Mineral prospecting prediction method and device based on fuzzy feature analysis of geological big data

Abstract

The present invention provides a prospecting prediction method and device based on fuzzy feature analysis of geoscience big data. The method includes: establishing a geoscience big data spatial database of a target prospecting area; dividing the target prospecting area to generate multiple grid cells; selecting multiple model cells from the multiple grid cells and selecting initial prediction variables; determining a fuzzy membership function for any initial prediction variable in each model cell; obtaining the fuzzy membership of the initial prediction variable of the model cell based on the fuzzy membership function; calculating a fuzzy matching coefficient between any initial prediction variable and any other initial prediction variable based on the fuzzy membership of the initial prediction variable of the model cell; determining a weight coefficient for each initial prediction variable based on the fuzzy matching coefficient; selecting a target prediction variable based on the weight coefficients of all initial prediction variables; and calculating the metallogenic favorability of the target prospecting area based on the fuzzy membership function and the weight coefficient of the target prediction variable. The present invention can improve prediction accuracy.

Description

Technical Field

[0001] The present application belongs to the field of big data prospecting technology, and specifically relates to a prospecting prediction method and device based on fuzzy feature analysis of geological big data. Background Art

[0002] In recent years, big data technology has been applied to mineral exploration. By analyzing massive amounts of multi-source, heterogeneous, and large-scale geological data, it is possible to more accurately predict the distribution and enrichment patterns of mineral resources, thereby reducing exploration risks. The characteristic analysis method is a multivariate statistical analysis method that uses in-depth analysis of geological big data to establish a characteristic model that reflects a specific type of mineral deposit, thereby achieving the goal of predicting that type of deposit. This method has a simple model structure, a wide range of applicability conditions, and strong geological interpretability. Therefore, it was widely used by many geologists and mineral exploration professionals in the early days.

[0003] However, this method has fallen out of favor in recent years, primarily due to a significant flaw: as a semi-quantitative method, it requires the predictor variables to be logical variables. Logical variables represent logically opposing states, such as the presence or absence of a geological factor or an abnormally high or low value. This is suitable for discrete variables like the presence or absence of a geological factor, but it is inappropriate for continuous data such as elemental concentrations obtained in geochemical exploration. The conversion to logical variables results in a loss of valuable detail, ultimately reducing the accuracy of the predictions.

[0004] Therefore, in view of the shortcomings of traditional feature analysis methods, there is an urgent need to propose an improved prospecting prediction method so that it can continue to maintain its technical advantages, and thus improve the prediction accuracy while being able to adapt to more mineral resource exploration scenarios. Summary of the Invention

[0005] In view of this, an embodiment of the present invention provides a mineral prospecting prediction method and device based on fuzzy feature analysis of geological big data, which can improve prediction accuracy while adapting to more mineral resource exploration scenarios.

[0006] The embodiment of the present invention provides a prospecting prediction method based on fuzzy feature analysis of geological big data, comprising:

[0007] Establishing a geoscience big data spatial database for the target prospecting area, wherein the geoscience big data spatial database includes a variety of different types of data;

[0008] Divide the target prospecting area and generate multiple grid cells;

[0009] Selecting a plurality of model units from a plurality of grid units, wherein the model units are grid units with a high degree of exploration and in which one or more mineral deposits have been discovered;

[0010] Selecting initial prediction variables from a geoscience big data spatial database corresponding to each model unit, wherein the initial prediction variables are at least two of a plurality of different types of data;

[0011] Determine the fuzzy membership function of any initial prediction variable in each model unit respectively;

[0012] Obtaining the fuzzy membership of the initial prediction variable of the model unit according to the fuzzy membership function; calculating the fuzzy matching coefficient between any initial prediction variable and any other initial prediction variable according to the fuzzy membership of the initial prediction variable of the model unit;

[0013] Determining the weight coefficient of each initial prediction variable according to the fuzzy matching coefficient;

[0014] Select the target predictor variable based on the weight coefficients of all initial predictor variables;

[0015] According to the fuzzy membership function and weight coefficient of the target prediction variable, the metallogenic favorableness of all grid cells in the target prospecting area is obtained.

[0016] Furthermore, the prospecting prediction method further includes: establishing a geoscience big data spatial database of the target prospecting area, wherein the geoscience big data spatial database includes a plurality of different types of data; the steps of establishing the geoscience big data spatial database of the target prospecting area include:

[0017] Acquire multiple different types of data in the target prospecting area, with each data having the same scale or accuracy, and including at least two of geology, geophysical prospecting, geochemical prospecting, remote sensing, and mineral location;

[0018] Based on the same coordinate system, various types of data are processed for spatial coordinate conversion and data format conversion to generate the geoscience big data spatial database.

[0019] Furthermore, the area of ​​the grid unit is calculated according to the following formula:

[0020]

[0021] Where d is the area of ​​the grid cell, D is the preset data scale or accuracy, and β is the impact factor.

[0022] Furthermore, when the initial prediction variable is a discrete variable, the fuzzy membership function of the initial prediction variable is determined using the following formula:

[0023]

[0024] Among them, μ A (x ij ) is the fuzzy membership degree of the j-th predictor variable of the i-th model unit to the fuzzy set A, x ij is the value of the jth predictor variable in the ith model unit, m is the number of initial predictor variables, and n is the number of model units;

[0025] When the initial prediction variable is a continuous variable, the fuzzy membership function of the initial prediction variable is determined using the following formula:

[0026]

[0027] Wherein, a and b are constants, and a<b, and a and b are determined based on the cumulative distribution function curve of each initial prediction variable.

[0028] Furthermore, the fuzzy matching coefficient is calculated using the following formula:

[0029]

[0030] Among them, r jk is the fuzzy matching coefficient between the jth initial predictor variable and the kth initial predictor variable, μ A (x ij ) is the fuzzy membership of the jth initial predictor variable of the i-th model unit, μ A (x ik ) is the fuzzy membership of the kth initial prediction variable of the ith model unit, m is the number of initial prediction variables, and n is the number of model units.

[0031] Furthermore, the weight coefficients of the initial predictor variables are determined using the following formula:

[0032]

[0033] Among them, a j is the weight coefficient of the jth initial predictor variable.

[0034] Furthermore, the step of selecting the target prediction variable according to the weight coefficients of all the initial prediction variables includes:

[0035] Based on the weight coefficients of all initial predictor variables, a weight coefficient change curve is drawn in descending order. According to the weight coefficient decreasing turning point on the weight coefficient change curve, all initial predictor variables before the decreasing turning point are selected as target predictor variables.

[0036] Furthermore, the calculation formula for the metallogenic favorableness is:

[0037]

[0038] Among them, f is the metallogenic favorable degree of the target prospecting area, μ A (x1), μ A (x2), ..., μ A (x p ) is the fuzzy membership of each target prediction variable, a1, a2, ..., a p is the weight coefficient of each target prediction variable, and p is the number of target prediction variables.

[0039] Furthermore, the prospecting prediction method further includes: classifying the types of various prospecting areas in the target prospecting area according to the metallogenic favorableness of the target prospecting area.

[0040] The embodiment of the present invention further provides a mineral prospecting prediction device based on fuzzy feature analysis of geological big data, comprising:

[0041] A construction unit is used to establish a geoscience big data spatial database of the target prospecting area, wherein the geoscience big data spatial database includes a plurality of different types of data;

[0042] A division unit is used to divide the target prospecting area into multiple grid units;

[0043] A first selection unit is configured to select a plurality of model units from a plurality of grid units, wherein the model units are grid units with a high degree of exploration and in which one or more mineral deposits have been discovered; and to select initial prediction variables from a geoscience big data spatial database corresponding to each model unit, wherein the initial prediction variables are at least two of a plurality of different types of data;

[0044] a processing unit, configured to respectively determine a fuzzy membership function of any initial prediction variable in each model unit; and obtain the fuzzy membership of the initial prediction variable of the model unit according to the fuzzy membership function, and calculate a fuzzy matching coefficient between any initial prediction variable and any other initial prediction variable according to the fuzzy membership of the initial prediction variable of the model unit, and determine a weight coefficient of each initial prediction variable according to the fuzzy matching coefficient;

[0045] The second selection unit is used to select the target prediction variable according to the weight coefficients of all the initial prediction variables;

[0046] The determination unit is used to calculate the metallogenic favorableness of all grid cells in the target prospecting area according to the fuzzy membership function and weight coefficient of the target prediction variable.

[0047] Compared to the prior art, the prospecting prediction method based on fuzzy feature analysis of geoscience big data provided by the embodiment of the present invention can obtain multiple different types of data used to characterize the mineral content of the target prospecting area by establishing a geoscience big data spatial database of the target prospecting area. Based on the multiple different types of data, the target prospecting area can be divided to generate multiple grid cells, which reduces computational complexity. In addition, this solution also selects at least one model cell from the multiple grid cells. Since the model cell is a grid cell with a high degree of exploration and one or more mineral deposits, it can be used to select initial prediction variables from it, which can reflect the favorable degree of mineralization. Then, the fuzzy membership function of any initial prediction variable in each model unit can be determined separately. Since the fuzzy membership function characterizes the relationship between different initial prediction variables and makes full use of the data in the geological big data spatial database, it avoids the loss of detailed information. Compared with the traditional prediction scheme based on logical variables, it can further improve the accuracy of the fuzzy matching coefficient, thereby increasing the weight coefficient of the initial prediction variable, so that the selected target prediction variable can more accurately reflect the mineral distribution of the current target prospecting area, and improve the calculation accuracy of the mineralization favorability. Moreover, this scheme is based on the traditional feature analysis method, which can improve the prediction accuracy while adapting to more mineral resource exploration scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0049] Figure 1 A flowchart of a mineral prospecting prediction method based on fuzzy feature analysis of geological big data in one embodiment of the present invention is shown;

[0050] Figure 2 A schematic diagram showing the principle of region division in one embodiment of the present invention is shown;

[0051] Figure 3 A schematic diagram showing a principle of determining a constant value in one embodiment of the present invention is shown;

[0052] Figure 4 A schematic diagram showing a curve change of a weight coefficient of an initial prediction variable in one embodiment of the present invention is shown;

[0053] Figure 5 A schematic structural diagram of a mineral prospecting prediction device based on fuzzy feature analysis of geological big data in one embodiment of the present invention is shown. DETAILED DESCRIPTION

[0054] As mentioned in the background, the introduction of feature analysis methods has also brought with it some new and challenging challenges in the field of mineral resource exploration. One of the more serious issues is that, as a semi-quantitative method, the predictor variables must be logical variables. Logical variables represent logically opposing concepts, such as the presence or absence of a geological factor or an abnormally high or low value. The most common representation is binary, using 1 and 0 to represent two opposing states.

[0055] This is suitable for discrete variables such as the presence or absence of a geological factor, but it is not suitable for element content data obtained from geochemical exploration, because they are continuous variables. In the process of converting them into logical variables, the data will lose a lot of useful detailed information, which will ultimately reduce the accuracy of the prediction results.

[0056] In order to solve the above technical problems, an embodiment of the present invention provides a prospecting prediction solution based on fuzzy feature analysis of geological big data. By establishing a geological big data spatial database of the target prospecting area, it is possible to obtain a large amount of data of various types used to characterize the mineral content of the target prospecting area. Based on the various types of data, the target prospecting area can be divided to generate multiple grid cells, which reduces the computational complexity. In addition, the present solution also selects multiple model cells from the multiple grid cells. Since the model cells are grid cells with a high degree of exploration and one or more mineral deposits have been discovered, initial prediction variables can be selected from them. The initial prediction variables can reflect the favorable degree of mineralization. Then, the fuzzy membership function of any initial prediction variable in the model unit can be determined. Since the fuzzy membership function makes full use of the data in the geological big data spatial database, it avoids the loss of detailed information. Compared with the traditional prediction scheme based on logical variables, it can further improve the accuracy of the fuzzy matching coefficient, thereby improving the accuracy of the initial prediction variable weight coefficient, so that the selected target prediction variable can more accurately reflect the mineral distribution of the current target prospecting area, and improve the calculation accuracy of the mineralization favorability. This scheme inherits the advantages of the traditional feature analysis method, and can adapt to more mineral resource exploration scenarios while also improving the prediction accuracy.

[0057] In other words, by improving the shortcomings of traditional feature analysis methods, we hope to enable them to continue to maintain their technological advantages, thereby improving prediction accuracy while adapting to more mineral resource exploration scenarios.

[0058] In order to enable those skilled in the art to better understand the inventive concept, working principle and advantages of the embodiments of the present invention, the prospecting scheme based on fuzzy feature analysis of geological big data in the embodiments of the present invention is described in detail below.

[0059] See also Figure 1 FIG. 1 is a flow chart of a mineral prospecting prediction method based on fuzzy feature analysis of geological big data in one embodiment of the present invention, as shown in FIG. Figure 1 As shown, you can perform the following steps:

[0060] S10, establishing a geoscience big data spatial database of the target prospecting area, wherein the geoscience big data spatial database includes a variety of different types of data.

[0061] In this embodiment, based on typical mineral deposits and regional mineralization laws (for example, time, space, genesis, and mineral types, etc.), and combined with existing regional geological and mineral data, areas with favorable mineralization are selected as target prospecting areas, which can increase the possibility of exploring potential resources.

[0062] Based on typical mineral deposits and regional metallogenic laws (time, space, genesis and mineral types), and combined with existing regional geological and mineral data, areas that are favorable for the mineralization of the target deposit type are selected as target prospecting areas.

[0063] For example, skarn-type copper polymetallic minerals are selected as target deposit types.

[0064] In this embodiment, the screening criteria for target prospecting areas favorable for mineralization include: 1) the target prospecting area is located within a known large-scale mineralization belt or mineral cluster; 2) there are known mineral deposits (for example, skarn-type copper polymetallic deposits) around the target prospecting area.

[0065] It should be noted that the selection criteria listed in the above examples are only illustrative and are used to indicate the requirements to be met in the target prospecting area, and should not be understood as a limitation to the present invention.

[0066] In this embodiment, the multiple different types of data refer to parameters used to characterize different parameter information of the target prospecting area. By obtaining multiple different types of data, the morphology and geological parameters of the target prospecting area can be more realistically reflected.

[0067] In this embodiment, step S10 may include: obtaining geological big data information of the target prospecting area; performing spatial coordinate conversion processing on various types of geological big data based on the same coordinate system, and performing data format conversion processing to generate the geological big data spatial database.

[0068] The multiple different types of geoscience big data may include at least two of geological data, geophysical data, geochemical data, remote sensing data, and mineral deposit data.

[0069] More specifically, geological data primarily relates to regional geological maps.

[0070] Geophysical data collection mainly involves regional gravity, magnetic and radioactivity measurement data obtained from geophysical scanning work.

[0071] The collection of geochemical data mainly involves the regional stream sediments, rock debris and soil measurement data obtained from geochemical scanning. The media are different at different scales and in different geochemical landscapes.

[0072] Remote sensing data collection involves a wide variety of remote sensing data. The specific type of data to be collected needs to be determined based on the target mineral deposit type and target task. It can be roughly divided into remote sensing data used for extracting alteration information and remote sensing data used for extracting information on ore-controlling structures and ore-bearing geological bodies.

[0073] The collection of mineral deposit data mainly involves the discovered mineral deposits (points) in the working area, including coordinates, cumulative proven resources and other information.

[0074] It should be noted that: in general, in order to facilitate similar analogies, the collection of geological, geophysical, geochemical and remote sensing data in the work area should adhere to the principle of scale equivalence, that is, similar data in the work area need to be fully covered, the scale needs to be as consistent as possible, and it is best to choose data with a larger scale.

[0075] In other words, the data scale or accuracy of each data is the same.

[0076] In this embodiment, current data is often stored in vector files, raster files, or table files. Some old data exists in the form of paper media and needs to be vectorized first.

[0077] Specifically, due to the different types, ages and sources of the above-mentioned geological, mineral, geophysical, geochemical and remote sensing data collected, there are also large differences in the geographic and projection coordinate systems. Among them, the common geographic coordinate systems corresponding to the projection coordinate system include Beijing 54 geographic coordinate system, Xi'an 80 geographic coordinate system, CGCS2000 geographic coordinate system and WGS84 geographic coordinate system; the projection coordinate systems include Gauss-Kriging (3 degree zone or 6 degree zone), Universal Mercator and Lambert projections.

[0078] To analyze data within a unified spatial coordinate system, the geographic coordinate system will be the CGCS2000 geographic coordinate system, currently used in my country, which is based on a centroidal coordinate system. For smaller data sets, the projected coordinate system will use the Gauss-Kriging (3-degree or 6-degree zone) projection, while for larger data sets, the Lambert projection will be used. Different spatial coordinate systems can be converted to a unified coordinate system using three-parameter or seven-parameter projection transformations within GIS software, ultimately establishing a unified spatial database for geoscience data.

[0079] In other words, based on the same coordinate system, performing spatial coordinate conversion processing on various types of data can improve the format consistency between various types of data, which is conducive to reducing the difficulty of subsequent data processing in the spatial database of geoscience big data.

[0080] In this embodiment, while unifying the coordinate system, a data conversion operation can also be performed.

[0081] For example, all data related to geology, mineral deposits (points), geophysical exploration, geochemical exploration, and remote sensing can be rasterized. Data in vector format can be directly converted to raster using the "area to raster" or "point to raster" methods. Data in table format can be interpolated using the "inverse distance weighted" method to generate rasterized data. Data in raster format can remain unchanged.

[0082] It should be noted that some special geological units in geological data, such as faults, external contact zones of rock bodies, and structural intersections, generally require first using the "buffer analysis" method to determine the impact range, and then using the "surface to grid" method to rasterize them. Ultimately, a unified geoscience big data spatial database is established.

[0083] It should also be noted that the multiple different types of data in the above examples are merely illustrative and are used to illustrate the acquisition of multiple data, and should not be construed as limiting the present invention. In some other embodiments, the data of the acquisition area may also include mask data.

[0084] S20, dividing the target prospecting area to generate a plurality of grid units.

[0085] Specifically, the grid unit is the most commonly used basic unit for prospecting and prediction in this plan. Dividing the grid unit in the target working area is a basic task for prospecting and prediction. By dividing the target prospecting area into multiple grid units, it is helpful to reduce the difficulty of executing subsequent steps.

[0086] In this embodiment, when executing step S20, the most critical issue is selecting the grid unit size, because different grid unit size divisions will have a significant impact on the prediction results.

[0087] In this embodiment, the data defines the morphology and geological parameters of the target prospecting area, and the target prospecting area already includes the mineral-bearing area. Based on the acquired data, multiple model units can be selected from them, and the selected model units are used as a reference for the prospecting prediction process.

[0088] In addition, when conducting prospecting predictions, the accuracy or scale of geological, geophysical, geochemical and remote sensing data should be as consistent as possible, and the corresponding accuracy or scale should be used as the basis for dividing the size of grid units.

[0089] Furthermore, the scales of different target deposit types vary greatly. For example, the distribution range of general endogenous deposits is approximately from a few square kilometers to several square kilometers, while the scale of exogenous deposits is often much larger, generally from several square kilometers to tens of square kilometers. Therefore, the spatial distribution of target deposit types is another important basis for dividing the grid unit size.

[0090] Based on this, step S20 may include: dividing the target prospecting area according to the data scale or accuracy of the data and the type of the target prospecting area to generate multiple grid units, each grid unit having its own area.

[0091] It should be noted that the preset data scale or accuracy can be obtained from a geographic information system (GIS) platform; or it can be understood as obtaining data at that data scale or accuracy.

[0092] Specifically, the data scale or accuracy defines the detection accuracy when acquiring the data, while the type of target prospecting area determines how the grid cells are affected by the type of ore deposit. Thus, after obtaining these two parameters, a meshing operation can be performed to form multiple grid cells, each with its own area.

[0093] More specifically, the area of ​​each grid cell can be determined as follows:

[0094]

[0095] Where d is the area of ​​the grid cell, D is the preset data scale or accuracy, and β is the impact factor.

[0096] It should be noted that, first, the β value is generally taken as 1 in the prediction of endogenous deposits, and generally takes a larger value in the prediction of exogenous deposits. The specific value should be determined according to the actual distribution of the target deposit type; second, the value of D should be determined according to the data scale or accuracy, generally representing the line distance of the route survey or the side length of the sampling area of ​​a single sample. Among them, the D value of the prospecting prediction of 1:50000 is generally taken as 0.5km, and the D value of the prospecting prediction of 1:200000 is generally taken as 2km; third, It is a correction to the D value, mainly to prevent the loss of local or detail information.

[0097] Furthermore, the grid cell area determined using the above formula can accurately scale the grid cells. This avoids the problem of overly large grid cells thinning out the data and losing useful information, thereby improving prediction accuracy. It also avoids the problem of overly small grid cells increasing the computational effort and affecting computational efficiency, thus preventing the "salt and pepper" effect in the calculation results.

[0098] As an example, see Figure 2 A schematic diagram of a principle of area division in an embodiment of the present invention is shown in FIG. Figure 2 As shown, the target prospecting area 10 can be divided according to the data scale or accuracy of the data and the type of the target prospecting area to generate multiple grid units (e.g. Figure 2 Schematic grid unit 12).

[0099] In some embodiments, the areas of the grid cells 12 may be the same.

[0100] S30, selecting a plurality of model units from a plurality of grid units, wherein the model units are grid units with a high degree of exploration and in which one or more mineral deposits have been discovered.

[0101] Specifically, multiple grid units together constitute the target prospecting area, and different areas in the target prospecting area correspond to different parameter characteristics. Therefore, multiple model units can be selected based on different types of data.

[0102] It should be noted that this step is performed on a Geographic Information System (GIS) platform, and sufficient evidence is required in the process of obtaining model units.

[0103] First, there is at least one mineral deposit within the grid, that is, after exploration, a mineral deposit with a small or larger scale of resources has been submitted; second, the exploration level of the mineral deposit should be high, and it is believed that the resource volume within the grid unit has been basically ascertained, so the exploration should at least reach the level of a general survey, preferably a detailed survey or exploration level.

[0104] For example, see Figure 2 , by performing the partitioning operation, 3 model units can be selected from multiple grid units.

[0105] In other words, it should be noted that the relationship between the target prospecting area, grid unit and model unit is that the grid unit covers the entire target prospecting area with a fixed size, and the model unit is the mineral-bearing unit in the grid unit.

[0106] It should be noted that the implementation of the characteristic analysis method is to determine the initialization prediction variables by summarizing the ore-controlling factors and prospecting signs of multiple model units. Therefore, the number of model units is preferably multiple, and the more the better.

[0107] S40, selecting initial prediction variables from the geoscience big data spatial database corresponding to each model unit, wherein the initial prediction variables are at least two of a plurality of different types of data.

[0108] Specifically, the model unit contains multiple different types of data, and these data play different roles in the prediction process. Therefore, based on the understanding of mineralization geology, initial prediction variables can be selected, and these initial prediction variables are data on mineralization-controlling factors and prospecting signs that are favorable to mineralization.

[0109] In this embodiment, there are at least two approaches to selecting favorable mineralization-controlling factors and prospecting markers as initial predictive variables. One approach is to base these selections on understanding of mineralization geology. For example, in the case of magmatic copper-nickel sulfide deposits, extensive research on the mineralization patterns of typical deposits reveals that these deposits require mantle-derived magma to intrude through large transcrustal faults to ultimately form. Therefore, deep large faults in geological maps can be selected as one of the initial predictive variables.

[0110] Another approach involves selecting predictive variables by comparing them on a graph to see if they indicate the location of the deposit. For example, in the case of low-temperature hydrothermal gold deposits, gold deposits tend to be found at medium-high values ​​of Au, As, Bi, and Hg, while low values ​​produce less. Therefore, the Au, As, Bi, and Hg contents in the geochemical data can be selected as part of the initial predictive variables.

[0111] It should be noted that the core of implementing the feature analysis method is to determine the importance of variables through the relationship between variables. Therefore, at least two initial prediction variables should be selected. Otherwise, the feature analysis method will not be able to determine the weight coefficient of the variable and the feature analysis method will not be applicable.

[0112] S50, respectively determining the fuzzy membership function of any initial prediction variable in each model unit.

[0113] In this embodiment, when conducting mineral exploration prediction for grid units, the model unit has multiple initial prediction variables. By introducing the fuzzy membership function, the continuous variables are converted into membership values. This can make full use of geological big data and avoid the loss of detailed information. While improving the accuracy of the prediction results, it also maintains the advantages of traditional feature analysis.

[0114] In this embodiment, the corresponding fuzzy membership function is determined according to the numerical type of each initial prediction variable.

[0115] More specifically, when the initial prediction variable is a discrete variable, the following formula is used to determine the fuzzy membership function of the initial prediction variable:

[0116]

[0117] Among them, μ A (x ij ) is the fuzzy membership degree of the j-th predictor variable of the i-th model unit to the fuzzy set A, x ij is the value of the jth predictor variable in the ith model unit, m is the number of initial predictor variables, and n is the number of model units.

[0118] Among them, x ij Read directly on the GIS software platform.

[0119] In other words, when the initial predictor variable is a logical variable expressing the existence or non-existence, it can take the value of 0 or 1 based on the existence or non-existence of the initial predictor variable.

[0120] For example, whether a granite body in a certain place exists or is missing, if it exists, the value is 1, if it is missing, the value is 0; for another example, whether a geomagnetic anomaly is high or low, if it is high, the value is 1, if it is low, the value is 0.

[0121] When the initial prediction variable is a continuous variable, in order to prevent the problem of information loss during the prediction process, another method can be used to determine the fuzzy membership function of the initial prediction variable.

[0122] Among them, the so-called continuous variable means that the value of the predicted variable is continuous. For example, the Cu element content data obtained from a geochemical exploration scan in a certain area is a continuous variable, and its value range can be any real number greater than 0.

[0123] Specifically, when the initial prediction variable is a continuous variable, the following formula is used to determine the fuzzy membership function of the initial prediction variable:

[0124]

[0125] Wherein, a and b are constants, and a<b, and a and b are determined based on the inflection point value of the cumulative distribution function curve of each initial prediction variable.

[0126] As a specific example, see Figure 3 A schematic diagram of a principle for determining a constant value in an embodiment of the present invention is shown in FIG. Figure 3The figure shows a cumulative distribution function curve, where the horizontal axis represents the initial predictor variable and the vertical axis represents the cumulative distribution function value of the initial predictor variable. Using the cumulative distribution function, by accumulating the distribution probabilities of each initial predictor variable, the required inflection points a and b of the membership function curve can be determined.

[0127] More specifically, the cumulative distribution function is as follows:

[0128]

[0129] Among them, F U (u) is the cumulative distribution function, u represents the upper limit of integration, x is the initial prediction variable, and f(x) is the probability density function of the prediction variable.

[0130] It should be pointed out that in the actual prediction process, a and b can also be obtained through other methods.

[0131] Therefore, the accuracy of the fuzzy membership function is improved by performing calculations in different situations based on the properties of the initial prediction variables.

[0132] S60, obtaining the fuzzy membership of the initial prediction variable of the model unit according to the fuzzy membership function; and calculating the fuzzy matching coefficient between any initial prediction variable and any other initial prediction variable according to the fuzzy membership of the initial prediction variable of the model unit.

[0133] In this embodiment, the initial prediction variables refer to the mineralization-related controlling factors and prospecting signs. The introduction of the fuzzy membership function prevents the loss of detailed information of the continuous prediction variables. Therefore, based on the membership of the initial prediction variables, the fuzzy matching coefficient between any two initial prediction variables in each model unit can be determined by the matching coefficient method.

[0134] In a specific embodiment, assuming that there are n model units and m initial prediction variables in the target prospecting area, the fuzzy matching coefficient between each initial prediction variable and other initial prediction variables in any model unit can be determined.

[0135] More specifically, step S60 may include: respectively calculating the product value between the fuzzy membership function of any initial prediction variable and the fuzzy membership function of any other one; and taking the sum of all product values ​​corresponding to the any initial prediction variable on the n model units as the fuzzy matching coefficient.

[0136] In this embodiment, the fuzzy matching coefficient between any one initial prediction variable and all other initial prediction variables is calculated using the following formula:

[0137]

[0138] Among them, r jk is the fuzzy matching coefficient between the jth initial predictor variable and the kth initial predictor variable, μ A (x ij ) is the fuzzy membership of the jth initial predictor variable of the i-th model unit, μ A (x ik ) is the fuzzy membership of the kth initial prediction variable of the ith model unit, m is the number of initial prediction variables, and n is the number of model units.

[0139] S70 , determining weight coefficients of the initial prediction variables based on the fuzzy matching coefficients between any initial prediction variable and all other initial prediction variables.

[0140] In this embodiment, the weight coefficient determines the importance of the initial prediction variable, and thus can be used as a determining factor for the target prediction variable based on the weight coefficient.

[0141] In one embodiment, step S70 may include: determining the sum of the squares of the fuzzy matching coefficients corresponding to any initial prediction variable; and using the square root of the sum as the weight coefficient of the initial prediction variable.

[0142] More specifically, the weight coefficients of the initial predictor variables can be determined as follows:

[0143]

[0144] Among them, a j is the weight coefficient of the jth initial predictor variable.

[0145] In this embodiment, the cumulative sum of the squares of the fuzzy matching coefficients is calculated, and the square root value is taken to calculate the weight coefficient of the prediction factor. The idea behind the square sum method is that the stronger the correlation between the initial prediction variable and other initial prediction variables, the more important the initial prediction variable is and the greater its impact on mineralization, which is consistent with the characteristics of geological mineralization.

[0146] S80: Select a target prediction variable from all the initial prediction variables based on the weight coefficients of the initial prediction variables in the model unit.

[0147] The weight coefficients of different initial predictors vary in size, reflecting their importance. Therefore, the initial predictor variables that serve as target predictors can be determined based on the weight coefficients of the initial predictors. Specifically, the weight coefficients of the initial predictors are sorted from largest to smallest, and a corresponding weight coefficient change curve diagram is established. The weight coefficient decrease turning point is determined based on the waveform changes indicated in the weight coefficient change curve diagram. The number of target predictor variables is determined based on the location of the weight coefficient decrease turning point. In other words, a weight coefficient change curve is drawn from largest to smallest based on the weight coefficients of all initial predictor variables. Based on the weight coefficient decrease turning point on the weight coefficient change curve, all initial predictor variables before the decrease turning point are selected as target predictor variables.

[0148] Since the turning point on the weight coefficient change curve represents a sudden decrease in the weight coefficient, the initial variable corresponding to the greatly reduced weight coefficient is no longer important and can be removed. Therefore, this embodiment determines the number of target prediction variables by the turning point on the weight coefficient change curve where the weight coefficient suddenly decreases, rather than arbitrarily deciding the number of variables in advance. For example, referring to Figure 4 As shown in the figure, the turning point P on the weight coefficient curve is determined. From the m initial predictor variables, the first p initial predictor variables with weight coefficients greater than the weight coefficient at the turning point P are used as the final predictor variables for constructing the model. In other words, the predictor variables before the turning point on the weight coefficient curve are retained, and the predictor variables after the turning point are discarded. Therefore, when there are m initial predictor variables, p target predictor variables can be selected from them.

[0149] S90, obtaining the metallogenic favorableness of all grid cells in the target prospecting area based on the fuzzy membership function and weight coefficient of the target prediction variable. The metallogenic favorableness is a comprehensive indicator of the possibility of mineral formation and enrichment within the geological units in the target prospecting area.

[0150] Exemplarily, step S90 includes:

[0151] S91: Obtain the fuzzy membership function of the target prediction variable, and use the fuzzy membership function of the target prediction variable as the target prediction function; obtain the fuzzy membership of each target prediction variable according to the target prediction function; calculate the fuzzy matching coefficient between any target prediction variable and any other target prediction variable according to the fuzzy membership of the target prediction variable.

[0152] Specifically, the target prediction variable inherits the main characteristics of the initial prediction variable, so that after determining the target prediction function, the fuzzy matching coefficient between a target prediction variable and any other target prediction variable can be re-determined.

[0153] S92: Determine a weight coefficient of any target prediction variable based on the fuzzy matching coefficient between any target prediction variable and all other target prediction variables.

[0154] S93. Based on the fuzzy membership of each target prediction variable obtained in step S91 and the weight coefficient of any target prediction variable obtained in step S92, calculate the sum W1 of the product of the fuzzy membership of the target prediction variable and the corresponding weight coefficient; and calculate the sum W2 of the weight coefficients of each target prediction variable.

[0155] S94, the ratio of the sum W1 of the product of the fuzzy membership of each target prediction variable and its corresponding weight coefficient to the sum W2 of the weight coefficients of each target prediction variable is used as the metallogenic favorable degree of the target prospecting area. Specifically, the metallogenic favorable degree is calculated according to the following formula:

[0156]

[0157] Among them, f is the metallogenic favorable degree of the target prospecting area, μ A (x1), μ A (x2), ..., μ A (x p ) is the fuzzy membership of each target prediction variable, a1, a2, ..., a p is the weight coefficient of each target prediction variable, and p is the number of target prediction variables.

[0158] In step S90 , for the specific calculation process of the fuzzy membership, fuzzy matching coefficient, and weight coefficient of the target prediction variable, refer to the aforementioned steps S50 - S70 .

[0159] It should be noted that in this solution, after determining the target predictor variable, the reason for determining the weight coefficient of the target predictor variable again is that the weight coefficient assigned to the predictor variable in the aforementioned example is determined based on the matching relationship between all predictor variables involved in the calculation, and it reflects the relationship between all predictor variables; however, the weight coefficient of the target predictor variable (p) finally obtained should still be determined by the matching relationship between these p predictor variables, so that the optimal model can be obtained.

[0160] In other words, this scheme can determine the parameters of the target prediction variables used to calculate the mineralization favorability by calculating the weight coefficients twice.

[0161] In this embodiment, after determining the mineralization favorableness, the types of prospective mineralization areas in the target mineralization area can be divided according to the mineralization favorableness of the target mineralization area (or the size of the mineralization favorableness) to further guide the mineralization process.

[0162] For example, in the target prospecting area, the above formula is used to calculate the metallogenic favorableness for all grid cells, and according to the size of the metallogenic favorableness, grid cells with a metallogenic favorableness greater than 0.8 are divided into Class A prospecting areas; grid cells with a metallogenic favorableness less than or equal to 0.8 and greater than 0.65 are divided into Class B prospecting areas; grid cells with a metallogenic favorableness less than or equal to 0.65 and greater than 0.5 are divided into Class C prospecting areas.

[0163] It should be noted that there are no strict regulations on the threshold value of mineralization favorableness used to classify prospective areas. Generally, it can be determined artificially, or the mineralization favorableness can be ranked and the threshold value for classifying prospective areas can be determined based on the inflection points that appear.

[0164] Therefore, the mineral exploration prediction method based on fuzzy feature analysis of geological big data in the above example is adopted. On the one hand, when dividing the grid units, the characteristics of the data scale and the target mineral deposit type are comprehensively considered, so that the prediction results have higher accuracy while ensuring the efficiency of the prediction; on the other hand, the fuzzy membership function in fuzzy mathematics is introduced into the feature analysis to replace the logical variable, thereby making full use of geological big data, avoiding the loss of detailed information, improving the accuracy of the prediction results while maintaining the advantages of traditional feature analysis.

[0165] The embodiment of the present invention also provides a device corresponding to the prospecting prediction method based on fuzzy feature analysis of geological big data, such as Figure 4 The schematic diagram of the structure of a potential resource estimation device in one embodiment of the present invention is shown in FIG. Figure 5 As shown, the mineral prospecting prediction device 100 based on fuzzy feature analysis of geological big data may include:

[0166] A construction unit 110 is used to establish a geoscience big data spatial database of a target prospecting area, wherein the geoscience big data spatial database includes a plurality of different types of data;

[0167] A division unit 120 is used to divide the target prospecting area into multiple grid units;

[0168] A first selection unit 130 is configured to select a plurality of model units from a plurality of grid units, wherein the model units are grid units with a high degree of exploration and in which one or more mineral deposits have been discovered; and to select initial prediction variables from a geoscience big data spatial database corresponding to each model unit, wherein the initial prediction variables are at least two of a plurality of different types of data;

[0169] The processing unit 140 is configured to determine a fuzzy membership function of any initial prediction variable in each model unit; obtain the fuzzy membership of the initial prediction variable of the model unit according to the fuzzy membership function; calculate the fuzzy matching coefficient between any initial prediction variable and any other initial prediction variable according to the fuzzy membership of the initial prediction variable of the model unit; and determine the weight coefficient of each initial prediction variable according to the fuzzy matching coefficient between any initial prediction variable and all other initial prediction variables;

[0170] A second selection unit 150 is configured to select a target prediction variable based on the weight coefficients of all initial prediction variables;

[0171] The determination unit 160 is configured to determine the metallogenic favorableness of all grid cells in the target prospecting area according to the fuzzy membership function and weight coefficient of the target prediction variable.

[0172] Among them, the specific working processes and principles of the construction unit 110, the division unit 120, the first selection unit 130, the processing unit 140, the second selection unit 150 and the determination unit 160 can be found in the relevant description of the above example.

[0173] It is understandable that the division of the above units is only a division of logical functions, and in actual implementation, they can be fully or partially integrated into one physical entity, or physically separated. In addition, the above units can be implemented in the form of a processor calling software.

[0174] It should be noted that “one embodiment” or “embodiment” referred to in the present invention refers to a specific feature, structure or characteristic that may be included in at least one implementation method of the present invention. And in the description of the present invention, terms such as “first” and “second” are used only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, features defined by terms such as “first” and “second” may explicitly or implicitly include one or more of the features. Moreover, terms such as “first” and “second” are used to distinguish similar objects, and are not necessarily used to describe a specific order or to express importance. It is understood that the terms used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than that illustrated or described.

[0175] Although the embodiments of the present invention are disclosed above, the present invention is not limited thereto. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention should be based on the scope defined by the claims.

Claims

1. A prospecting prediction method based on fuzzy feature analysis of geological big data, characterized in that: include: Establish a geoscience big data spatial database for the target prospecting area; Divide the target prospecting area and generate multiple grid cells; Selecting a plurality of model units from a plurality of grid units, wherein the model units are grid units with a high degree of exploration and in which one or more mineral deposits have been discovered; Selecting initial prediction variables from a geoscience big data spatial database corresponding to each model unit, wherein the initial prediction variables are at least two of a plurality of different types of data; Determine the fuzzy membership function of any initial prediction variable in each model unit respectively; Obtaining the fuzzy membership of the initial prediction variable of the model unit according to the fuzzy membership function; calculating the fuzzy matching coefficient between any initial prediction variable and any other initial prediction variable according to the fuzzy membership of the initial prediction variable of the model unit; Determining the weight coefficient of each initial prediction variable according to the fuzzy matching coefficient; Select the target predictor variable based on the weight coefficients of all initial predictor variables; According to the fuzzy membership function and weight coefficient of the target prediction variable, the metallogenic favorableness of all grid cells in the target prospecting area is obtained.

2. The mineral prospecting prediction method based on fuzzy feature analysis of geological big data according to claim 1 is characterized in that: Before dividing the target prospecting area, the method further includes: establishing a geoscience big data spatial database of the target prospecting area, the steps of which are: Acquire multiple different types of data in the target prospecting area, with each data having the same scale or accuracy, and including at least two of geology, geophysical prospecting, geochemical prospecting, remote sensing, and mineral location; Based on the same coordinate system, various types of data are processed for spatial coordinate conversion and data format conversion to generate the geoscience big data spatial database.

3. The mineral prospecting prediction method based on fuzzy feature analysis of geological big data according to claim 1 is characterized in that: The area of ​​the grid cell is calculated as follows: Where d is the area of ​​the grid cell, D is the preset data scale or accuracy, and β is the impact factor.

4. The mineral prospecting prediction method based on fuzzy feature analysis of geological big data according to claim 3 is characterized in that: When the initial prediction variable is a discrete variable, the fuzzy membership function of the initial prediction variable is determined using the following formula: Among them, μ A (x ij ) is the fuzzy membership degree of the j-th predictor variable of the i-th model unit to the fuzzy set A, x ij is the value of the jth predictor variable in the ith model unit, m is the number of initial predictor variables, and n is the number of model units; When the initial prediction variable is a continuous variable, the fuzzy membership function of the initial prediction variable is determined using the following formula: Wherein, a and b are constants, and a<b, and a and b are determined based on the cumulative distribution function curve of each initial prediction variable.

5. The mineral prospecting prediction method based on fuzzy feature analysis of geological big data according to claim 4 is characterized in that: The fuzzy matching coefficient is calculated using the following formula: Among them, r jk is the fuzzy matching coefficient between the jth initial predictor variable and the kth initial predictor variable, μ A (x ij ) is the fuzzy membership of the jth initial predictor variable of the i-th model unit, μ A (x ik ) is the fuzzy membership of the kth initial prediction variable of the ith model unit, m is the number of initial prediction variables, and n is the number of model units.

6. The mineral prospecting prediction method based on fuzzy feature analysis of geological big data according to claim 5 is characterized in that: The weight coefficient of each initial predictor variable is determined using the following formula: Among them, a j is the weight coefficient of the jth initial predictor variable.

7. The mineral prospecting prediction method based on fuzzy feature analysis of geological big data according to claim 1 is characterized in that: The step of selecting the target prediction variable according to the weight coefficients of all the initial prediction variables includes: Based on the weight coefficients of all initial predictor variables, a weight coefficient change curve is drawn in descending order. According to the weight coefficient decreasing turning point on the weight coefficient change curve, all initial predictor variables before the decreasing turning point are selected as target predictor variables.

8. The mineral prospecting prediction method based on fuzzy feature analysis of geological big data according to claim 6 is characterized in that: The calculation formula of the metallogenic favorable degree is: Among them, f is the metallogenic favorable degree of the target prospecting area, μ A (x1), μ A (x2), ..., μ A (x p ) is the fuzzy membership of each target prediction variable, a1, a2, ..., a p is the weight coefficient of each target prediction variable, and p is the number of target prediction variables.

9. The mineral prospecting prediction method based on fuzzy feature analysis of geological big data according to claim 1, characterized in that: Also includes: According to the metallogenic favorableness of the target prospecting area, the types of prospecting areas in the target prospecting area are divided.

10. A mineral prospecting prediction device based on fuzzy feature analysis of geological big data, characterized in that: include: A construction unit is used to establish a geoscience big data spatial database of the target prospecting area, wherein the geoscience big data spatial database includes a plurality of different types of data; A division unit is used to divide the target prospecting area into multiple grid units; A first selection unit is used to select a plurality of model units from a plurality of grid units, wherein the model units are grid units with a high degree of exploration and in which one or more mineral deposits have been discovered; and selecting initial prediction variables from a geoscience big data spatial database corresponding to each model unit, wherein the initial prediction variables are at least two of a plurality of different types of data; a processing unit, configured to respectively determine a fuzzy membership function of any initial prediction variable in each model unit; and obtain the fuzzy membership of the initial prediction variable of the model unit according to the fuzzy membership function, and calculate a fuzzy matching coefficient between any initial prediction variable and any other initial prediction variable according to the fuzzy membership of the initial prediction variable of the model unit, and determine a weight coefficient of each initial prediction variable according to the fuzzy matching coefficient; The second selection unit is used to select the target prediction variable according to the weight coefficients of all the initial prediction variables; The determination unit is used to calculate the metallogenic favorableness of all grid cells in the target prospecting area according to the fuzzy membership function and weight coefficient of the target prediction variable.

Images