two constraints (Formula 5-6), the optimal weight λ i and λ j Solved by the Lagrange multiplier method while minimizing the cokriging variance.

[0078] Under the condition that both the main variable z1 and the covariate z2 obey or approximately obey the normal distribution, the sequence-inertial Gaussian co-simulation process is as follows:

[0079] (1) Let the number of unsampled grid nodes be p, and define a random path so that each grid node at the unsampled point position u0 is calculated only once;

[0080] (2) At the kth node u0(k) (1≤k≤p), the conditional data is the neighboring points of the grid node u0(k), that is, the main variable z1(u i ), covariate z2(u j ), and the first (k-1) simulation values composition;

[0081] (3) Estimate the parameters of the local Gaussian conditional cumulative distribution function (ccdf) of the primary variable at u0(k), namely the mean and variance, where the mean is equal to the estimated value at u0(k) The variance is equal to the variance σ at u0(k)2 ;

[0082] (4) Using Monte Carlo simulation, a simulation value z(u0(k)) is randomly selected from the ccdf, and then added to the kth conditional simulation data set for simulation of other grid nodes;

[0083] (5) If k<p, repeat steps (2)-(4) until k=p, that is, all nodes u0(k) are simulated.

[0084] The above steps only complete the first simulation realization. In order to generate a series of simulation realizations, steps (1)-(5) will be simulated using different random paths. From a series of simulation realizations, the mean simulation realization can be calculated.

[0085]

[0086] in, is the estimated value z * The lth simulation realization of (u0), L is the total number of simulation realizations.

[0087] This method uses root mean square error (RMSE) and mean absolute error (MAE) as evaluation indicators of model prediction accuracy:

[0088]

[0089]

[0090] The smaller the RMSE and MAE, the closer the estimated value is to the observed value, the better the simulation effect is, and the higher the prediction accuracy is.

[0091] Based on mean-value simulation The cumulative area percentage and the cumulative mineral point percentage are plotted using the PA (prediction-area) method to determine the mean simulation implementation. The optimal threshold is the minimum area containing the largest number of mineral deposits, and areas greater than this optimal threshold indicate favorable prospecting areas.

[0092] Based on a series of simulations and optimal thresholds, uncertainty evaluation technology is used to calculate the probability expression pattern of structural geochemical anomalies greater than the optimal threshold, draw probability distribution maps, delineate hidden ore prospecting target areas, and evaluate the exploration risks of mineral deposits. Uncertainty evaluation technology includes local uncertainty algorithms and spatial uncertainty algorithms.

[0093] ① Local uncertainty: It is the probability evaluation of each simulation point in the study area exceeding its own optimal threshold.

[0094]

[0095] Among them, z(x * ) is the simulation point x * The value at z c is the optimal threshold achieved by the mean-value simulation; N is the total number of conditional simulations; n(x * ) is the number of times the simulation value of the point exceeds the optimal threshold in all N simulations.

[0096] ②Spatial uncertainty: It is to evaluate the joint probability of uncertainty of all points to be simulated in the study area:

[0097]

[0098] Among them, z c The best threshold achieved for mean-type simulations; is the spatial position x k The simulated value at n(x1 * , x2 * , ..., x m * ) is the number of times the simulated values ​​of all m simulated points exceed the optimal threshold; N is the total number of conditional simulations. This algorithm calculates the probability that all points to be simulated exceed the optimal threshold at the same time during the simulation process. Specific embodiment:

[0100] This method uses the main mineralization elements (main variables) in tectonic rocks as the model object, and introduces structural elements closely related to mineralization as auxiliary variables to achieve tectonic geochemical collaborative simulation. This method takes into account the spatial variability characteristics of the main variables and covariates, as well as the joint spatial variability characteristics between the two, which can improve the detection capability of hidden mines and reduce the risk of mineral exploration.

[0101] The present invention will be further described below in conjunction with the accompanying drawings.

[0102] Data acquisition: The geochemical data of tectonic rocks used in this paper come from a certain lead-zinc ore cluster area, with a total of 1238 tectonic rock samples. The contents of 19 elements such as Pb, Zn, Cu, and Au were tested and analyzed. The spatial distribution of the lead content is as follows: Figure 2 As shown in a.

[0103] (1) The 19 original geochemical variables were transformed by central log ratio (clr) and the transformed variable Pb was defined as the main variable z1(Pb). Its spatial distribution pattern is shown in Figure 2 As shown in b.

[0104] (2) Using GIS spatial analysis, we can obtain the structural line density map of the mining area ( Figure 3 ), extract the structural line density value at the same location as the structural rock sample as the covariate z2(FD);

[0105] (3) Calculate the semivariogram of the main variable z1 and the covariate z2 through Matlab programming ( Figure 4 a-4b), and the cross semivariogram between the two ( Figure 4 c), the obtained semivariogram function parameters include nugget value, sill value and range.

[0106] (4) Based on (2) and (3), a series of equally probabilistic tectonic geochemical co-simulations can be obtained by using the sequential Gaussian co-simulation technique to characterize the spatial volatility characteristics of the main variable z1(Pb) ( Figure 5 a-5b), reproduces the global spatial correlation and local variability characteristics of the tectonic geochemical field. Thus, the mean-valued simulation can be calculated to achieve z E (Pb), such as Figure 6 The ROC (receiver operating characteristic) curve was used to evaluate the three variables of lead element, namely Raw(Pb), z1(Pb) and z E The classification performance of (Pb), AUC value shows that the mean type simulation achieves z E (Pb) significantly enhances the ability to identify lead mineralization anomalies, such as Figure 7 shown.

[0107] (5) Figure 8 As shown, the PA curve is used to evaluate the mean-value simulation to achieve z E (Pb) prediction ability for known lead-zinc deposits and determine the best z E The abnormal threshold of (Pb) is 0.67, and its corresponding probability value is 0.48. On this basis, uncertainty evaluation technology is used to establish Fig. 9 The distribution pattern of structural geochemical anomalies greater than the threshold value shown in the figure provides decision-making for the detection of hidden lead-zinc deposits and the risk assessment of mineral exploration.

[0108] The beneficial effects of the present invention are as follows: the present invention has a guiding role in inferring the migration direction of ore-forming fluids during tectonic activities and studying the activation, migration, enrichment and dispersion laws of ore-forming elements, and provides technical means and decision-making services for identifying ore-causing anomalies, locating and predicting hidden ore deposits, and assessing the risks of ore exploration.

[0109] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for risk assessment of concealed mineral exploration based on tectonic geochemical collaborative simulation, characterized in that: include: S1: The main variables were obtained by constructing geochemical data and using the central logarithmic ratio transformation; S2: Draw the spatial distribution map of tectonic line density and extract the tectonic line density value with the same location as the tectonic rock sampling point as the covariate; S3: Calculate the semivariogram of the main variable and the covariate, and the cross semivariogram between the two, and obtain relevant geostatistical parameters, wherein the relevant geostatistical parameters include nugget value, sill value and range; S4: Based on step S3, the main variables and covariates are integrated through the global cokriging method to perform cokriging estimation; S5: Execute sequential Gaussian co-simulation to obtain a series of simulation realizations with equal probability distribution and mean-valued simulation realizations; Under the condition that both the main variable z1 and the covariate z2 obey or approximately obey the normal distribution, the sequence-inertial Gaussian co-simulation process is as follows: (1) Let the number of unsampled grid nodes be p, and define a random path so that each grid node at the unsampled point position u0 is calculated only once; (2) At the kth node u0(k), 1≤k≤p, the conditional data is the neighboring points of the grid node u0(k), that is, the main variable z1(u i ), covariate z2(u j ), and the first (k-1) simulation values composition; (3) Estimate the parameters of the conditional cumulative distribution function of the main variable at u0(k), namely the mean and variance, where the mean is equal to the estimated value at u0(k) The variance is equal to the variance σ at u0(k) 2 ; (4) Using Monte Carlo simulation, a simulation value z(u0(k)) is randomly extracted from the local Gaussian conditional cumulative distribution function, and then added to the k-th conditional simulation data set for simulation of other grid nodes; (5) If k<p, repeat steps (2)-(4) until k=p, that is, all nodes u0(k) are simulated; The above steps only complete the first simulation implementation. In order to generate a series of simulation implementations, steps (1)-(5) will be simulated using different random paths; Mean-valued simulation implementation The formula is: Among them, z *(l) (u0) is the cokriging estimate z * (u0) is the,th simulation realization, l = 1, 2, ..., L, L is the total number of simulation realizations, u0 is the grid node in the sequential-inertial Gaussian collaborative simulation process; S6: Based on the mean-value simulation realization, cumulative area percentage and cumulative mineral point percentage, the PA method is used to draw the change relationship curve between the three and determine the optimal threshold; S7: Using local uncertainty and spatial uncertainty algorithms, the probability expression pattern of structural geochemical anomalies with simulation values ​​greater than the optimal threshold is calculated to conduct risk assessment of concealed mineral exploration; The formula for the local uncertainty algorithm is: Among them, z(x * ) is the simulation point x * The value at z c is the optimal threshold achieved by the mean-value simulation; N is the total number of conditional simulations; n(x * ) is the number of times the simulation value of the point exceeds the threshold in all N simulations; The formula for the spatial uncertainty algorithm is: Among them, z c The best threshold achieved for mean-type simulations; is the spatial position x k The simulated value at, k = 1, 2, ..., m; n (x1 * , x2 * ..., x m * ) is the number of times the simulated values ​​of all m simulation points exceed the optimal threshold; N is the total number of conditional simulations.

2. The method for risk assessment of concealed mine exploration based on structural geochemical collaborative simulation according to claim 1, characterized in that: In step S1, the centered logarithmic ratio transformation is used to transform the structural geochemical data into an open Euclidean space to eliminate the closure effect of the composition data. The transformed data approximately obeys the normal distribution. The centered logarithmic ratio transformation formula is as follows: Among them, clr(X) is the matrix composed of D variables after the central logarithmic ratio transformation, y i is the original geochemical variable, is the geometric mean of D original geochemical variables, i = 1, 2, ..., D.

3. The method for risk assessment of concealed mine exploration based on structural geochemical collaborative simulation according to claim 1, characterized in that: In step S2, a spatial distribution map of the structural line density is drawn through GIS spatial analysis. If the covariate does not obey the normal distribution, a logarithmic transformation is performed to meet the assumptions of the sequential Gaussian co-simulation.

4. The method for risk assessment of concealed mine exploration based on structural geochemical collaborative simulation according to claim 1, characterized in that: In step S3, the semivariogram formula is: Among them, γ is the semivariogram; h is the spatial separation distance of the sample points, that is, the lag distance; n(h) is the total number of sample pairs under the interval h; z(x k ) is the spatial position x k The value at z(x k +h) is the spatial position (x k +h) value.

5. The method for risk assessment of concealed mine exploration based on structural geochemical collaborative simulation according to claim 1, characterized in that: In step S3, the cross semivariogram function formula is: Among them, γ ij (h) is the cross semivariogram of variables i and j; h is the spatial separation distance of sample points, i.e., the lag distance; n(h) is the total number of sample pairs under interval h; z i (x k ) is the spatial position x k The value of variable i at position z i (x k +h) is the spatial position (x k +h) the value of variable i; z j (x k ) is the spatial position x k The value of variable j at z j (x k +h) is the spatial position (x k +h) the value of variable j.

6. The method for risk assessment of concealed mine exploration based on structural geochemical collaborative simulation according to claim 1, characterized in that: In step S4, by establishing the neighboring point main variable z1(u i ) and covariate z2(u j ) to calculate the co-kriging estimate of the main variable z1 Its expression is as follows: Constraints: Among them, is the estimated value of the main variable z1 at u; λ i and λ j are the weights of the main variable z1 and the covariate z2 respectively; n(u) and m(u) are the number of neighboring points of the main variable z1 and the covariate z2 at u, that is, the number of neighboring points within the range; u i and u j are the sampling positions of z1 and z2 respectively.

Images